^ B£> 9 P*si)CHEMICAL RESE/' • ' 1912-13 UC-NRLF Sri 93 THE STAFF OF THE BIOCHEMICAL LABORATORY THE UNIVERSITY OF CHICAGO ALBERT P. MATHEWS Editor UNIVERSITY CHICAGO 1914 GIFT OF Prof. Jacques Loeb BIOCHEMICAL RESEARCH 1912-13 By THE STAFF OF THE BIOCHEMICAL LABORATORY THE UNIVERSITY OF CHICAGO 'I ALBERT P. MATHEWS Editor CHICAGO 1914 .V «?. - .<••• : .% ' "> BIOLOGY LHL/:A» ujBftA.-.y BIOLOGf UBRARY G DEDICATED TO THE MEMORY OF WALDEMAR KOCH (April 8, 1875 — February i, 1911) CONTENTS 1. WALDEMAR KOCH By Albert P. Mathews (Biochemical Bulletin, I, 372-76, 1912) 2. STUDIES ON THE PHYSICAL PROPERTIES OF PROTOPLASM By G. L. Kite (American Journal of Physiology, XXXII, 146-64, 1913) 3. ON THE NATURE OF THE IODINE CONTAINING COMPLEX IN THYREOGLOBULIN By Fred C. Koch (Journal of Biological Chemistry, XIV, 101-16, 1913) 4. A COMPARISON OF THE BRAIN OF THE ALBINO RAT AT BIRTH WITH THAT OF THE FETAL PIG By Mathilde L. Koch (Journal of Biological Chemistry, XIV, 267-79, 1913) 5. A COMPARISON OF TWO METHODS OF PRESERVING NERVE TISSUE FOR SUBSEQUENT CHEMICAL EXAMINATION By W. Koch and M. L. Koch (Journal of Biological Chemistry, XIV, 281-82, 1913) 6. THE CHEMICAL DIFFERENTIATION OF THE BRAIN OF THE ALBINO RAT DURING GROWTH By W. Koch and M. L. Koch (Journal of Biological Chemistry, XV, 423-46, 1913) 7. ADAPTATION FROM THE POINT OF VIEW OF THE PHYSIOLO- GIST By A. P. Mathews (American Naturalist, XL VII, 90-103, 1913) 8. A METHOD OF DETERMINING "A" OF VAN DER WAALS'S EQUATION By A. P. Mathews (Journal of Physical Chemistry, XVII, 154-61, 1913) 9. THE RELATION OF THE VALUE "A" OF VAN DER WAALS'S EQUATION TO THE MOLECULAR WEIGHT AND NUMBER OF VALENCES OF THE MOLECULE By A. P. Mathews (Journal of Physical Chemistry, XVII, 181-204, 1913) 10. THE VALENCE OF CHLORINE AS DETERMINED FROM THE MOLECULAR COHESION OF CHLORINE COMPOUNDS By A. P. Mathews (Journal of Physical Chemistry, XVII, 252-63, 1913) iii 801295 iv CONTENTS 11. THE VALENCE OF OXYGEN, SULFUR, NITROGEN AND PHOS- PHORUS DETERMINED FROM THE MOLECULAR COHESION By A. P. Mathews (Journal of Physical Chemistry, XVII, 331-36, 1913) 12. THE VALENCE OF THE ARGON GROUP AS DETERMINED FROM THE MOLECULAR COHESION By A. P. Mathews (Journal of Physical Chemistry, XVII, 337~43, 1913) 13. A NOTE ON THE STRUCTURE OF ACETYLENE By A. P. Mathews (Journal of Physical Chemistry, XVII, 320-21, 1913) 14. DO MOLECULES ATTRACT COHESIVELY INVERSELY AS THE SQUARE OF THE DISTANCE? By A. P. Mathews (Journal of Physical Chemistry, XVII, 520-36, 1913) 15. THE SIGNIFICANCE OF THE RELATIONSHIP BETWEEN MOLECULAR COHESION AND THE PRODUCT OF THE MOLECU- LAR WEIGHT AND THE NUMBER OF VALENCES By A. P. Mathews (Journal of Physical Chemistry, XVII, 481-500, 1913) 16. THE INTERNAL PRESSURES OF LIQUIDS By A. P. Mathews (Journal of Physical Chemistry, XVII, 603-28, 1913) 17. THE QUANTITY OF RESIDUAL VALENCE POSSESSED BY VARIOUS MOLECULES By A. P. Mathews (Journal of Physical Chemistry, XVIII, 474, 1914) 18. AN IMPORTANT CHEMICAL DIFFERENCE BETWEEN THE EGGS OF THE SEA URCHIN AND THOSE OF THE STAR FISH By A. P. Mathews (Journal of Biological Chemistry, XIV, 465-67, 1913) 19. A NEW METHOD AND APPARATUS FOR THE ESTIMATION OF EXCEEDINGLY MINUTE AMOUNTS OF CARBON DIOXIDE By Shiro Tashiro (American Journal of Physiology, XXXII, 136-44, 1913) 20. CARBON DIOXIDE PRODUCTION FROM NERVE FIBRES WHEN RESTING AND WHEN STIMULATED; A CONTRIBUTION TO THE CHEMICAL BASIS OF IRRITABILITY By Shiro Tashiro (American Journal of Physiology, XXXII, 107-35, 1913) [Reprinted from THE BIOCHEMICAL BULLETIN, 1912, i, pp. 372-376. March.] WALDEMAR KOCH Herman Koch, a mining engineer of international reputation, lived in Clausthal, Hannover, Germany. His father, his grand- father and his grandfather's father had been mining engineers be- fore him. One of his nine sons was Robert Koch, the great bac- teriologist ; another was Hugo Koch, also a mining engineer ; another, Arnold Koch, came to this country in 1867 with letters of intro- duction from Alfred Nobel, who was a friend of Herman Koch. Arnold Koch settled in St. Louis, where his only son, Waldemar* was born April 8, 1875. The first part of Dr. Koch's college life was spent in Washington University, St. Louis, but his last year he spent in Harvard, from, which he received his undergraduate degree, and two years later, in 1900, the degree of Ph.D. in organic chemistry. He was then for one year, assistant in physiology in the Harvard Medical School with Professor Porter. He began at that time the study of the chem- istry of the nervous system, which he continued until his death. He came to the University of Chicago as an associate in physiological chemistry in 1901, and with a short interregnum spent in teaching" pharmacology and physiological chemistry in the University of Mis- souri, he remained in the University of Chicago continuously there- after, where at the time of his death he was associate professor of pharmacology. He was for a time with Schmiedeberg in Strass- burg; and during his vacations he worked for several years, part of the time under a grant from the Rockefeller Institute, in the labo- ratory of Dr. Mott in the Claybury Asylum for the Insane, near London; afterwards for one season he was on the staff of the new hospital for the insane at Long Grove, near London; and for the past year he had been connected, also, with the Wistar Institute of Anatomy in Philadelphia. In these various institutions he had un- usual opportunities, which he utilized to the utmost, for the study of pathological and normal nervous material. Dr. Koch's work on the chemistry of the nervous system is 372 373 Waldemar Koch [Mar. £> a;ll physiological chemists. The chemistry of the brain is 'a very 'difficult field and the separation of the lipoid substances is stifl hardly possible. He spent the greater part of these years in devising accurate methods of quantitative analysis. These methods had been so perfected that he had secured more complete and more accurate quantitative analyses of nervous tissue than any hitherto made. By means of these methods he was attacking the problem of the differentiation of the brain during growth; the distribution of various substances in different parts of the brain; variations in composition during disease; and the differences between the brains of different animals. He was also engaged in separating carefully the various lipoids, such as kephalin and lecithin, and in examining their composition. He showed the very important fact that kepha- lin exists as a potassium salt, whereas lecithin has more of an affinity for sodium. His method of purification of the lipoids by precipitation with chloroform and hydrochloric acid was extremely useful. Among his other important contributions must be men- tioned his work on the behavior of lecithin and kephalin emulsions towards various salts, anesthetics and drugs, work which showed one way in which these substances might influence irritability. He discovered, also, that the brains of persons having the very obscure insanity, dementia praecox, contained less of a certain sulfur frac- tion than usual, and in his further examination of the sulfur dis- tribution in the brain he isolated a lipoid-sulfur compound of very interesting nature. He had prepared a considerable quantity of this compound and he was engaged in studying its nature at the time of his death. Dr. Koch's interest from the first had been in the problem of the action of drugs on the nervous system. He taught pharma- cology almost from the time of his graduation. He fitted himself for his duties as a teacher by taking the regular medical courses in pathology, anatomy, embryology and many of the clinical courses, as well as by studying with Schmiedeberg in Germany. He thus had a very unusually broad training and was able to look at his subject from all sides, the chemical, the biological, and the clinical. There are very few men in pharmacology to-day, who possess his extensive knowledge and his sane, scientific and broad point of view. i9i2] Albert P. Mat hews 374 The work he was doing on the nervous system he regarded as fundamental work in pharmacology, necessary before any real sci- ence of pharmacology could be constructed. He had just reached a point when the direct application of his methods to the solution of the problem of how drugs combine with nerve cells could be begun. To die at such a time was particularly cruel. Had he lived a few years longer his recognition as one of the leading pharmacologists of the world would undoubtedly have been assured. His work, like that of his uncle, Robert Koch, was very thorough and exact, and he proceeded in logical order to overcome one difficulty after another. Dr. Koch's personality won him many friends. He was a true, loyal and courageous friend, entirely honest and of sane judgment; he avoided making enemies, as far as possible. He was fond of out of doors, and loved the hills, rivers and fields, and tramping in the dunes near Chicago was his main recreation. He had a keen appreciation of what was fine in music and in art. His was an open, frank, kindly nature, considerate of others and slow to anger. His death by pneumonia, February ist, at the early age of thirty-six, was an irreparable loss to his friends, to his University and to science. PAPERS PUBLISHED BY DR. WALDEMAR KOCH KOCH (and C. LORING JACKSON). Die Einwirkung des Jods auf das Bleisalz des Brenzcatechins. Berichte der deutschen chemischen Gesellschaft, 1898, xxxi, 1457. (and C. LORING JACKSON). On the action of sodic ethylate on tribromdinitrobenzol. Proceedings of the American Academy of Arts and Sciences, 1898, xxxiv, 119. (and C. LORING JACKSON). On certain derivatives of ortho- benzoquinone. Ibid., 1900, xxxvi, 1*97. The physiological action of formaldehyde. American Journal of Physiology, 1902, vi, 325. Zur Kenntniss des Lecithins, Kephalins, und Cerebrins aus Ner- vensubstanz. Zeitschrift fiir physiologische Chemie, 1902, xxxvi, 134. The lecithans. Their function in the life of the cell. Decennial Publications, Chicago (University of Chicago Press), 1902, x, 93. Some corrosions found on ancient bronzes. Science (n. s.), 1903, xvii, 152. 375 Waldemar Koch [Mar. Die Lecithane und ihre Bedeutung fiir die lebende Zelle. Zeit- schrift fiir physiologische Chemie, 1903, xxxvii, 181. Methods for the quantitative chemical analysis of the brain and cord. American Journal of Physiology, 1904, xi, 303. — On the origin of creatinin. Ibid., 1905, xiii, 19. — On psychic secretion. Ibid., 1905, xiii, 37. - Relation of creatinin excretion to variations in diet. Ibid., 1905, xv, 15. On the presence of a sulphur compound in nerve tissue. Science (n. s.), 1905, xxi, 884. Lecithin in infant feeding. St. Louis Courier of Medicine, 1905, xxvi, June. A study of the metabolism of the nervous system. Ibid., 1906, xv ; Proceedings of the American Physiological Society p. xv (December, 1905). (and W. H. GOODSON). A preliminary study of the chemistry of nerve tissue degeneration. Ibid., 1906, xv, 272. - (and HERBERT S. WOODS). The quantitative estimation of the lecithans. Journal of Biological Chemistry, 1906, i, 203. tiber den Lecithingehalt der Milch. Zeitschrift fiir physio- logische Chemie, 1906, xlvii, 327. (and HOWARD S. REED). The relation of extractive to protein phosphorus in Aspergillus niger. Ibid., 1907, iii, 49. The relation of electrolytes to lecithin and kephalin. Ibid., 1907, iii, 53. The quantitative estimation of extractive and protein phos- phorus. Ibid., 1907, iii, 159. Some chemical observations on the nervous system in certain forms of insanity. Archives of Neurology, 1907, iii, 331. Zur Kenntniss der Schwefelverbindungen des Nervensystems. Ibid., 1907, liii, 496. — (and S. A. MANN). A comparison of the chemical composition of three human brains at different ages. Journal of Physiology (English), 1907, xxxvi; Proceedings of the Physiological Society, p. xxxvi (Nov. 23). - Phosphorus compounds as brain foods. Journal of the Amer- ican Medical Association, 1909, Hi, 1381. - (and S. A. MANN). A chemical study of the brain in healthy and diseased conditions, with special reference to Dementia praecox. Archives of Neurology and Psychiatry, 1909, iv, 31. • Albert P. Mathews 376 • Methods for the quantitative chemical analysis of animal tissues. i. General principles. Journal of the American Chemical Society, 1909, xxxi, 1329. (and S. A. MANN). 2. Collection and preservation of material. Ibid., 1335. (and EMMA P. CARR). 3. Estimation of the proximate constit- uents. Ibid., 1341. (and F. W. UPSON). 4. Estimation of the elements, with special reference to sulphur. Ibid., 1355. — — Die Bedeutung der Phosphatide (Lecithane) fiir die lebende Zelle. II. Zeitschrift fiir physiologische Chemie, 1909, Ixiii, 432. (and F. W. UPSON). The distribution of sulphur compounds in brain tissue. Proceedings of the Society for Experimental Biol- ogy and Medicine, 1909, vii, 5. Phosphorus compounds as brain foods. Journal of the Amer- ican Medical Association, 1909, Hi, 1381. Zur Kenntniss der Schwefelverbindungen des Nervensystems. II. tiber ein Sulfatid aus Nervensubstanz. Zeitschrift fiir physiologische Chemie, 1910, Ixx, 94. Pharmacological studies on the phosphatids. i. Methods for the study of their combinations with drugs and other substances. Journal of Pharmacology and Experimental Therapeutics, 1910, «, 239. (and F. H. PIKE). 2. The relation of the phosphatids to the sodium and potassium of the neuron. Ibid., 245. (and F. C. MCLEAN). 3. The relation of the phosphatids to Overton and Meyer's theory of narcosis. Ibid., 249. — (and A. W. WILLIAMS). 4. The relation of brain phosphatids to tissue metabolites. Ibid., 253. — (and H. T. MOSTROM). 5. The function of the brain phospha- tids in the physiological action of strychnin. Ibid., 265. — Recent studies on lipoids. Journal of the American Medical As- sociation, 1911, Ivi, 799. ALBERT P. MATHEWS. University of Chicago. STUDIES ON THE PHYSICAL PROPERTIES OF PROTOPLASM Reprinted from the American Journal of Physiology Vol. XXXII -June 2, 1913 -No. II STUDIES ON THE PHYSICAL PROPERTIES OF PROTOPLASM I. THE PHYSICAL PROPERTIES OF THE PROTOPLASM OF CERTAIN ANIMAL AND PLANT CELLS BY G. L. KITE [From the Hull Laboratories of Biochemistry and Pharmacology, University of Chicago] INTRODUCTION \ LTHOUGH the living substance of animal and plant cells -^A- was correctly interpreted by Dujardin and von Mohl in the second quarter of the nineteenth century, almost nothing is definitely known of the physical state of protoplasm. Properties described by such adjectives as glutinous, slimy and hyaline were recognized by the early micro^copists, who were forced to study liv- ing cells and tissues. During the last fifty years an extensive literature has grown up on the subject of the structure of protoplasm. For the purposes of this paper, these investigations may be divided into two groups. The first group comprises those studies on the structure of protoplasm, made with the aid of fixatives. A large part of our knowledge of the morphology of the cells and tissues of animals and plants is the direct result of the development of fixing methods. The errors involved in the attempts to determine the true molar structure of protoplasm by the use of fixing reagents have been pointed out particularly by Flemming,1 Berthold,2 Schwarz,3 Fischer,4 and Hardy.5 In this connection, it will suffice to state that Hardy's con- clusion that fixing reagents always cause structural changes in pro- 1 FLEMMING: Zellsubstanz, Kern und Zelltheilung, 1882, Leipzig. 2 BERTHOLD: Studien iiber Protoplasmamechanik, 1886, Leipzig. 3 SCHWARZ: Cohn's Beitrage zur Biologic der Pflanzen, 1887, v, p. i. 4 FISCHER: Archiv fur Entwicklungsmechanik, 1901, xiii, p. i. 5 HARDY: Journal of physiology, 1899, xxiv, p. 158. Physical Properties of Protoplasm 147 toplasm that are frequently very different from the normal living substance, has never been refuted. Hence, at least, it does not seem that more than an approximation of the actual structure of pro- toplasm can be attained by the use of fixatives. Besides, this method is worthless as a means of investigating the physics of protoplasm. The papers that fall in the second group deal with the study of living cells. Strassburger,6 Wilson,7 Foot and Strobell,8 Lundegardh9 and others have shown that many of the structural elements of the mi to tic figure can be seen in living animal and plant cells. Numerous investigators have pointed out the presence of granules, vacuoles, and fibrils in various types of unfixed cells; but for the most part, the studies on living cells have been made for the purpose of decreas- ing the error due to the use of fixing reagents. All such investigations are open to several sources of error. Hardy 10 writes that the process of dying produces structural changes in the cell substance, since coagulation appears to occur in all dying cells. Many cells are certainly quickly asphyxiated when mounted for microscopical examination in the usual manner. I have been able to overcome, largely, this source of error, by the use of an open- end moist chamber, that does not appear to interfere with normal respiration. A second source of error is due to the nature of the optical prin- ciples involved in microscopical vision. Many years since Abbe u demonstrated that the optical image is a diffraction pattern produced by the object and that under certain conditions the image may be quite different from the object. More recently, Porter,12 experi- menting under ordinary working conditions, has described a number of interesting examples of this sort. Porter12 says, "Images were formed which were utterly false in their smaller details, and other images were profoundly modified by the presence of structure lying 6 STRASBURGER: Zellbildung und Zelltheilung, Jena, 1880, iii. Auflage. 7 WILSON: Journal of morphology, 1899, xv, Suppl. 8 FOOT and STROBELL: American journal of anatomy, iv, p. 199. 9 LUNDEGARDH: Jahrbiicher fur wissenschaftliche Botanik, li, p. 236. 10 HARDY: Journal of physiology, 1899, xxiv, p. 158. 11 ABBE: Archiv fur mikroskopische Anatomic, 1874, ix, p. 413; Gesam- melte Abhandlungen, 1904, i, p. 45. 12 PORTER: Philosophical magazine, 1906, xi, p. 154. i48 G. L. Kite entirely beyond the focal plane." Such facts should serve to make it evident that one can easily fall into error in interpreting the optical image of a living cell. Porter12 recommends " a working knowledge of the phenomena and laws of diffraction," as a safeguard against this form of error. The third and by far the most important source of error is due to the peculiar and little known optical properties of living matter. The phenomena of reflection, refraction, absorption, dispersion, inter- ference, diffraction 13 and a scattering action on light 14 are all exhib- ited by this substance, with the result that a correct interpretation of the image of a living cell is frequently impossible. Further- more, many cells are so opaque and turbid that the interior is not visible. Cloudiness or turbidity is almost a universal property of protoplasm and appears to be due chiefly to dispersion, refraction, diffraction, and the scattering action on light of the colloidal par- ticles which may be considered as the real structural units of all protoplasm. Globules, granules and cell walls frequently show diffraction halos that are difficult to interpret in undissected cells. The aim of this investigation is to determine the physical state and the molar structure of protoplasm. The methods are radically different from those heretofore used, and are believed to be adequate for this purpose. Dissection and vital staining are used to deter- mine the truthfulness of the optical image and the actual structure of cells. Unfortunately, the amount of the error involved in the employment of these methods depends entirely on the skill of the experimenter; but it is believed that the error becomes quite small with complete mastery of the methods. The structural changes that cells may undergo during the time 13 Excellent expositions of the principles of physical optics are given by: WOOD, R. W: 1911, Physical Optics; DRUDE, P.: 1912, Lehrbuch der Optik; PRESTON: 1901, Theory of Light. 14 Lord Rayleigh (Philosophical magazine, xli, p. 107) has pointed out that reflection and refraction have no application unless the surface of the disturbing body is larger than many square wave-lengths. The turbidity of protoplasmic sols, then, is due entirely to the scattering action on light of the minute aggre- gates of the disperse phase, while reflection, refraction, diffraction, dispersion and a scattering action on light are all seemingly involved in the production of turbidity by gels. Physical Properties of Protoplasm . 149 required for their dissection is a possible source of error that may appear, at first sight, to be very difficult to control. Certainly many biologists hold the view that cells rapidly undergo important morphological changes following mechanical injury. With few exceptions it has not been found difficult to follow the structural changes that occur in cells that are being dissected; but the really remarkable fact is the marked slowness of such death changes as granulation, fragmentation and general coagulation, following mechanical injury. It seems best to limit this introductory paper to a description of selected types of widely different cells and in future publications to treat systematically, selected types of the principal phyla of animals and the chief groups of plants. The special literature bearing -on this investigation will be dis- cussed in subsequent papers. , A review of such well-known theories as those of Butschli, Flemming and Altmann lies outside the province of this paper. METHODS AND MATERIAL The development of a really adequate method for the dissection of living cells, under the highest powers of the microscope, has made possible this study. The principles of this method are simple. The dissecting instrument is a glass needle that may measure less than one micron in diameter and is drawn on the end of a piece of special Jena glass tubing about 200 mm. long and 4 mm. in diameter. The needle is held in a three-movement Barber pipette holder. The cell chosen for dissection is mounted in a hanging drop in an open-end moist chamber and held in place by water-glass surface tension, which can be varied at will. Both diffuse sunlight and artificial light are used as sources of illumination. For the latter a Nernst Glower has been found satis- factory, but all the light waves outside of 450 and 670 p>p* are cut out by the use of appropriate ray screens. The same means is used to remove enough of the orange and red rays to make the transmitted light perfectly white. Such light is composed of the waves that are least injurious to living cells. A special condenser 15 of a focal dis- 15 The condenser was made by E. Leitz & Co., Wetzlar. G. L. Kite stance of about 20 mm., a 2 mm. apochromatic objective, compensat- ing oculars, and a number of vital stains, are necessary additions. An open-end moist chamber 25x60x15 mm. has been found satisfactory. The bottom is separated into three compartments by very small glass rods and water is placed in the end compartments. If water be put in the middle compartment it may decrease the efficiency of the condenser. The chamber is held in a mechanical stage and most of the dissections are made by quick movements of the chamber and therefore of the cell being dissected, the needle remaining fixed. The use of acetylene which can be burned in a glass micro-burner has greatly simplified the making of extremely fine needles. An acetylene flame that is so small that it is invisible in a well-lighted room can be kept alive. The more important of. the vital stains used include methylene blue, new methylene blue N (Cassella Color Co.), new methylene blue GG (Cassella Color Co.), new methylene blue R (Cassella Color Co.), janus green (Metz & Co.), pyronin (Griibler), vusuvin (Griibler), toluidin blue (Griibler), neutral red (Griibler). The chief structural components of a cell can usually be quickly brought out by using a large enough number of vital stains, thus effecting a great economy of time, when the dissection of the unstained cell is made. Barber's 16 isolation and intracellular injection methods, vari- ously modified, are frequently employed to supplement and control the data obtained by dissection. Nomenclature Employed. — The current nomenclature of de- scriptive physics, physical optics and colloidal chemistry will be employed. Such physical properties as solidity, tenacity, elasticity, hardness and viscosity have been determined for the cells, so far studied. In general, the term viscosity will be used to designate the degree of rigidity of protoplasmic structures, but such a structure as a vitelline membrane may be comparatively soft and yet have what must be considered as a high internal friction or viscosity. Elasticity is determined by transfixing a selected piece of a cell and stretching it and observing the power of resumption of the original form. Dis- 16 BARBER: University of Kansas Science Bulletin, 1907, iv, p. 3;. Journal of infectious diseases, 1911, viii, p. 248; ibid., 1911, ix, p. 117. Physical Properties of Protoplasm 151 section is the method employed for determining such properties as solidity, hardness and tenacity or cohesiveness. All physical proper- ties that have been enumerated are relative and it is hoped at a later time to increase the accuracy of description by the selection of arbi- trary standards. The usage of the terms employed in this paper is based on the dissection of many widely different types of animal and plant cells. Living matter occupies an intermediate position between true solids and true liquids and has many of the properties of both as well as properties peculiar to itself. It belongs to the class of colloids known as emulsoids and exists in either a gel (hydrogel) or a sol (hydrosol) state.17 The term gel will be used to designate the amor- phous semi-solid state and sol the apparently homogeneous liquid state, of living substance. Protoplasmic sols usually appear as hazy homogeneous liquids on account of the very minute size of the protein aggregates that compose the solid phase. On the other hand pro- toplasmic gels are characterized by the large size of the particles of the solid phase which set to form the gel. Hence, living gels may exhibit either a homogeneous or heterogeneous molar structure. It should now be clear that the term homogeneous is used in a relative sense to describe the optical image and refers only to the molar structure that can be brought out by the usual microscopical powers and further that heterogeneity is the universal distinguishing characteristic of colloidal sols and gels. In this connection it may be noted that Pauli18 states that the " unfixed" gel of gelatine is not structured in the sense of being composed of threads, networks, granules and vacuoles; it has the molar structure of a one-phase sys- tem,'which is precisely what is meant by the term homogeneous as used in this paper; the molecular structure is unknown. The present unsettled state of the problem of phase relations of colloidal 17 For discussion of the classification of colloids see: No YES, A: 1905, Journal of the American Chemical Society, 1905, xxvii, p. 85; OSTWALD, Wo.: 1907, Zeitschrift fur Chemie und Industrie der Kolloide, 1907, i, p. 291; PERRIN, J.: 1905, Journal de la chemie physique, iii, p. 50; FREUNDLICH and NEU- MANN, 1908, Kolloid Zeitschrift iii, p. 80; VON WEIMARN, P. P.: ibid., 1908, iii, p. 26. 18 PAULI: Der Kolloidale Zustand und die Vorgange in der lebendigen Sub- stanz, Braunschweig, 1902. 152 G. L. Kite solutions has been ably discussed in a recent paper by Hardy.19 It is usual to regard colloidal systems as consisting of two phases, a solid and a liquid, which have been termed by Wo. Ostwald 20 the dis- perse phase and the dispersion medium, respectively. For convenience of description arbitrary meanings will be given the terms microsome and globule; the former will be restricted to minute dense masses of gel, the latter to suspensions in protoplasm that show many of the physical properties of oil droplets and besides are usually free of protoplasm when dissected out of a cell. Most of the suspensions so far found in cells fall into one or the other of these groups, but intermediate forms have been observed. THE EGG or ASTERIAS The egg of Asterias is surrounded by a mass of either transparent or translucent jelly which is soft and somewhat elastic and glutin- ous; but it can be cut and torn to pieces and removed from the egg with little difficulty. Thirty-four and six-tenths microns is the average thickness of this jelly. This structure has a low viscosity for a gel and is therefore extremely dilute. On many eggs, the jelly has become turbid and undergone a change in refractive power and as a result is visible in the usual microscopical examination. The inner surface of the jelly envelope is closely applied to the outer surface of the vitelline membrane which is invisible except in eggs that have maturated. The vitelline membrane of the immature starfish egg is a transparent and invisible solid of about two microns in thickness. The physical properties of this structure are • very definite since it exhibits extraordinarily high viscosity, elasticity and tenacity. A small piece can be drawn out into a mere thread and when freed the thread contracts to a more or less rounded mass. During maturation the vitelline membrane swells to two and three times its original thickness, undergoes a change in refractive index, and becomes quite cloudy and hence visible. In this state it is softer, more glutinous and less rigid. The inner surface of this 19 HARDY: Proceedings of the Royal Society, Series A, 1912, Ixxxvi, p. 601. 20 OSTWALD: Zeitschrift fur chemie und Industrie der Kolloide, 1907, i, p. 291. Physical Properties of Protoplasm 153 membrane is tightly glued to the surface of the cytoplasm, from which it can be dissected only with considerable difficulty. The misleading optical phenomena that are involved in a study of the cytoplasm are of great interest. It is usual for cytologists to consider the echinoderm egg a classi- cal example of the alveolar structure of protoplasm. No one can question the fact that beautiful round spaces with hazy, protoplasmic walls in which are embedded minute granules, can be seen in such eggs. Biitschli supposed these spaces to be filled with a watery fluid. What is the true structure of the cytoplasm of the egg of Asterias? Careful dissections give a clear-cut answer to this question. The cytoplasm is a quiet translucent gel of comparatively high viscosity; it can be drawn out into large strands, but is not cohesive and elastic enough to form small threads. It can be cut into small pieces with comparative ease. Fragments usually become spherical, though in some cases water is slowly taken up and the mass changes into the sol state. Minute granules measuring little more than one micron are scattered plentifully throughout the cytoplasmic gel. It has been found impossible to free these structures completely from the gel in which they are embedded. They are optically more dense and have a different refractive index from the surrounding living substance. A part of the total mass of cytoplasm is composed of what appears to be alveoli or spaces; but a careful dissection of such an alveolus reveals the presence of a globule that has many of the optical properties of an oil drop. Such a globule, freed from cyto- plasm, does not dissolve in sea water and in a light of low intensity exhibits the usual diffraction halo. The invisibility of liquid drop- lets of rather high viscosity when embedded in the cytoplasm might at first sight appear difficult to explain. This invisibility is due to the fact that the refractive index and dispersive power of the globules is very near that of sea water; also, the optical density of the cyto- plasm is evidently higher than that of the globule. No diffraction rings could be seen surrounding the globules when they were imbedded in cytoplasm. Centrifugal force dislodges the globules, proving them to be merely suspended in a living gel. The minute granules respond much less readily to centrifugal force. Besides they show optical properties — their index of refraction is certainly higher than that of the surrounding gel — that ally them to highly concentrated particles 154 G. L. Kite of the cytoplasmic gel. Yet it seems likely that all such structures as granules and globules must be considered as having separated out of the disperse phase and to be therefore of the nature of suspensions. The living cytoplasm, then, is an apparently homo- geneous and very viscous gel in which microsomes and globules are suspended. If the nucleus of the immature starfish egg be dissected out in sea water it undergoes no appreciable change. Dissection of the highly- translucent nuclear membrane shows this structure to be a very tough viscous solid, and, in fact, closely allied physically to the vitelline membrane and not at all the delicate structure of the conventional descriptions. With the exception of the nucleolus, the nuclear sub- stance is all in the sol state. The nuclear sol is apparently a homo- geneous liquid. The nucleolus is a small mass of quite rigid and cohesive granular gel that is suspended in the nuclear sol. The polar body is a granular, elastic and highly viscous gel. In order to make it possible to observe the structural components of the starfish egg and of the eggs of other common marine inverte- brates, without having to use my tedious methods, vital staining was resorted to. The jelly envelope can be stained a beautiful light blue with dimethyl-safranin-azo-dimethyl-anilin; the vitelline membrane a very dark blue with isamin blue; the globules or droplets from yellow to orange with vusuvin; and the extremely small granules a slate blue with diethyl-safranin-azo-dimethyl-anilin. The dead or dying asterias egg shows remarkable morphological changes. The whole egg becomes almost opaque. The cytoplasm separates into a large number of more or less rounded masses which still adhere to each other. Such masses vary greatly in size, some being as small as five microns in diameter. If the formation of such small masses be observed, one is easily misled into believing that fusion of the globules is occurring. Dissection of such a mass frees the original globules. The dead gel does not stick to a glass needle and can no longer be drawn out into strands; it has lost much of its viscidity and cohesiveness. The nuclear fluid has set and the result- ing gel is more voluminous than was the nuclear fluid in the living egg. The nuclear membrane shows little change in its physical properties, while the nuclear gel is elastic and quite viscous and granular. The physical properties of the dead nuclear gel are very similar to those Physical Properties of Protoplasm 155 exhibited by the living cytoplasm. Small fragments of the dead nuclear gel do not go into solution when dissected out in sea water. AMEBA PROTEUS Small pieces of ectoplasm of proteus can be cut off in distilled water and show no change. This living substance has a moderately high viscosity and cohesiveness ; it does not stick to glass needles very readily and little difficulty is experienced in cutting it into pieces as small as the limit of microscopical visibility. Pieces of all sizes appear perfectly homogeneous. The cloudiness of the ectoplasmic gel is a well-known property. The inner three or four microns of the hya- line ectoplasm and particularly the interior of the outer end of small pseudopods, contain varying numbers of minute granules and glob- ules that may measure as much as four or five microns. If these gran- ules and globules are dissected out they do not go into solution. The globules show confusing diffraction rings ; but, both globules and granules can be brought out by light staining with diethyl-safranin- azo-dirnethyl-anilin. The endoplasm contains a large contractile vacuole in which the presence of protein has not been demonstrated, as yet, and numerous food vacuoles which contain either liquid or liquid and food masses. The same kind of granules and globules are found in the endoplasm as are found in the ectoplasm and the number of these structures varies in different animals. The sub- stance forming the walls of the vacuoles is of much higher viscosity and cohesiveness. The living endoplasmic substance is a very dilute and apparently homogeneous gel that possesses a remarkable affinity for water. The ectoplasm of ameba then is a quite concentrated gel while the interior is quite dilute and is continously changing its water- holding power in different regions. New methylene blue R and trypan blue are of great value in bringing out the globules, granules and vacuoles. The nuclear membrane is an extremely thin and moderately tough solid substance. It shows some elasticity and is quite viscous. The whole of the nuclear substance is a highly rigid and granular gel, the minutest pieces of which show no appreciable change when dissected out in distilled water. A slight elasticity and a definite G. L. Kite glutinicity are exhibited by this substance. There are variations in concentration of the nuclear gel that produce a characteristic but misleading optical image. The nucleus appears to contain an irregu- lar "network with granules imbedded in it. The interstices of the network are very small luminous spots which have been misinter- preted to be vacuoles. Many dissections have shown that the granules are very concentrated masses of gel; the network irregularly disposed masses of a diluter gel; and the interstices or light spots the most dilute gel in the nucleus. The so-called network is a part of the nuclear gel that forms a concentration gradient; the interstices and granules may be considered constants connected by the grading network. It should be clearly understood that the network is not made up of definite threads of fibres but of irregular masses of hydrogel that are very dense immediately surrounding the granules, from which they grade into the dilute gel of the interstices. No free liquid was found in the nuclear substance. When the granules are in focus they appear gray and cloudy or opalescent; when out of focus as dark spots. They measure from less than one to about two microns in diameter. It seems that a part of the luminosity of the interstices of the network is due to diffraction and not simply to slight absorption of light by this portion of the nuclear substance. The structural details of the nucleus can be brought out with con- siderable vividness by staining with janus green (diethyl-safran — in azo-dimethyl-anilin) . Slight cuts in the surface of proteus qukkly close. Extensive cuts frequently cause an ameba to explode — in as short a time as two seconds nothing but the nucleus may remain. If the contrac- tile vacuole be cut and its liquid content caused to mix with the cytoplasm the Ameba is immediately destroyed with explosive vio- lence. A relatively large dose of distilled water and even | to i molar cane sugar solution or one molar sodium chloride or potassium nitrate give a like result. It is not usually possible to produce more than a temporary vacuole with two molar cane sugar; a large dose of sugar of this concentration usually causes the appearance of granules, globules, fibrils and a hyaline appearance in any portion of the endo- plasm into which the injection is made. The doses that were injected varied from about 270 cubic microns to 30,000 cubic microns. Physical Properties of Protoplasm 157 A large number of indicators have been injected into the interior of proteus with the idea of determining a possible relation between an excess of H+ or OH~ ions and the extraordinary water-holding-power of the endoplasm. Azolitmin, sodium alizarin sulphonate, tropeolin ooo No. i, methyl orange and congo red, dissolved in from J to f molar cane sugar have been so far employed. A neutral to slightly alkaline reaction is shown by all the indicators. It seems probable then that the concomitant variation in water-holding-power of dif- ferent regions of the cytoplasm is the mechanism by which Ameba proteus moves and is associated with an excess of OH~ ions. A number of operations were performed on the ectoplasm of Ameba proteus for the purpose of determining the relation between movement and surface tension changes. The results of shallow and deep cuts in the ectoplasm have already been given. The outer 5 to 7 microns of the pseudopods were cut away in some animals, and in others small doses of distilled water were injected into the ectoplasm. The removal of the outer end of a pseudopod was usually followed by rapid closure of the incision. The injection of distilled water into the ectoplasm had no noticeable effect on the formation of pseudopods. By means of such operations the rigid ectoplasm was either removed, for a short time, from a given area of the surface or at least greatly weakened; yet, no tendency to the formation of pseudopods was ever observed, in such weakened surface areas. These facts seem to justify the conclusion that surface-tension changes play a negligible role in the movement of Ameba proteus. Furthermore, it may be recalled, that the outer surface of Ameba proteus is a semi-rigid solid of from 5 to 12 or more microns in thickness, and it has still to be shown, that the changes, in the tension of the surface film, that are commonly assumed to occur, can appreciably affect the underlying semi-rigid ectoplasm. The nutrient solution in which the amebae were grown was slightly alkaline in reaction. Proteus usually recovers from the large doses of neutral salts and sugar in much less than an hour, almost certainly by throwing them off. 158 G. L. Kite PARAMECIUM The living substance of Paramecium is a soft, elastic and somewhat glutinous gel which can be drawn out into strands. It is filled with a large number of vacuoles of various sizes the walls of which are more dense than the surrounding gel. The surface layer is more viscous and cohesive than the interior. Small cuts usually close quickly, exten- sive deep cuts are either followed by a loss of cytoplasm or a rapid change of the whole cytoplasm into the sol state with almost explosive violence. If the fluid in the contractile vacuole be caused to mix with the cytoplasm a rapid change of this substance into the sol state results. Suspended in the living and apparently homogeneous and rather dilute gel are varying numbers of extremely small granules and small globules. Many of the granules are recently ingested bacteria. Neither the granules nor globules go into solution when dissected free from the cytoplasm. The food masses are granular gels of rather high viscosity. The optical properties of the meganucleus render its study ex- tremely tedious. It is almost transparent and invisible. Therefore its refractive index and its dispersion are very close to those of water. Dissection has proved the meganucleus to be a gel of higher viscosity than the cytoplasm and to be slightly glutinous and elastic. The meganuclear gel has areas, more dense than the surrounding sub- stance, that may be considered granules. A complete study of the micronucleus has not been made. NECTURUS The Striped Muscle Cell. — The. living substance of the striped muscle cell of Necturus is the most viscous, elastic and cohesive of the living gels we have so far considered. The muscle substance sticks to a glass needle and can be drawn out into extraordinarily long threads which when released almost regain their previous shape. The absorptive power and turbidity of this substance are compara- tively high. When the whole or a piece of a muscle cell is stretched the stria- tions become faint or disappear — • only to reappear when the tension Physical Properties of Protoplasm 159 is removed. Beautiful but misleading diffraction phenomena are to be observed when a piece of muscle cell is stretched. If the point of a very minute needle be pushed into a muscle cell, it can be moved in one direction about as easily as another. The optical image of striped muscle is very misleading. Dissec- tions have shown that the dark bands seen in living muscle are pro- duced by concentrated areas of muscle substance which absorb enough transmitted light of low intensity to appear as dark bands in the optical image. I have been unable to dissect out definite fibrils. The substance lying between the concentrated regions and appearing as light bands is a highly viscous, elastic gel and has no physical properties that serve to distinguish it from the surrounding sarco- plasmic gel. By cutting the dark band to pieces, small masses of highly concentrated muscle substance, frequently less than one micron in diameter, are partially freed from the dilute enveloping gel and in light of low intensity show well-defined diffraction halos. The appearance of dark bands in the optical image, then, is produced by absorption of light waves by the concentrated muscle substance; the light bands, by the low absorptive power of the diluter inter- mediate gel, and the diffraction of the light waves by the edges of the concentrated substance. Striking changes in the optical image that are well known can be produced by increasing the intensity of illumi- nation. The dark band becomes cloudy and more or less opalescent and the light band may show an intersecting dark line or well-defined diffraction fringes just outside the geometrical shadow of the con- centrated substance. Hence, absorption, diffraction, refraction and dispersion are involved in the formation of the optical image of striped muscle and the former two particularly when the illumination is of a relatively high intensity. The nuclear substance is a gel that is for the most part compara- tively dilute but contains more concentrated areas in the form of granules and an imperfect network. The appearance of a network in the optical image is due not to definite fibrils but to more con- centrated parts of the gel that grade into the diluter nuclear substance. On the outer surface of the muscle cell is found a highly trans- lucent membrane, the sarcolemma, which is extremely elastic and measures about one micron in thickness. It is stuck to the whole outer surface of the muscle cell and is viscous and cohesive enough 160 G. L. Kite to offer an appreciable resistance to a glass needle a micron or less in diameter. The disagreement among investigators concerning the presence of a sarcolemma is due to the fact that it is transparent and that its refractive and dispersive powers are so nearly the same as those of water. Instead of being the delicate structure of the con- ventional descriptions, the sarcolemma of the striped muscle cell of Necturus exhibits physical properties that are very similar to those of the vitelline membrane of an echinoderm egg. If a concentrated solution of isamin blue made by boiling in dis- tilled water or .8 per cent sodium chloride be added to freshly teased muscle cells, blue staining of the sarcolemma occurs in ten to fifteen minutes. An Epidermal Cell. — The epidermal cells are embedded in an intercellular gel of extremely high viscosity and considerable elas- ticity. The substance is tough but softer than many nuclear mem- branes and shows a relatively high absorptive power. It is also quite turbid. A few globules and granules, varying in size from about one to four microns, that can be easily stained with diethyl-safranin- azo-dimethyl-anilin are to be seen scattered through the intercellular gel- The whole cell substance is a gel of even higher rigidity than the muscle substance of the same animal. Small pieces cut out of the nucleus or cytoplasm, in distilled water or .8 per cent sodium chloride, show no appreciable change. The cytoplasm exhibits a high absorptive power and a definite elasticity. Very small granules that seem to be denser cytoplasmic areas are to be seen scattered throughout the turbid cytoplasm. Many cells show radially arranged fibrils, in the outer part of the cytoplasm, which can be partially freed from the surrounding gel by dissection. Such a fibril is physically and optically more dense than the remainder of the cytoplasm. The nuclear membrane is thin, clear, and* quite cohesive and elas- tic, and 1ms a different index of refraction from the cytoplasm and nucleus. The nuclear gel is of a higher viscosity than the cytoplasm. The appearance of a network in the optical image of the nucleus is due to concentrated areas in the form of granules and imperfect threads which are not sharply separated from, but grade into, the surrounding diluter Physical Properties of Protoplasm 161 gel. The whole nuclear substance is quite glutinous. No trace of free liquid could be found in the nucleus. SPIROGYRA The cellulose wall of Spirogyra is enormously cohesive; it is cut or punctured with extremely fine Jena glass needles with considerable difficulty. The outer surface is covered by an almost invisible soft gel, that frequently measures five or more microns in thickness and can be stained red with sodium alizarin sulphonate in a neutral or slightly alkaline solution. A layer of dilute granular gel covers the inner surface of the cellulose wall and is connected by a number of strands of an elastic gel to a central mass of living substance, in which a small nucleus is imbedded. The central mass of gel contains a few granules and is of a higher viscosity and cohesiveness than the surface cytoplasm. This mass also has a higher refractive index and higher absorptive power than the surface cytoplasm. The anchoring strands of gel decrease in viscosity from within outwards. Much of the surface layer of cytoplasm is usually invisible. Hence, it is quite translucent and has refractive and dispersive powers very close to those of water. If the cell wall be cut across the surface cytoplasm shrinks. The chloroplasts either shrink or separate into rounded masses. The chloroplasts have a higher viscosity and elasticity than the gel in which they are imbedded. The pyrenoid is a complex structure. Dissection shows the presence of an optically dense but fragile wall which, when broken, frees a globule that is of considerable interest. This globule shows many of the optical properties of an oil droplet but has too high a viscosity to round up under the influence of surface tension; there- fore it seems to be a true gel. None of the cytoplasm goes into solution very readily even when cut into very minute pieces. The nucleus of Spirogyra is a gel that has higher viscosity and refractive and absorptive powers than the cytoplasm. It is also more cloudy than the cytoplasm. There are denser areas in the nuclear substance in the form of granules and threads that form a sort of network. Small pieces dissected from all parts of the nucleus 162 G. L. Kite into water, not only do not go into the sol state but remain too rigid to show surface tension effects. Pieces of broken glass needles stick firmly to the nuclear gel when imbedded in it. The image of the nucleus is false in important details. The denser areas, when in the focal plane, appear as grayish or slightly opalescent . granules and threads and when above or below the focal plane as dark spots and lines. Besides, if the intensity of the illumination be increased the network appears much finer. Very small dense masses of gel could be partly freed from the remaining nuclear substance. It seems proper to term such structures granules. On the other hand, the dense masses that produce the appearance of a network in the image are not actual threads that are sharply separated from the surrounding gel but irregularly shaped dense areas that grade into the immediately contiguous diluter gel. The light spots that change their position at different focal planes seem to be due chiefly to two factors, viz., a relatively low absorptive power of the gel occupying the interstices of the network and diffraction by the edges of the denser areas. It seems certain that the vacuolar fluid of Spirogyra contains protein and must be considered a hydrosol. Much evidence has been adduced in support of this statement. A number of injections of Millon's fluid into the vacuole were made with positive results. Extremely small solid particles appeared in the cell sap after the injection of such precipitating agents for proteins, as saturated subli- mate, 40 per cent formaldehyde, saturated picric acid and saturated phosphotungstic acid containing 5 per cent sulphuric acid. The vacuolar fluid is cloudy. This is positive proof of the pres- ence of ultramicroscopic particles which would ordinarily be con- sidered protein even in the absence of a positive color test for protein. The cell sap of Chara seems to be richer in protein than that of Spirogyra. This conclusion is based on the fact that a compara- tively heavy precipitate results from the in tra vacuolar injection of saturated sublimate or 40 per cent formaldehyde. Hence, it is proba- ble that cell sap containing protein is very common in plants. Mucor, Saprolegnia, Hydrodictyon, Chara and the parenchymatous cells of the leaves of Tradescantia have been dissected for comparison with animal cells. In general, it may be stated that the cellulose walls of plants are extremely cohesive and are cut and punctured Physical Properties of Protoplasm 163 with considerable difficulty. The protoplasm of plant cells is much more dilute or less rigid than that of animal cells. RESTING AND DIVIDING MALE GERM CELLS OF THE SQUASH BUG (ANASA). GRASSHOPPERS AND CRICKETS A brief note has been published on this subject.21 The whole cell substance of resting and dividing spermatogonia and spermatocytes is a moderately viscous gel. Cutting away pieces of the cytoplasm and nucleus in Ringer's fluid shows that these struc- tures are far too rigid to flow or change shape under such experi- mental treatment. The appearance of a network is due to denser masses of nuclear gel that grade into the diluter surrounding substance. No definite threads or fibrils could be dissected out of resting nuclei. Some of the optical principles involved in a study of the living nuclei of spermatogonia and spermatocytes were discussed in connection with the nucleus of proteus. Very definite statements can be made about the physical proper- ties of chromosomes and spindle fibres. The chromosome has been found to be the most highly concentrated and rigid part of the nuclear gel. Such a mass of gel is less translucent and has a higher refractive index and absorptive power than the diluter homogeneous gel in which it is imbedded. A chromosome when dissected out shows no affinity for water and does not disintegrate readily. tPieces of it stick to the glass dissecting needle but when drawn out show no marked elas- ticity. The spindle fibre is an elastic concentrated thread of nuclear gel and its absorptive power and refractive index are also different from those of the diluter gel in which the spindle fibre is imbedded and from which it cannot be entirely freed. Metaphase spindle fibres that were dissected out with great care seemed continuous with the ends of the chromosomes. The homogeneous gel in which a telophase spindle is imbedded is so rigid, that all the surrounding cytoplasm can be cut away and the spindle and chromosomes show no appreciable change; metaphase, anaphase and telophase spindles can be cut to pieces in Ringer's fluid and the pieces are so rigid that they undergo no change in shape. 21 KITE and CHAMBERS: 1912, Science, N. S., xxxvi, p. 639. 1 64 G. L. Kite Many of the physical and chemical changes of cell-division are reversible. Pressure on the cell plate of spermatocytes in telophase has caused rapid fusion of the daughter cells and extensive swelling and loss in rigidity of the protoplasmic gel in which the spindle fibres are imbedded. If the displaced spindle fibres and chromosomes are dissected out, after such a partial reversal, they are found to have undergone no appreciable change in rigidity. From a preliminary study of mitosis, a few conclusions, that are probably general, can be drawn. It seems that cell-division results primarily from concomitant shrinking and swelling or change in water- holding-power of different portions of the cell protoplasm. Many of the structural elements of the mitotic figure separate out of the pro- toplasm and change in rigidity according to their water-content. During the prophase, the nuclear substance becomes so soft that movement of the components of the nucleus is affected by flowing of the nuclear gel. The mechanism at the basis of this flowing seems to be a change in water-holding-power of the nuclear components. I wish here to thank Dr. A. P. Mathews for the very helpful interest that he has shown in this investigation. Reprinted from THE JOURNAL OF BIOLOGICAL CHEMISTRY. VOL. XIV, No. 2, 1913 ON THE NATURE OF THE IODINE-CONTAINING COMPLEX IN THYREOGLOBULIN. BY FRED C. KOCH. (From the Hull Laboratories of Biochemistry and Pharmacology, University of Chicago.} (Received for publication, January 27, 1913.) In this paper are given the results of an attempt to determine the nature of the active complex in the iodine-containing active principle of the thyroid gland. Although the nature of this group was not determined, the quantitative physiological results here reported serve to establish certain predicted and other unexpected facts and to eliminate certain hitherto considered probabilities. The problem was taken up both by analytical and by synthetic methods. In the former method the physiological activity and iodine content of the dried thyroid tissue, the globulin therefrom and various products of hydrolysis from this globulin were deter- mined quantitatively. In the second method two iodized amino- acid derivatives, not previously tested by quantitative methods, were prepared synthetically and their physiological activity stud- ied quantitatively. In thus tracing the active complex a number of important as- sumptions were made. First, that the activity of unaltered thy- roid tissue depends quantitatively on its iodine content. Second, that the best method known for measuring this activity directly and quantitatively is the Reid Hunt acetonitrile test.1 Third, that in case the iodine is present in the products of hydrolysis in the same combination as in the globulin then, per unit of iodine, these will still possess an activity comparable with the original globulin. Fourth, that in case the iodine complex is an iodized amino-acid and that in case this is decomposed in the process of hydrolysis then the synthetic preparation of various iodized amino-acids or derivatives thereof and the quantitative testing of 1 This Journal, i, p. 33, 1905. THE JOURNAL OF BIOLOGICAL CHEMISTRY, VOL. XIV, NO. 2. 102 The Iodine Complex of Thyreoglobulin these per unit of iodine may determine the probable nature of the iodine complex. In other words, the actual quantitative physio- logical activity per unit of iodine as measured by the Reid Hunt method was taken as the crucial test for the presence or absence of the unaltered iodine complex. The historical development of the relation of thyroid activity to iodine content need not be considered at this time, especially in view of the thorough reviews and extensive confirmatory experi- ments made by Reid Hunt and A. Seidell,2 as well as the compara- tive histological and chemical studies by Marine in cooperation with Lenhardt and Williams.3 A careful study of these papers justifies the first assumption. The second assumption is also well taken provided the proper precautions are observed as shown by Reid Hunt and A. Seidell.4 Other methods for testing the physiological activity of thyroid substance, based on changes in blood pressure,5 on increasing the irritability of the depressor nerve,6 on changes in nitrogen metabolism7 and on curative effects in cretinism8 have been employed, but are not applicable in a quantitative study, nor are they as specific reactions. Of the third and fourth assumptions we had no definite proof. The studies of Oswald9 and others show that during hydrolysis of thyreoglobulin only 30 per cent or less of the iodine remains in organic combination. The iodine thus combined is in the vari- ous fractions and qualitatively it has been determined10 that prob- ably the greater activity remains in the more complex products 2 Bulletins 47 (1908) and 69 (1910) of the Hygienic Laboratory, U. S. Pub- lic Health and Marine Hospital Service. 3 Johns Hopkins Hospital Bull, xviii, p. 359, 1907; Journ. Inf. Dis., iv, p. 417, 1907; Archives of Internal Med., i, p. 349, 1908; Ibid., iii, p. 66, 1909; ibid., iv, p. 440, 1909; ibid., vii, p. 506, 1911; ibid., viii, p. 265, 1911; Journ. of Exp. Med., xiii, p. 455, 1911. 4 Loc. cit.; Journ. of Pharmacol. and Exp. Ther., ii, p. 15, 1910. 5 von Fiirth and Schwarz: Pfluger's Archiv, cxxiv, p. 113, 1908. 6 von Cyon and Oswald: Pfliiger's Archiv, Ixxxiii, p. 199, 1901; Asher and Flack: Zeitschr. f. Biol, Iv, p. 83, 1910. 7 Baumann: Zeitschr. f. physiol. Chem., xxi, p. 487, 1896; ibid., xxii, p. 1, 1896; Munch, med. Wochenschr., xl, 1896. 8 E. Pick and F. Pineles: Zeitschr. f. exp. Path. u. Ther., vii, p. 518, 1909- 10. 9 Arch.f. exp. Path. u. Pharm., Ix, p. 115, 1908. 10 Pick and Pineles: loc. cit. Fred C. Koch 103 of hydrolysis where also the greater part of the organically com- bined iodine is found. What relation the activity bears to the iodine content therein has however not been determined. As stated above we have evidence that some of the iodine is split off as iodide, but we have no direct evidence that all the organ- ically combined iodine found in the products of hydrolysis is still in the same complex or in the same structural relationship as in the original thyreoglobulin. A number of iodized amino-acids have been studied qualitatively as to physiological activity. In no case has thyroid activity been detected. The most conclusive results as to the inactivity of 3,5-iodo-laevo-tyrosine are those reported by Strouse and Voegtlin.11 Other observations on the inactivity of various iodized proteins, which on hydrolysis yield 3,5-iodo-tyrosine, also bear out these conclusions. The studies on other iodized amino-acids do not lead to definite conclusions. Thus von Fiirth and Schwarz12 prepared and studied what they con- sidered iodized phenylalanine, histidine and tryptophane. They reported all these substances as physiologically inactive, but gave no data indicating that they had really separated iodo-deriva- tives of these substances. Pauly13 however actually separated pure tetra-iodohistidine anhydride and tri-iodo-imidazol and re- ported that these substances increased the respiratory and pulse frequencies, although uniodized imidazol had no such action. These considerations lead us to conclude that for the present the validity of the third and fourth assumptions is unknown to us and that the true answers thereto are part of the problem in hand. EXPERIMENTAL PART. The mode of attack has already been outlined above. The details as to the methods employed and the preparation of the substances studied are given below. A. Preparations. Dried hog thyroids. Hog thyroids14 were freed mechanically from fat as much as possible and dried on glass plates in a current of air at 30-35 °C. 11 Journ. ofPharm. andExp. Ther., i, p. 123, 1909. 12 Pfliiger's Archiv, cxxxiv, p. 113, 1908. 13 Ber. d. deutsch. chem. Gesellsch., xliii, p. 2243, 1910. 14 The raw material for this research was supplied by the Armour Labora- tory Department. 104 Iodine Complex of Thyreoglobulin The mass was then ground to a coarse powder and fat removed by ether in the cold. The remaining dry mass was then finely powdered. Duplicate determinations on this gave 0.243 and 0.250 per cent iodine. Thyreoglobulin. This was prepared as previously described.15 Dupli- cate determinations on this gave 0.462 and 0.468 per cent iodine. lodothyrin (a) was prepared by the usual Baumann process from the above thyreoglobulin. The extraction with 95 per cent alcohol of the melanoidin precipitate was made in a continuous hot extractor. Duplicate determina- tions gave 5.81 and 5.85 per cent iodine. lodothyrin (6) was obtained in the same way from the melanoidin precipi- tate which separated in the complete hydrolysis of some of the same thyreo- globulin by 30-35 per cent sulphuric acid. This on analysis gave 7.51 per cent iodine. lodothyrin (c) was obtained from 40 grams of the same globulin by hydroly- sis for three days at room temperature and for twenty-four hours at boil- ing temperature with 20-25 per cent phosphoric acid. Phosphoric acid was used as it was thought that possibly the oxidative action of sulphuric acid might have an injurious effect. This amount of globulin yielded 3.30 grams of melanoidin, containing 1.75 per cent iodine. The iodothyrin extracted from this represented 24 per cent of the weight and contained 4.44- 4.46 per cent iodine. Thus only 61 per cent of the iodine in the melanoidin fraction was recovered in the alcohol extract. Metaprotein (A4). The nitrate from the melanoidin fraction above was neutralized with NaOH and the metaprotein separated and dried over sul- phuric acid in a vacuum desiccator. This weighed 1 .62 grams and contained 1.51-1.53 per cent iodine. Primary albumose (A6). The filtrate from above was half saturated with zinc sulphate after slightly acidifying with sulphuric acid. The precipitate obtained was dialyzed until free from sulphate. In this fraction there were recovered 3.1 grams containing 0.22-0.225 per cent iodine. Secondary albumose (A6). Obtained from the filtrate from above by com- plete saturation with zinc sulphate. The precipitate after dialyzing as above yielded 4 grams dry substance containing 0.069 per cent iodine. The table (I) below gives a summary of the distribution of iodine in the different fractions above. TABLE I. WEIGHT KBCOVERED PER CENT OF IODINE THEREIN WEIGHT OF IODINE PER CENT OF TOTAL IODINE IN THE GLOBULIN Melanoidin precipitate Metaprotein 3.30 1 62 1.74 1 52 0.0575 0 0246 30.9 13 2 Primary albumose Secondary albumose. . . Undetermined iodine . . 3.01 4.0 0.22 0.0695 0.0066 0.0027 3.5 1.5 50.9 15 This Journal, ix, p. 121, 1911. Fred C. Koch 105 Phosphotungstic acid precipitate. Another 40 grams of thyreoglobulin were boiled with 25 per cent phosphoric acid for ninety-three hours. The filtrate from the melanoidin precipitate and metaprotein, after removal of the phosphoric acid by Ba(OH)2 and the excess of barium by sulphuric acid, was concentrated under diminished pressure to about 250 cc. This was then freed from proteose and peptone by the Kutscher tannin method.16 The filtrate finally obtained here after removal of the excess of lead was boiled with BaCO3 to remove the ammonia. The dissolved barium was again removed by sulphuric acid. The filtrate after acidifying with H2SO4 to 5 per cent strength was precipitated with phosphotungstic acid in the usual way. The precipitate after thorough washing with 2.5 per cent phos- photungstic acid solution was freed from phosphotungstic acid, barium and sulphate in the usual way. Duplicate determinations on the dry amino- acid mixture gave 0.0107 per cent and 0.0093 per cent iodine. Another phosphotungstic acid precipitate from a hydrolysis by H2SO4 was worked up in the same way. This dry residue contained 0.0068 per cent iodine. The two samples were mixed and designated as P.T.A. Ppt. 1. This mixture contained 0.0073 per cent iodine. Phosphotungstic acid filtrate (1). This was freed from phosphotungstic acid in the usual way. The amino-acid solution was evaporated to dryness. Duplicate determinations on the dry amino-acid mixture gave 0.0024 per cent iodine. Phosphotungstic acid precipitate (2) . This was obtained in the same way as the above from the partial hydrolysis by 10 per cent sulphuric acid of 141. 6 grams of thyreoglobulin containing 0.511 per cent iodine. The puri- fied dry residue by analysis contained 0.0043 per cent iodine. Phosphotungstic acid filtrate (2}. The filtrate from the above was treated in the usual way. The dry purified amino-acid mixture left gave in dupli- cate determinations 0.0045 and 0.0043 per cent iodine. Tetra-iodohistidine anhydride. Histidine was prepared from ox erythro- cytes by the method of Frankel.17 Various methods were employed in trying to iodize the dichloride or the base itself but in no case were there indications of true absorption of iodine, but rather decomposition of the histidine. While this work was under way Pauly18 published his obser- vations with the same conclusions as to the difficulty or inability to iodize histidine directly. At the same time, as stated above, he published his observations on tetra-iodohistidine anhydride. Following the methods given by Pauly19 the preparation of the methyl ester of histidine dichloride was carried out and from this the histidine anhydride by the Pauly modification20 of the Fischer and Zuzuki method. The histidine anhydride was recrystallized from hot water a number of times to obtain the more "Zentralbl. f. PhysioL, xix, p. 504, 1905. 17 Monatsh.f. Chem., xxiv, p. 230, 1903. 18 Ber. d. deutsch. chem. Gesellsch., xliii, p. 2243, 1910. 19 Zeitschr. f. physiol. Chem., Ixiv, p. 75, 1910. 20 Loc. cit. io6 The Iodine Complex of Thyreoglobulin readily soluble laevorotatory form. This was then iodized according to the Pauly method. One determination on the snow-white product gave 63 per cent iodine (theoretical 65 per cent). The slightly lower value may be due to an admixture of a small amount of di-iodohistidine anhydride. Iodized tryptophane. Tryptophane was prepared from commercial casein by the Hopkins-Cole method.21 Several attempts were made to iodize the pure crystals by the method of Neuberg,22 but in no case was a substance obtained containing more than 6.3 per cent iodine. The preparation finally made for physiological testing was obtained by dissolving one milligram molecule of tryptophane in 4 cc. of | NaOH, cooling by immersing in ice water and, while keeping cool and stirring well, adding drop by drop 6 cc. of aqueous N iodine solution. The mixture was allowed to stand at ice box temperature for twenty-four hours, then filtered off. The precipitate was well washed with cold water and dried over sulphuric acid in a vacuum desic- cator. The product obtained is light brown in color, readily soluble in alka- lies, reprecipitated on acidifying and liberates a very small amount of iodine to chloroform on shaking therewith. Duplicate determinations on this gave 41.5 and 41.9 per cent iodine (the theoretical for mono- and di-iodo- tryptophane are 38.4 per cent and 55.7 per cent respectively). B. Methods. Determination of iodine. The Hunter23 method with slight modifications was employed. The material to be analyzed, taken in quantities of 0.05- 2 grams was mixed with 15 grams of fusion mixture and covered with 10 grams of fusion mixture as suggested by Hunter. To conduct the fusion the Roger's ring burner was found to be much more satisfactory in ensuring a uniform rapid heating without overheating. With the size of the flame once determined one "finds ten minutes to be ample time to give a satisfactory, easily removable fusion. In the treatment with alkaline hypochlorite it was considered best to warm to 40°C. for ten minutes. In acidifying it is very important to make sufficiently acid and then always to the same degree. Sulphuric acid of 25 per cent strength was used here and since the same amounts of fusion mixtures and hypochlorite were used in each case the acidity was well controlled by always adding the same amount of acid. In removing the excess of chlorine gentle boiling was continued for forty minutes after the negative test of the vapors by starch iodine paper. In this way the blank test on the reagents never was more than 0.1 cc. of a ^ Na2S2O3.5H2O solution. Physiological testing by the Hunt method. The method employed was that of feeding the same quantity of iodine, in the different combinations, to white mice in such a manner as to make as certain as possible the entire consumption of the material fed. In order to do this each mouse was first 21 Journ. of PhysioL, xxvii, p. 418, 1901. 22 Biochem. Zeitschr., vi, p. 276, 1907. 23 This Journal, vii, p. 321, 1910. Fred C. Koch 107 fed for three or f >ur days with cracker dust made into pellets of known weight. At the close of this preliminary feeding the unconsumed material was weighed and from this the average amount eaten per day determined. For ten days following this period each mouse then received this weight of cracker dust, with the incorporated iodine-containing substance, in the form of pellets. The control mice were fed in the same way with plain cracker dust pellets. At the end of the 10-day feeding period the acetonitrile was injected subcutaneously. Each dose administered in series I, II and III was contained in 1 cc. of fluid; in series IV-IX, in 0.5 cc.; in series X, in 0.66 cc. In most cases the animals consumed the food very well. All the mice used were raised in the laboratory building on a diet of milk and crackers with occasional bits of lettuce until used for the experiment. Care was taken to compare mice of as nearly the same age as possible. In the tables below the litter number of each mouse is given. The ages of the mice of the various litters were as follows : Litter 2, 119 days; litter 3, 102 days; Utter 4, 100 days; litter 5, 80 days; litters 6 and 10, 99 and 113 days respectively; litter 9, 115 days; litters 11-12, 125 and 135 days respectively; litters 13-14, 144 and 151 days respectively; litter 28, 85 days; litters 29-30, 59 and 66 days respectively; litters 31, 32 and 34, 95, 85 and 97 days respectively; litters 33-35, 101 and 89 days respectively; litters 36-37, 89 days; litter 38, 79 days; litters 56-60, 91-103 days. C. Discussion of the physiological tests. Thyreoglobulin. Series I shows that thyreoglobulin possesses the full activity per unit of iodine when compared with the dried thyroid from which it was prepared. This is also confirmed by series IV where a decomposition product obtained from the globulin still shows the complete activity per unit of iodine. The whole of the physiological activity of the gland is therefore quantitatively in the thyreoglobulin. Metaprotein. As stated above, this still shows the full activity per unit of iodine although the percentage concentration of iodine has increased from 0.465 per cent in the thyreoglobulin to 1.52 per cent in the metaprotein. lodothyrin. None of the iodothyrin preparations tested was found to bring about a resistance to acetonitrile more than three- fourths of that produced by the thyroid-tissue fed mice. The indi- cations are that these preparations are all about equally inactive, lodothyrin is therefore less active per unit of iodine than the thyreo- globulin. See series III and V. Primary albumose. This is still very active, as shown by series IV and VII; although the full activity per unit of iodine is not io8 The Iodine Complex of Thyreoglobulin shown to be present in every case tested. In this connection it may be mentioned that the results in series VI are of no value. This series VI, however is an illustration of irregular results, due in all probability to impure acetonitrile. The acetonitrile was taken from a freshly opened bottle and found to smell decidedly of hydrocyanic acid. Before using, it was shaken twice with saturated potassium carbonate solution, dehydrated with P2O6 and twice distilled from fresh P205. Finally it was redistilled and the fraction collected between 79 and 83°C. This distillate was used in series VI. For the later series this distillate was again purified in the same way three times and finally redistilled twice without the addition of P2O6. Here the distillate was collected between 80.5 and 81.5°C. Secondary albumose. This is much less active per unit of iodine than either the iodothyrin preparations or the primary proteoses. Series VII shows this, where the maximum dose resisted is only 40 per cent of the maximum dose resisted by the thyroid-tissue fed mice. Ammo-acids from the phosphotungstic acid precipitate and the phosphotungstic acid filtrate respectively. The results in series IX indicate that the former possess very little physiological activity as measured by the Hunt method. On the whole, however, the results here are very unsatisfactory as the mice did not eat the amino-acid mixtures well, there being two or more days' feeding left. The results indicate that these amino-acid fractions contain very little thyroid activity. This is better shown in series X where only one-tenth the quantity of iodine-containing substances was fed. Although the mice fed with dried thyroid tissue resisted an amount over two and a half times that of the control mice, still the mice fed with the same amount of iodine, but in the form of amino-acids, resisted very little, if any, more of the acetonitrile than the control mice. In other words, these amino-acid fractions show a very slight physiological activity, if indeed they possess any activity whatever. Tetra-iodohistidine anhydride and iodotryptophane. These sub- stances when fed in amounts representing ten times the amount of iodine fed as thyroid tissue do not appreciably increase the resistance to acetonitrile. See series II and VIII. Fred C. Koch 109 Table II gives a summary of the results above. The relative physiological activity is expressed (on the basis of feeding the same amount of iodine in each case) as follows : representing in each case, by 100, the largest dose of acetonitrile from which the thyroid- tissue fed mice recovered, then the other figures represent the proportions the limiting doses of the otherwise fed mice bear thereto. TABLE II. RELATIVE ACTIVITY IODINE IN THE SUBSTANCE TOTAL IODINE IN THB TISSUE Thyroid tissue Thyreoglobulin . . 100 100 •per cent 0.247 0 465 per cent 100.0 100 0 Metaprotein 100 1 520 12 0 lodothyrin 50-75 4 46-7 51 18 3 Primary albumose . . 80-100 0 220 3 *> Secondary albumose 40 0 0695 i z Amino-acids precipitated by phos- photungstic acid Amino-acids not precipitated by phosphotungstic acid 0(+?) 0(+?) 0.0043 0 0044 Tetra-iodohistidine anhydride lodotryptophane 0 o 65.00 41 70 These results show that both the thyroid activity and iodine may be concentrated from thyroid tissue in the thyreoglobulin as well as in the metaprotein and iodothyrin from the latter. Per unit of iodine, however, we have full activity retained in the thyreo- globulin and metaprotein only. In the primary albumose frac- tion we have a lowering in the percentage concentration of iodine and also a slight lowering in the physiological activity per unit of iodine. In the secondary albumose this is still more marked. In the amino-acid fractions the activity is extremely low if present. In view of the researches of Hunt and Seidell with various iodine compounds and in view of the results obtained here, we cannot attribute the protective action in any of these cases to iodine itself, but to a specific iodine-containing complex in the thyreoglobulin. It is significant to note that the highest physiological activity per unit of iodine is found in the original protein and in the more com- plex products of hydrolysis. Since the lowest products of hydroly- sis are still less active per unit of iodine than the secondary albu- f no The Iodine Complex of Thyreoglobulin mose it indicates either that the iodine group is altered in the hydrolysis, or that the iodine-containing group when in simpler combination or when separated, does not possess the full specific thyroid activity. That the iodine-containing group when once separated would not possess the full activity is not at all unlikely, but we would be inclined to expect it to show some activity; at least when given in amounts such as were employed by Strouse and Voegtlin with iodotyrosine and by the author in the experi- ments with tetra-iodohistidine anhydride and iodotryptophane. The indications as to the presence of tyrosine and tryptophane in iodothyrin are very favorable, both from the chemical studies on iodothyrin and also from similar studies on iodine-free melanoi- dins.24 It is not likely that the iodine is split off and then later added to the melanoidin fraction; it is more likely that it is already present in the globulin in the melanoidin-forming groups and re- mains in the original position in these groups, but that the groups themselves are changed in regard to each other and thus the activ- ity affected to some extent; a poly-iodo derivative may be changed to a mono-iodo derivative and then may show decided differences in physiological activities. If this were not the case we would expect artificially iodized melanoidins to show a decided thyroid activity. Furthermore, it is not likely that sufficient hydriodic acid is split off in the early stages of the hydrolysis to yield as much iodine as is contained in the melanoidin fraction. Finally, it is not at all improbable that we here have to do with a specific iodophore group just as in hemoglobin we have the chromophore group containing the iron. The negative results with artificially iodized proteins speak strongly in favor of this view. CONCLUSIONS. 1. The full activity of thyroid tissue is contained in the thyreo- globulin fraction when this activity is measured by the Hunt method. 2. The full activity per iodine unit is still present in the meta- protein fraction from this globulin, although the iodine content in the metaprotein fraction has been increased over threefold that of the globulin itself. 24 Samuely: Hofmeister's Beitr.age, ii, p. 355, 1902. Fred C. Koch in 3. The other products of the hydrolysis studied, primary albu- mose, iodothyrin and secondary albumose, show a gradual decrease in activity per unit of iodine in the order given. 4. The amino-acid fractions precipitated and not precipitated by phosphotungstic acid from the partially hydrolyzed thyreo- globulin still contain very small amounts of iodine and per unit of iodine are either extremely low in activity or entirely inactive. 5. Tetra-iodohistidine anhydride and iodotryptophane do not possess thyroid activity as determined by the Hunt method. I wish to express my thanks to Prof. A. P. Mathews for sug- gestions made in the course of the work. H2 The Iodine Complex of Thyreoglobulin SERIES I. February 25-March 8. MOUSE (")cf (6) 9 LITTER NO. 4 4 4 4 FED DAILY WITH CRACKER DUST PLUS FATAL DOSE OK ACETO- NITRILE DEATH OCCURRED AFTER DOSE OF ACETO- MTHILE FROM WHICH RECOVERY OCCURRED mg. per gm. 0.4 Gravid, Escaped krg. n not injec from cag mg. per gm. 0.35 ted g 3 79 3.50 (c) 9 (d) 9 0) 9 4 3 3 !1 mg. dried hog thy- roid (=0.00247 mg. I) 4.49 died whi 4.0 3* le feedin 30 (/) cf (a) cf .. (h) 9 . 3 3 3 j 0.531 mg. thyreo- f globulin J ( =0.00247 mg. I) 4.0 U (i) 9 (7) 9 SERIES II. April Si-May 1. (a) cf 3 0 55 41 5 4 (c)j (Ser.I)... 3 0 (rf)i (Ser.I).. 3 0 (e) cf 2 )1 mg. dried hog thy- 5.0 31 (/) cf 2 roid (=0.00247 4.0 4 (o) cf 2 mg I) 4 ^ 00 (/O 9 4 0 00392 mg tetra- (i) 9 4 iodohistidine an- 0.60 (fc) cf 3 10.0392 mg. tetra-l 0.55 5 (09 4 iodohistidine an- 3.55 48 (m) b (Ser.I). 4 hydride 0 ( =0.0247 mg. I) SERIES III. May 19-29. (a)rr "H 3.0 3 i died whi le feedin 2.5 g (c) roid ( = 0.00247 j mg. I) died whi le feedin • 2.0 2.5 3.0 Wcf :. (e) cf 31-34 31-34 31-34 31-34 11.123 mg. primary albumose (As) (=0.00247 mg.I) 2.8 3.0 > 6 > 8 2.5 2.0 (/) 9 (g} 9 . . (h) 9 .... (»)cf 0') 4 24 1.2 Fred C. Koch SERIES VIII. February 17-27. MOUSE LITTER NO. FED DAILY WITH CRACKER DUST PLUS FATAL DOSE OF ACETO- NITRILE DEATH OCCURRED AFTER DOSE OF ACETO- NITRILE FROM WHICH RECOVERY OCCURRED (a) 9 33-35 mg. per gm. hrs. mg. per gm. 0.45 (6) 9 (c} 9 (d) 9 . 33-35 33-35 33-35 0 55 <18 0.40 0.35 (e) & 33-35 4.0 2 (/)c7 (a) 9 . . 36-37 36-37 1 mg. dried hog thy- roids (=0.00247 died whi 3.0 le feedin 6 g (/O 9 36-37 mg. I) 2.0 (o cf ; . . O')cf (k) 9 . . 33-35 33-35 36-37 10.0059 mg. iodotryp- tophane 0.55 1.6 1 0 >36 2* 18 (09 36-37 ( = 0.00247 mg. I) 0.45 >24 (TO) cf 33-35 ) 0.059 mg. iodotryp- 10 < 3 (n)cf (o) 9 33-35 33-35 tophane ( =0.0247 mg. I) 0 70 <18 0.5 SERIES IX. March 12-22. (a) a Ser. VIII (6) 6 Ser. VIII (c) c Ser. VIII (d)n Ser. VIII 33-35 33-35 33-35 33-35 11 mg. dried hog thy- roid ( = 0.00247 mg.I) 4.0 did not < 6 eat; not 3.5 injected 3.0 («) 9 (/) 9 38 38 38 ] 33.6 mg. P. T. A. [ Ppt. 1 (=0.00247 j mg. I) died whi died whi le feedin le feedin g g 0.8* (fc)cf (l) cf 38 38 38 ] lOOmg. P.T.A. Filt. [ 1 (=0.0024 mg. J D 1.0 0.8 0.6 < 3* <18* <18* • * Two or more days feeding left. This experiment is not reliable as animals were used which had recovered in previous experiments and the differences in age were too great for such young animals. n6 The Iodine Complex of Thyreoglobulin SERIES X. MOUSE LITTER NO. FED DAILY WITH CRACKER DUST PLUS FATAL, DOSE OP ACETO- NITRILE DEATH OCCURRED AFTER DOSE OF ACETO- NITRILB FROM WHICH RECOVERY OCCURRED (a)rf (6) d* 56-60 ' 56-60 56-60 56-60 mg. per gm. 0.5 died whi hrs. <10 e feedin mg. per gm. g 0.4 0.35 (c) 9 (d} thyroid (= 0.000247 mg. I) 1.2 not injec 3* ted, grav 1.1 1.0 id (/) cf . (0) 9 • • (A) 9 (i) d1 . . . . 56-60 56-60 56-60 56-60 1 5.74 mg. P. T. A. > Ppt. 2 (=0.000247 j mg. I) 0.5 1.0 0.8 48* < 6f <16 0.4 C/)c? (fc)rf1 (/)9 (w)cf (n) d\ 56-60 56-60 56-60 56-60 , 15.61 mg. P. T. A. Filt. 2 (=0.009247 mg. I) 1.0 0.8 0.7 3| <12 24 0.5 (o) d" (p) d*.... * Two days' feeding left. t About one day's feeding left. Reprinted from THE JOURNAL OF BIOLOGICAL CHEMISTRY, VOL. XIV, No. 3, 1913. CONTRIBUTIONS TO THE CHEMICAL DIFFERENTIA- TION OF THE CENTRAL NERVOUS SYSTEM. I. A COMPARISON OF THE BRAIN OF THE ALBINO RAT AT BIRTH WITH THAT OF THE FETAL PIG. BY MATHILDE L. KOCH. (From the Hull Laboratory of Biochemistry and Pharmacology, University of Chicago, and the Wistar Institute of Anatomy, Philadelphia.) (Received for publication, February 20, 1913.) INTRODUCTION. For the study of the progressive changes in the central nervous system during growth and senescence, the albino rat, on account of its small size, short span of life and its powers of rapid reproduc- tion1 is especially suited. Its growth processes are, moreover, strikingly like those of man, as has been brought out by the exten- sive investigations of Dr. Donaldson within the past few years. It was, therefore, decided to use this animal for a study of the chemical differentiation of the central nervous system during growth. The youngest brains which could be conveniently collected for chemical analysis were those of rats just born. As it was not certain that the rat at this period of development was sufficiently immature (chemically undifferentiated) to serve as the starting point for such a growth series, it was suggested by my brother, Dr. Waldemar Koch, that the brain of the new born rat be com- pared chemically with the brain of the fetal pig, collected at vari- ous stages of fetal life. By such a comparison we hoped to determine the physiological age of the rat at birth in terms of fetal pig material; and to obtain, possibly, from the pig fetus, material which would be more imma- ture than the new born rat. 1 H. H. Donaldson: President's Address, Journ. of Nervous and Mental Disease, xxxviii, p. 258, 1911. 267 268 Chemical Differentiation of the Brain MATERIAL AND METHODS. The rat material was obtained from the Wistar Institute of Anatomy, which supplied the brains of rats of known age, from animals which had been raised under constant conditions; two factors which are absolutely essential for such a study. The material was collected by Dr. Hatai and the method used was that adopted by the Institute. This in brief is as follows : the rat was chloroformed; the skull opened from the dorsal side; the division between the brain and the cord made at the tip of calamus scrip- torius; and the brain removed. The meninges of the brain were left intact. Such blood as it contained was, therefore, included in the weight. Immediately after removal the brain was placed in a closed weighing bottle, quickly weighed to within 10 mgms.2 and transferred to a wide mouthed bottle of 300 cc. capacity con- taining absolute alcohol. The weighing bottle was weighed back and the difference recorded as the weight of the sample. As the weight of one brain from rats at this early age is 0.2 — 0.3 gram and as it takes at least from 25 to 50 grams to make one sample for analysis, a large number of brains had to be collected (100 brains of the rat at birth for one 25-gram sample). As this cov- ered a period of several weeks it was necessary to heat the sample from time to time in a water bath kept at a temperature of 70°C. to insure a thorough penetration of the alcohol and sterilization. The amount of alcohol was so adjusted as to make the final concen- tration not less than 80-85 per cent. A well fitting cork stopper covered with tin-foil was now inserted and the bottle carefully shaken to insure a uniform mixture. The dates of collection of the samples were recorded, as the time a sample has been kept in some cases influences the analytical results.3 The tightly corked bottles were then shipped to the Laboratory of Biochemistry of the Uni- versity of Chicago, where the samples were analyzed according to Koch's methods of tissue analysis.4 2 A coarse weighing of the brain was permissible in this instance as it was not the exact brain weight that was sought but merely data for indicating roughly when the required amount of material had been obtained. 3 W. Koch: Methods for the Quantitative Chemical Analysis of Animal Tissue, Journ. of the Amer. Chem. Soc., xxxi, p. 1340, 1909. 4 Ibid., pp. 1329-64. Mathilde L. Koch 269 The fetal pig material was collected by my brother at the Chicago Stock Yards. The fetuses selected were 50, 100, and 200 mm. in length. The pregnant uterus was opened : and the fetuses removed. The neck-rump length of each litter was taken and if the average length corresponded to one of the three sizes mentioned above, the entire litter was taken, placed upon ice and in this chilled condition taken to the laboratory, where the brains were immedi- ately removed, preserved and later analyzed according to the same methods used for the rat material. RESULTS OF ANALYSES. The results from the chemical analysis of the brains from the new born rat and the adult rat are given in table I: those of the 50, 100, and 200 mm. pig fetuses are recorded in table II, and table III gives the summary of all the averages which have been taken from the figures which were most consistent. The brain of the 200 mm. pig fetus was plainly more differentiated, it is there- fore left out of table III and of the final discussion of results. DISCUSSION OF CHEMICAL RESULTS. Before taking up a comparison of the new born rat with the pig fetus it may be well to state, briefly, the chief chemical changes in nervous tissue during growth. It is well known that the chemi- cal composition of a tissue varies with age and that the water content is one of the most important variables. Donaldson states that, "the progressive diminution of the percentage of water in the brain is a function of age and is not significantly modified by any conditions to which the animals have been thus far experi- mentally subjected."5 He suggested that this "is to be regarded as an index of fundamental chemical processes, which take place in the more stable constituents of the nerve cells."6 The principal chemical differences due to growth, noted by my brother, are, "a decrease in moisture,- proteins, extractives, and ash as the brain increases with age, and an increase in cerebrosides, sulphatides, 5 H. H. Donaldson : On the percentage of Water in the Brain and in the Spinal Cord of the Albino Rat, Journ. of N enrol, and Psychol., xx, p. 143, 1910. 6 Ibid. 270 Chemical Differentiation of the Brain phosphatides, and cholesterol; in other words, an increase in sub- stances which predominate in the fibres (medullated sheath) during growth."7 These same differences are to be found in all TABLE I. Relative proportion of the proximate constituents of the brain of the albino rat at birth and when adult. ALBINO RAT (AT BIRTH) ALBINO RAT (ADULT) Moist weight of or Solids in per cent. le brain 0.25 10.42 0.25 10.42 1.667 21.9 Dry weight of one Number of brains brain : . . . in sample 0.026 . • 100 0.026 100 0.380 31 In relative proportions of solids. Proteins 57.16 14.8 0.0* 1.5 16.5 (10.04) 0.96 1.82 57.30 15.6 0.0 1.4 19.3 (6.4) 1.04 1.92 48.5 22.0 9.0 4.6 9.8 (6.1) 0.58 1.39 Phosphatides • • • Cerebrosides Sulphatides Organic extracts 1 Inorganic const . J Cholesterol | ,. Undetermined J Total S Total P Distribution of sulphur in per cent of total S. Protein S 31.02 3.2 49.14 16.6 30.02 2.8 47.26 19.95 64.2 15.6 14.2 6.0 Lipoid S NeutralS.. . Inorganic S Distribution of phosphorus in per cent of total P. Protein P 13.3 33.2 53.5 33.0 53.6 6.8 67.6 25.6 Lipoid P Water-soluble P •Mendel, L: Amer. Journ. of Physiol., xxi, p. 104, 1908. t By difference. 7 W. Koch and S. A. Mann: A Comparison of the Chemical Composition of Three Human Brains at Different Ages, Journ. of Physiol, xxxvi, pp. 1-3. (From the Proceedings of the Physiological Society, November 23, 1907). Mathilde L. Koch 271 TABLE II. Relative proportions of the proximate constituents of the brain of the fetal pig at different ages. 50 MM. PIG FETUS 100 MM. PIG FETUS 200 MM. PIG FETUS Year of analysis ' '11 '12 '12 '11 '12 '12 '11 10.1 Moist weight of one brain 0.40 8.75 0.43 9.87 0.47 9.04 1.8 9.1 1.91 8.98 2.15 8.98 Solids in per cent Dry weight of one brain 0.035 0.042 0.042 0.164 0.171 0.193 Number of brains in sample 65 111 109 35 27 27 In relative proportions of solids. Proteins 56.6 13.0 2.4? 22.20 2.4* (3.4) 0.67 1.74 58.2 15.04 0.8 20.5 2.4* (3.06) 0.59 1.85 54.61 15.79 1.05 23.84 2.4* (2.31) 0.58 1.90 51.5 15.7 1.8? 24.2 4.4* (2.4) 0.59 1.76 51.81 16.31 0.96 24.92 4.4* (1.6) 0.57 1.91 52.34 14.85 0.84 25.44 4.4* (2.13) 0.55 1.82 43.8 17.2 0.00? 23.2 (8.42) 0.55 1.45 Phosphatides Cerebrosides Sulphatides Organic extract 1 Inorganic const. / Cholesterol Undetermined f TotalS Total P Distribution of sulphur in per cent of total S. ProteinS Lipoid S Neutral S 54.3 7.2? 29.3 9.0 57.3 2.67 29.4 10.67 55.8 3.59 27.6 13.0? 58.4 6.1? 26.7 9.0 57.8 3.36 28.97 9.93 55.66 2.98 31.68 9.6 60.9 0.00? 25.5 13.33 Inorganic S Distribution of phosphorus in per cent of total P. Protein P Lipoid P 1 31.6 53.6 15.7 32.4 51.8 14.0 29.6 56.2 35.5 14.5 34.5 50.9 14.6 31.6 53.8 5.2 46.3 48.5 Water-soluble . . . * Mendel, L.: Amer. Journ. of Physiol., xxi, p. 103, 1908. t By difference. (?) Indicates doubtful result. THE JOURNAL OF BIOLOGICAL CHEMISTRY, VOL. XIV, NO. 3. 272 Chemical Differentiation of the Brain TABLE III. Relative proportions of the proximate constituents of the brain of the fetal pig at different ages compared with the brain of the albino rat at birth. (Aver- ages of the foregoing determinations.} PIG FETUS ALBINO RAT 50 mm. 100 mm. at birth adult Moist weight of one brain 0.433 9.22 0.039 95 1.90 8.99 0.171 30 0.25 10.42 0.026 100 1.667 21.9 0.380 31 Solids in per cent Dry weight of one brain Number of brains in sample In relative proportions of solids. Proteins . . 56 47 51 88 57 23 48 5 Phosphatides 15.41 15.62 15.2 22.0 Cerebrosides* .... 0 0 0.0 0 0 9.0 Sulphatides 0 92 0 90 1 45 4 6 *•« Organic extract 1 22 18 24 69 17 9 9 8 Inorganic const. / Cholesterol 2.4f 4.4t Undetermined! : Totals (2.59) 0 585 20.49 0 57 (8.22) 1 00 (6.1) 0 58 Total P. . . 1 83 1 83 1 87 1 39 Distribution of sulphur in per cent of total S. Protein S 55 8 57 28 30 52 64 2 Lipoid S . . 3 13 3 17 3 0 15 6 Neutral S 28 7 29 11 48 2 14 2 Inorganic S. . 9 83 9 51 18 27 6 0 Distribution of phosphorus in per cent of total. Protein P 14.8 14.55 13.3 I 6.8 Lipoid P. 31.2 33.8 33.1 67.6 Water-soluble P 53 8 52 3 53 55 25 6 1 ^°'D Cerebrosides not determined in fetal brains. Not present according to Mendel, t Mendel, L: Amer. Journ. of Physiol., xxi, p. 103, 1908. J By difference. Mathilda L. Koch 273 nervous tissue during growth and may therefore be used in making a comparison between the brain of the new born rat and that of the fetal pig to determine which is the more immature. We may now proceed to consider in detail the comparison of the various constituents8 in the brain of the new born rat and the pig fetus. Water. The per cent of water in the brain of the new born rat is closely similar to, but a little lower than, that of either the 50 mm. or 100 mm. pig fetus. This would indicate that the rat is of about the same physiological age as these fetuses, since the differences are within the limits of error. Protein. The per cent of protein in the total solids is higher in the brain of the new born rat than in either that of the 50 or the 100 mm. pig fetuses. Since the per cent of protein is highest in the youngest material, this is an indication that the rat's brain is less mature than that of the 100 mm. pig fetus, but not very different from the 50 mm. fetus. Phosphatides. The per cent of phosphatides is the same in the new born rat as it is in the 50 and the 100 mm. pig fetus. This would indicate a close agreement in physiological age between these two. This is the lowest phosphatide content yet obtained in an analysis of the brain tissue and approaches that observed in the suprarenal, which, among all the organs, comes closest to that of nervous tissue in chemical composition. Cerebrosides. These are absent in both the new born rat, and in the pig fetus, as is to be expected in nervous tissue before medul- lation. Sulphatides. The percentage of sulphatides is about the same in the new born rat as in the pig fetus, which indicates the same age. Organic extractives and inorganic constituents. These are some- what higher in the pig fetus than in the new born rat and, except as this is associated with the greater per cent of lymph in the embry- onic material, it would indicate it to be more immature than the new born rat. 8 The nature and significance of the constituents will be discussed in the third paper of this series. 274 Chemical Differentiation of the Brain Cholesterol. The figures for cholesterol were not determined by me, but were taken from Mendel9 and incorporated here for the sake of completeness. This leaves undetermined from 2 to 3 per cent which is not more than would be expected in the errors involved in making so many determinations from one tissue and calculating approximate con- stituents from assumed factors. The distribution of sulphur in per cent of total sulphur is widely different in the two forms, but as this is not correlated with age but is apparently a species peculiarity, the results are not out of harmony with the foregoing conclusions. The distribution of phosphorus between the protein, lipoidand water-soluble phosphorus is closely similar in the rat and the 50 and 100 mm. pig fetus, showing the physiological ages to corres- pond. The remarkably high figure for neutral and inorganic sulphur in the rat at birth requires an explanation but it is not possible to give this with the data so far at hand. The general conclusions from these figures are, that from a chemical standpoint the brain of the new born rat is about as imma- ture as that of the 100 mm. pig fetus, being on the whole a little less differentiated than the latter. The differences between the brain of the 50 mm. and the 100 mm. pig fetus are not marked, and this would indicate that there oc- curs between these ages an increase in weight unaccompanied by any significant change in chemical composition. This would corres- pond with the results of Mendel10 and Raske11 who found that in the brains from these young fetuses there is no chemical dis- tinction between grey and white matter. Moreover the brain of the 50 mm. pig fetus is the youngest which it is practicable to obtain for analysis and even at this age the tissues are so watery and filled with lymph that some error is thereby introduced in the analysis of the constituent tissues. Since the brain of the 50 mm. pig fetus shows no material differ- ences from that of the 100 mm. pig fetus and the latter is no more 9 L. B. Mendel and Charles S. Leavenworth : Chemical Studies on Growth. IX. Notes on the Composition of Embryonic Muscular and Nervous Tissues. Amer. Journ. of PhysioL, xxi, p. 103, 1908. 10 Ibid. 11 Raske: Zeitschr.f. physiol. Chem., x, p. 340, 1886. Mathilde L. Koch 275 immature chemically than that of the new born rat, it appears that the new born rat's brain is as young nervous material as can conveniently be analyzed at present: and it forms, therefore, a convenient starting point for the study of the chemical differen- tiation of the central nervous system during growth. CHEMICAL RESULTS COMFIRMED BY PHYSIOLOGICAL AND ANATOMICAL DATA. It is astonishing that chemically the brain of the new born rat should be as immature as that of the 100 mm. pig fetus, but, surprising as this fact is, it is substantiated by a comparison of the structure of the cerebellum of these two animals and of their behavior at the time of birth. It is a well-known fact that the rat is born in a very immature state, with its eyes shut, and when first born, is capable only of movements involved in sucking, bending the body and tail and making a squeaking noise.12 The pig, on the other hand, "is born with its eyes open and requires no assistance as a rule in making its start in life. It is more or less able to walk around as soon as born."13 Such a state of activity in the rat is not reached until the period of weaning 17-21 days after birth. This difference in physiological behavior is correlated with the relative development of the cerebellum of the two animals; par- ticularly as indicated by the development and transformation of the outer granular layer of cells. A comparison of this layer in both animals, founded on the observations of Addison14 who studied the different layers of the cerebellum in the albino rat, and of Takasu15 who studied these same layers in the pig fetus, brought out the following facts : 12 Wm. H. F. Addison: The Development of the Purkinje Cells and the Cortical Layers in the Cerebellum of the Albino Rat, Journ. of Comp.N enrol., xxi, p. 476, 1911. 13 Forbes: personal communication. 14 Wm. H. F. Addison: Journ. ofComp. Neural., xxi, p. 464, 1911. 15 K. Takasu: Zur Entwicklung der Ganglienzellen der Kleinhirnrinde des Schweines, Anat. Anz., xxvi, pp. 225-32, 1905. 276 Chemical Differentiation of the Brain BAT PIG First appearance of cells in outer granular layer Division of layer into two zones inner and outer 19-day fetus At birth 50 mm. fetus 100 mm. fetus Disappearance of cells from inner zone of layer 21 days after birth 300mm. fetus The first appearance of the cells in the outer granular layer of the cerebellum in the rat is in the 19-day fetus, and in the .pig in the 50 mm. pig fetus. At this time the cells are settled in a thin layer (two rows deep) around the outer edge of the cerebellar cortex. This layer increases until a considerable depth is filled in with cells which soon separate into two strata, an outer and an inner; this separation takes place in the rat at birth, and in the pig fetus when it is 100 mm. in length. The cells from the outer granular layer now begin to migrate to the inner granular layer and the dis- appearance of the cells from this outer granular layer, which corres- ponds with the time of securing motor control in an animal, occurs in the rat at the twenty-first day of life, and in the pig when this is from 200 to 300 mm. in length, or at birth. These facts show, therefore, that the new born rat is as developed with respect to motor activity as the 100 mm. pig fetus, and that the rat at wean- ing (17-21 days after birth) and the pig at birth are at correspond- ing physiological ages. The conclusion from this anatomical com- parison, namely, that the 100 mm. pig fetus and the rat at birth are of like physiological age, fully confirms, therefore, the conclu- sion drawn from both chemical and physiological evidence already adduced. We now ask the question, how far this result, that the nervous system of the new-born rat is chemically as old as that of the 100 mm. pig fetus, agrees with observations made by Donaldson, that the rate of growth and percentage of water in the mammalian nervous system (represented by the brains of man and the rat) agree in the two forms at equivalent ages, thus indicating that the nervous systems are in corresponding physiological states at equal fractions of the life cycles.16 18 H. H. Donaldson: A Comparison of the White Rat with Man in respect to the Growth of the Entire Body, Boas Anniversary Volume, 1906, pp. 5-26. Mathilde L. Koch 277 It remains therefore to inquire whether the chemical and behav- ior relations between the rat and pig which have just been pointed out, occur at equivalent ages in these two forms. Great difficulty was experienced in finding any statements con- cerning the age of the pig fetuses. The statements of different authors did not always agree, but the two which agreed closest were those of Bradley17 and Coe.18 Bradley compared the length of the embryos with the time from coition; Coe estimated the age from the rate of development of embryos of other mammals. While considerable uncertainty thus attaches itself to these figures19 it may be assumed that the 50 mm. pig fetus is about 40 days old from conception: the 100 mm. fetus is 55-62 days; and the 200 mm. fetus is from 88-90 days from conception. We find in the rat the period of gestation is 21 days and its span of life three years (Donaldson), or a total age of 1116 days; in the pig the period of gestation is 125 days and its normal span of life, as far as could be ascertained, is 20 years,20 or 7425 days. The rat, therefore, lives about one-sixth as long as the pig. Assuming that the rat at birth has lived TTTG", or ^V, of its total life, the 60-day pig fetus will have lived yfio, or TTS, of its life. It appears then, if the total length of life given is correct for both animals and the numbers used for the divisors in each case are really com- parable as they stand, that we do not have corresponding physio- logical conditions of the brain at equivalent ages, for these brains 17 O. C. Bradley: On the Development of the Hind Brain of the Pig, Journ. of Anat. and PhysioL, xl, Part I, p. 1. 18 Mendel refers to Professor Coe in Chemical Studies on Growth. I. The Inverting Enzymes of the Alimentary Tract, especially in the Embryo, Amer. Journ. of PhysioL, xx, p. 90, 1907-1908. 19 Bradley makes the statement, that "although the age of the different embryos is given, it is not intended that it should signify more than the time which elapsed between the time of coition and the time when the mother was destroyed In embryos taken from two litters it not infre- quently happens that those which should be further advanced in develop- ment, judging from the period which has elapsed since sexual congress took place, are as backward, or even more backward, than those of the 'younger* litter." A more definite way to determine the age of an embryo would be, according to Mall, by ossification. No data were available however for a comparison between the rat and the pig. 20 Longevity, Encyclopaedia Britannica (Eleventh Edition), xvi, p. 975. 278 Chemical Differentiation of the Brain are found to be in corresponding states at the -&V and the yiur part of the total life cycles. Had the relation, in the form stated above, held, these fractions should have been identical. It is only fair to add, however, that in view of the absence of precise information concerning the pig and in view of the fact that the early days of gestation are used for cell division accompanied by only slight differ- entiation, too much stress should not be laid on the relation here given. On the other hand, instead of taking the end of life as the fixed point of 'our calculations, we may consider the time when motor control is obtained to indicate closely corresponding states of the central nervous system. In the rat, motor control is obtained at 42 days from conception, and in the pig at 125 days, that is, at birth. If the law of corresponding states is correct, the nervous system of these two animals should be in corresponding conditions at the same fractions, either J, J, or J of these periods. This is found to be the case, for the rat is born after 21 days' gestation. This would be just half way between the two fixed points of concep- tion and time of gaining motor control and this corresponds in the pig to just one-half of its gestation period or about 62 days, which is the age of the 100 mm. pig. It was actually found, both chemically and anatomically, that the nervous systems of these two animals were in the same state of development at these respective periods and it appears from these observations that Donaldson's law may hold, if put in the form: the nervous systems of mammals are in the same physiological state at equal fractions of their total periods of development. In conclusion it gives me great pleasure to thank Dr. H. H. Don- aldson and Prof. A. P. Mathews for their many suggestions in connection with this problem and for their aid in getting this paper ready for publication. The problem itself was suggested to me by my brother and forms the first of a series of papers which are to follow from time to time, on the chemical differentiation of the central nervous system, and on which he was engaged at the time of his death. The work was carried out in the Laboratory of Bio- chemistry and Pharmacology of the University of Chicago and was aided by h grant from the Wistar Institute of Anatomy and Biology, Philadelphia. Mathilde L. Koch 279 SUMMARY. 1. A quantitative determination of the constituents of the brain of the albino rat at birth shows it to be chemically as undifferenti- ated as the brain of a 50 mm. or 100 mm. pig fetus. 2. There is little difference in chemical composition between the 50 mm. and the 100 mm. pig fetus brain. 3. Since the 50 mm. fetus brain is the youngest which can be analyzed and this closely resembles the 100 mm. fetus, and this in turn is no more immature than the new born rat, it appears that the brain of the new born rat is sufficiently immature to serve as a starting point in a study of the chemical differentiation -of the brain during growth. 4. That the brain of the new born rat is as immature as the 100 mm. pig embryo is shown, also, by the similarity of the changes in the outer layer of cells of the cerebellar cortex in both animals previous to gaining motor control, and by the animal's behavior at this period. 5. If the nervous systems are assumed to be in corresponding states when motor control is obtained, and Donaldson's law is correct, that the nervous system is in the same state at correspond- ing physiological ages, then the brain of the rat at birth should correspond chemically with the 100 mm. pig fetus brain. This is found to be the case. Reprinted from THE JOURNAL OF BIOLOGICAL CHEMISTRY, VOL. XIV, No. 3, 1913 CONTRIBUTIONS TO THE CHEMICAL DIFFERENTIA- TION OF THE CENTRAL NERVOUS SYSTEM.. II. A COMPARISON OF TWO METHODS OF PRESERVING NERVE TISSUE FOR SUBSEQUENT CHEMICAL EXAMINATION. BY W. KOCH AND M. L. KOCH. (From the Hull Laboratories of Biochemistry and Pharmacology, University of Chicago, and the Wistar Institute of Anatomy, Philadelphia.) (Received for publication, February 20, 1913.) With valuable biological material it is sometimes desirable to make water estimations and the estimations of the other constitu- ents on the same sample. This can be done somewhat indirectly if the method already described1 of placing the fresh, weighed tissues immediately in 95 per cent alcohol is used. To see whether tissues in which the water had been determined by drying could thereafter be analyzed by the methods referred to above and would yield the same proportion of the various con- stituents as these same tissues treated by the alcohol method, an analysis was made of the brains and spinal cords of albino rats which had been dehydrated in these two ways. The possible drawbacks of the heat method of determining moisture, namely, the oxidation or decomposition of part of the material and the evapora- tion of volatile constituents, are obvious, but we had no definite knowledge of how serious the errors involved in the method might be in practice. To determine to what extent these changes took place and what they were, we analyzed material which had been dried at 95°C. for one week and which at the end of this time had been placed in alcohol, and compared it with similar material which had been placed directly in alcohol. The results are given in the table. It may be seen by a comparison of the results of the two analyses in the table, that decompositions seriously affecting the analyses are produced by heat drying, particularly in the case of the brain. The differences are most marked in the phosphorus compounds. 1 Koch, W. : Journ. Amer. Chem. Soc., xxxi, pp. 1353-4, 1909. 281 282 Preservation of Nerve Tissue C tmpirison of brains and cord* dried at 95°C. with brain* arid cur da plnccd directly in alcohol without ENCEPHALON CORDS Direct Into alcohol Dried at 95°C. Direct Into alcohol Dried at 95°C. Laboratory number W 13 W 18 W 9 W 20 In per cent of total solids. Proteins 48.5 47.9 32.8 29.6 Phosphatides 22.0 16.2 25 3 22 1 Cerebrosides 8.4 (?) 12 5 14 4 Sulphatides 4 5 4.6 70 6.7 Organic ext. } 9 8 12.4 7.6 8 0 Inorganic const./ Undetermined lipoids 6 8* 14.8* 19 2* Total S 0.58 0.59 0.45 0 42 Total P 1 39 1 31 1.44 1.42 Distribution of sulphur in per cent of total S. ProteinS 63.8 64.6 53.7 53.3 Lipoid S 15.6 15.6 30.9 32.1 Neutral S 14 5 14 0 10.3 9.5 Inorganic S 6.1 6.0 5.0 5.0 Distribution of phosphorus in per cent of total P. Protein P 6 8 8 1 5 6 5 0 Lipoid P 67 6 55 1 77 4 69 2 Water Sol. P 25 6 36 8 17.0 25.8 ' Obtained by difference. By drying there has been a destruction of the phosphatides involv- ing a change in the distribution of phosphorus in per cent of total phosphorus; that is, a considerable amount of lipoid phosphorus is changed to water-soluble phosphorus. There was no change in the sulphur distribution. It will be noticed that the phosphatides of the cord are more resistant to heat than those of the brain, a point of sufficient interest to justify repetition. We conclude, then, that the determination of water by drying at 95°C. cannot safely be used, if it is desired to determine in the same sample the relative proportions of the solid constituents; and that the indirect method already described is far superior for this pur- pose. Reprinted from THE JOURNAL op BIOLOGICAL CHEMISTRY, Vol. XV, No. 3, 1913 CONTRIBUTIONS TO THE CHEMICAL DIFFERENTIATION OF THE CENTRAL NERVOUS SYSTEM. III. THE CHEMICAL DIFFERENTIATION OF THE BRAIN OF THE ALBINO RAT DURING GROWTH. BY W. KOCH AND M. L. KOCH. (From the Hull Laboratory of Biochemistry and Pharmacology, University of Chicago, and the Wistar Institute of Anatomy, Philadelphia.) (Received for publication, July 1, 1913.) The transformations which occur in the brain during growth offer a particularly enticing field for the study of chemical differ- entiation, not alone because of the very great interest attaching to the solution of the problem of the chemical basis of its func- tions, but because its structural differentiation during growth is very marked. On the one hand there is the formation of a large amount of new material composing the medullary sheath of the nerve fibers, and, on the other hand, the appearance of a quantity of a peculiar supporting tissue, the neuroglia. The chemical changes during growth should, therefore, be very marked; and it is of interest to discover how far our chemical methods enable us to follow such obvious structural modifications. 1 Waldemar Koch died at Chicago February 1, 1912. As Associate in Bio- logical Chemistry at the Wistar Institute of Anatomy and Biology, he spent the autumn of 1910 and of 1911 in Philadelphia working mainly on matters connected with this research. This paper has been prepared in considerable part from results of analyses made by me under the direction of my brother, and from a manuscript written by him. Many additional analyses, which he had planned, have been made and incorporated into the series. The inter- pretations of the results have been left to a large extent, in his words. I have been assisted in its preparation for publication by Professor A. P. Mathews and Dr. H. H. Donaldson, the aid of both of whom I gratefully acknowledge.— M. L. KOCH. 423 424 Chemical Differentiation of the Brain The selection of chemical methods for such a study was largely guided by the principle now coming to be generally accepted, namely, that in living matter we are not dealing with an aggre- gation of more or less similar, highly organized and necessarily complex molecules (Riesenmolekul of Pfltiger), but rather, with a more or less heterogeneous substratum in which dissimilar and not necessarily highly complex molecules, or their dissociated par- ticles, are engaged in a series of correlated chemical reactions. The larger aggregates may be conceived as either not taking part directly in chemical activity, or as helping in the control and localization of the chemical reactions, just as in a photo- graphic dry plate, the presence of the gelatin makes possible a high degree of localization of the photo-chemical reaction. We aimed, therefore, to stop the chemical activities at definite given stages during the growth period and then to observe the differ- ences which could be demonstrated. From such data we then drew conclusions as to the nature of the transformations which had occurred in the interval. The methods of collecting the material were devised with this end in view, namely, to stop all chemical activity as rapidly and completely as possible. The sources of error due to post mortem changes then became constant, and we are in reality following a principle that has long been in use in histological studies. For the preserving agent, alcohol was selected, as it is the least apt to interfere with the further chemical procedure, and, in fact, treatment with alcohol represents a step in the process. In the selection of the chemical methods for this series two points were kept in mind: 1. The necessity of correlating the chemical observations with the known facts of structure, to the interpretation of which they should add a greater precision. As an example of this, there were studied the sulphatides (lipoid sulphur) which are intimately associated in the nervous system with the sheaths of the medul- lated nerve fibers. 2. The collection of data, which, correlated with function, would give the physiologist a better knowledge of the nature of his material and thus enable him to do more than speculate as to the probable nature of the processes involved in the phenomena W. Koch and M. L. Koch 425 he is observing. As an example of this, there was studied the ratio between neutral sulphur and protein sulphur, a ratio which correlates closely with the decrease in metabolic activity associated with the growth of the nervous system from birth to maturity. The general plan of the chemical technique has been first to block out the material into larger groups of substances and then carry the procedure of separation into greater detail. The neces- sity of working with data which represent something definite from the point of view of the chemist, has also been kept in mind. The following outline illustrates the extent to which the chemi- cal procedure has been carried up to the present. Outline illustrating the separation of constituents ~by the method employed, z classified according to their state of aggregation. ENCEPHALON DIVIDED BY STATE OP AGGREGATION INTO: COLLOIDAL (FRACTION 1 AND 4) NON-COLLOIDAL (FRACTION 2 AND 3) Proteins (Fract. 4) Lipoids (Fract. 1) Organic Extractives (Fract. 2 and 3) Inorganic Constituents (Fract. 2 and 3) Proximate c o n - stituents (Include sup- porting structures) Phosphatides Cerebrosides Sulphatides ( Cholesterol \Undetermined Sodium Potassium Calcium Magnesium Chlorides Sulphur c o m bi - nations Protein S Lipoid S Neutral S Inorganic S (sulphates) Phosphorus com- binations Protein P Lipoid P Organic e x - tractives P Inorganic P (phosphates) For an explanation of the chemical procedure followed for this separation the following outline has been inserted. 2 The method employed for this separation is described in an earlier paper by W. Koch and coworkers: Journ. of the Amer. Chem. Soc., xxxi, pp. 1342-1361, 1909. 426 Chemical Differentiation of the Brain Moist Brain Tissue: Add alcohol and extract alternately with alcohol and ether.* EXTRACT (FRACTION 1 AND 2) RESIDUE (FRACTION 3 AND 4) EVAPORATE TO DRYNE8S, EMULSIFY WITH WATER, PPT. WITH CHClsIN 0.5 PER CENT HC1 SOLUTION DRY, WEIGH AND EXTRACT WITH HOT WATER Ppt. (Fract. 1) : Filtrate (Fract. 2) : Filtrate (Fract. 3): Residue (Fract. 4) : Lipoids Organic extrac- tives Inorganic consti- tuents Organic extrac- tives Inorganic consti- tuents Proteins Organic extractives in Fraction 2 and 3 are equal to total organic ex- tractives. Inorganic constituents in Fraction 2 and 3 are equal to total inorganic constituents. Fraction 1 and 2 are soluble in alcohol (85-95 per cent). Fraction 3 is insoluble in alcohol; soluble in hot water. Fraction 4 is insoluble in alcohol and hot water. For a clearer understanding of the terms used in this series of papers, the following interpretation of the chemical nature, anatomical distribution, and physiological significance of the sub- stances determined, with special reference to the nervous system based both on the studies already made and those presented in this paper, is given below. Proteins. Chemistry. These represent complex combinations of ammo-acids ren- dered insoluble in water by coagulation with hot alcohol. This fraction has been exhaustively extracted with hot alcohol and should retain only traces of lipoids and fats. The nucleoproteins and the neurokeratin are included in this fraction. Anatomical distribution. In the part of the nervous system rich in cells (cortex) the proportion of the proteins is larger than in the white matter. Some of the nucleoproteins are supposed to be associated with the chro- matin and Nissl substance of the nerve cell. The remainder of the nucleo- proteins are represented by the nuclei of the glia cells scattered through- 3 Although ether is used in the extraction following the first alcohol, it does not remove any considerable amount of material and need not be con- sidered in the above scheme. W. Koch and M. L. Koch 427 out the nervous system. Neurokeratin occurs in the medullated sheath of the nerve fiber. The other proteins occur in the axon of the nerve fiber as well as in the cell body and its dendrites. Physiological significance. The proteins have usually been considered as the essentially living part of the protoplasm, but some of them, like neurokeratin, are undoubtedly inactive and represent supporting struc- tures. The same may be said of the proteins which make up the fibers of the glia cells. It is therefore impossible to tell at the present time to just what extent and in what proportion the proteins are involved in the chemical activities of the nervous system. The significance of the neutral sulphur compounds, which represent simpler cleavage products of the larger protein aggregates, will be discussed later as having an important bearing on this point (see p. 431). Phosphatides. Chemistry. These represent complex combinations of fatty acids, phos- phoric acid, glycerin, and nitrogen complexes of the nature of choline, and include among other things lecithin and kephalin. The chemistry of this group is very much in need of revision, as some of its members are not so simple as the older work of Hoppe-Seyler has led us to infer. The group does not include lecithin in combination with sulphur or cerebrin. The phosphatides as here given are calculated from the phosphorus of the lipoid fraction on the assumption that they have an average molecular weight of 800. Correction must be made for the phosphorus of the sulphatides.4 Anatomical distribution. Comparison of cortex and corpus callosum5 in- dicates that the phosphatides are not very differently distributed between cell body and nerve fiber. Analyses of the brain at a period when medullation has not begun, but when the cell processes are growing freely, indicate that the phosphatides are largely associated with the axon. If mitochon- dria consist largely of phosphatides, as has been suggested, the observa- tions of Cowdry would give us a picture of their distribution in the cell body. The absence of mitochondria in the axon, which is known to con- tain phosphatides, would not argue against this, as there is some evidence that the phosphatides of the processes and the cell body are different in their behavior. Physiological significance. The phosphatides, like the proteins, may be considered to be intimately associated with the vital processes of the living protoplasm. Their colloidal nature and relation to inorganic ions, as well 4 Calculations for phosphatides. The total lipoid phosphorus found times 25.77 gives the phosphatides, on the basis that 3.88 per cent of the phospha- tides consist of phosphorus. Since 51.2 per cent of the sulphatides are phos- phatides, that amount was deducted from the total phosphatides found. The difference was considered as free phosphatides. 5 Koch, W.: Amer. Journ. of Physiol, xi, pp. 326-328, 1904. 428 Chemical Differentiation of the Brain as their instability towards heat,6 lend support to this idea. They prob- ably occur largely in the cytoplasm, cell body and its branches, where they may act as oxygen carriers, as has been suggested by the work of Koch and Mostrom.7 Cerebrosides. Chemistry. Complex combinations of fatty acids, galactose, and possi- bly other hexoses with a nitrogen complex of the nature of sphingosine. The cerebrosides are calculated from the lipoid sugar on the assumption that they yield on hydrolysis 21.8 per cent of reducing sugar, the amount found by Thierfelder in his cerebron. Correction must be made for the cerebrin content of the sulphatides.8 Anatomical distribution. Although the cerebrosides are occasionally met with in other tissues, they occur in largest amount in the medullated nerve fiber, and their quantity increases as medullation proceeds. The rather large amount found in the cortex9 on chemical analysis indicates that they may predominate in the fibers of that region. Physiological significance. As laid down in the medullated nerve fiber, the cerebrosides most probably serve only a mechanical function and are not available as sources of energy in spite of their carbohydrate and fatty acid content. Sulphatides. Chemistry. These represent the combination of a phosphatide with a cerebroside by means of a sulphuric acid group in ester combination.10 The sulphatides are estimated from the lipoid sulphur on the basis of a sulphur content of 2 per cent, based on the analysis of a purified compound.11 6 Koch, W and Koch, M. L. : this Journal, xiv, pp. 281-282, 1913. 7 Koch, W and Mostrom, H. T. : Journ. of Pharm. and Exp. Ther., ii, No. 3, p. 265, 1910. 8 Calculations for cerebrosides. The cerebrosides, from the lipoid frac- tion, on hydrolysis for twenty-four hours with a weak solution of HC1 (75 cc. of water containing 3 cc. concentrated HC1), yield 21.8 per cent by weight galactose. The calculations for cerebrosides were made on the assumption that galactose and glucose were equivalent in reducing power and the weight of galactose was thus determined from Munson and Walker's tables for glucose. (Journ. Amer. Chem. Soc., xxviii, p. 663) . The corrected weight of total galactose to cerebrosides was then made on the basis that 21.8 per cent of the latter is galactose. Finally since 42.9 per cent of the sulphatides consist of cerebrosides this amount was deducted from the total cerebrosides found. The difference was considered as cerebrosides. 9 Koch, W. and Mann, S. A. : Archives of Neurology and Psychiatry, iv, p. 33, 1909. 10 Koch, W.: Zeitschr.f. physiol Chem., Ixx, p. 94, 1910. 11 Calculations for sulphatides. These are considered to be of the gen- eral formula: W. Koch and M. L. Koch 429 If it were desirable to recognize the chemical identity of the much abused protagon, the sulphatides might be considered as purified products. Pro- tagon could be much more safely calculated from the lipoid sulphur than from the lipoid sugar as Noll12 has attempted. The sulphur content of protagon preparations, when it has not been simply ignored, is variously reported as 0.5 and 1.0 per cent. Anatomical distribution. The sulphatides, like the cerebrosides, increase parallel with the growth of the medullary sheath and may be considered as essential constituents of that structure. The fact that the sulphatides, as the result of more recent work, have been found to be pretty generally distributed in other tissues, indicates that they might occur in the cell body of the neurone, although a comparison of the analyses of cortex and corpus callosum does not make this very probable. The sulphatides have an important function in the maturing of the nerve fiber and give the Weigert staining reaction in a very characteristic manner. Physiological significance. Their colloidal nature and the peculiar com- bination into which the sulphatides enter with potassium, suggests that they may have an important relation to the nerve impulse and to the phe- nomena of conductivity in general. Organic extractives and inorganic constituents. Chemistry. This group represents essentially the water-soluble, non- colloidal constituents of the nervous system. The older method of esti- mating the inorganic constituents by the ash has been abandoned as too inaccurate. The principal reason for reporting the above group is to give an idea of the ratio between the colloidal and non-colloidal constituents. Anatomical distribution. The group occurs in large quantity in the cell body, although some is also present in the axon of the nerve fiber. Physiological significance. This group is a rough index of the amount of metabolic activity going on in the tissue, as it represents at the same time the end products of chemical activity, as well as the culture media from which the more complex combinations are built up. Undetermined (cholesterol). This fraction is represented in the nervous system to a certain extent by cholesterol, which has not been directly estimated. Besides this, how- O II (Phosphatides) x — O — S — O — (Cerebrosides) y II O containing 2.0 per cent sulphur, 42.9 per cent cerebrosides, and 51.2 per cent phosphatides. Then we have, (Lipoid sulphur X 50) = ^ rf su, hat;des ;n dry substance. weight of dry substance 12 Noll, A. : Zeitschr. f. physiol. Chem., xxvii, p. 370, 1899. 43O Chemical Differentiation of the Brain ever, all the errors of analysis, as well as of such calculations as are based on assumed factors, enter into this fraction. After accounting for. the cholesterol in the brain of the 50 and 100 mm. pig fetus,13 in which this was estimated directly by Mendel, there remained undetermined 2 to 3 per cent of the total solids. Considering the number of groups estimated, this is not a very discouraging result. (In other tissues, which contain little cholesterol, the undetermined is recorded as neutral fat.) Anatomical distribution. Cholesterol is principally of interest as a con- stituent of the medullary sheath to which it adds a sort of mechanical stability. But it is present in the cell bodies also, possibly contributing to the cell membranes. According to Lorrain Smith14 it is one of the sub- stances responsible for the color which the medullary sheath gives with Weigert's stain. Total sulphur and total phosphorus. It may not be out of place to state briefly the reasons for selecting these two elements for special determination in preference to others. As far as the phosphorus is concerned, the importance of the nucleins to all liv- ing cells and the phosphatides to the nervous system in particular, amply justify its selection. The reason for selecting sulphur in preference to the much more generally studied nitrogen, may,- however need a word of explanation. Nitrogen is studied for two reasons: because it is an important element in the building up of the proteins, and because it is easy of estimation. Sulphur is just as characteristic of proteins, in fact more so, as it does not enter into the non-protein groups such as the nucleic acids. Among the lipoids, too, sulphur enters into only one group, the sulphatides, while nitrogen occurs in all except cholesterol. In other words, to estimate sulphur in the protein fraction is to esti- mate an element essentially characteristic of the more truly protein part. To estimate it in the lipoid fraction, enables one to distinguish one par- ticurar, and, as growth curves show, a very interesting group of lipoids. Besides, as has already been pointed out in a previous paper15 sulphur occurs in the tissues in several states of oxidation and thus gives us some indication of the intensity of reactions of oxidation which are so impor- tant to growing tissues, and about which we know so little. It seems wise therefore to estimate the sulphur, and in case there are any special reasons to study nitrogen, to study it rather in the form of one of its definite groups of compounds such as the purine bases or the amino-acids. 13 Koch, Mathilde L. : this Journal, xiv, pp. 267-279, 1913. 14 Smith, Lorrain: Journ. of Path, and Bact., xv, pp. 179-181, 1911. 15 Koch, W. and Upson, F. W. : Proc. Soc. for Exp. Biol. and Med., vii, pp. 5-6, 1909. W. Koch and M. L. Koch Distribution of sulphur. Chemistry. PROTEIN S. This group represents sulphur in various amino- acid combinations such as cystine or cysteine. The proteins in which this sulphur fraction is found have been coagulated and rendered insoluble in water by treatment with hot alcohol. LIPOID S. Ethereal sulphuric acid combinations are discussed under sulphatides. NEUTRAL S. This group of compounds represents the total non-col- loidal, water-soluble combinations of sulphur, minus the inorganic sul- phates. As far as studied, they resemble in all their reactions a similar group found in the urine and called by Bondzynski proteinic acids. They represent, probably, larger cleavage products of the protein molecule or complex non-coagulable, water-soluble polypeptides somewhat altered by processes of oxidation. The sulphur of this fraction is represented essentially by compounds included among the organic extractives, and the sulphur is most often in an oxidized form like taurine or ethereal sulphate. INORGANIC S (inorganic sulphates). Derivatives of sulphur directly precipitated by barium chloride in hydrochloric acid solution. Anatomical distribution and physiological significance. The LIPOID S, as has already been mentioned under the sulphatides, represents an essen- tial constituent of the medullary sheath. The proportion in which it occurs in the sheath can be considered as a measure of the maturity of the sheathing substance. THE PROTEIN S AND NEUTRAL S will be considered together as they bear an important relation to one another and as the combinations in which they occur are essential constituents of all living cells. As has already been stated, the study of these two groups of sulphur compounds gives us a means of investigating the protein metabolism of the central nervous system during its growth period. TABLE I. A comparison of neutral sulphur with protein sulphur in the brain of the albino rat at different ages (figures in per cent of total sulphur}. PROTEIN SULPHUR NEUTRAL SULPHUR 1 day 30.5 48.2 10 days 44.2 45.4 20 days 56.4 28.6 40 days . 63.7 18.2 120 days . 61.8 18.7 210 days . ... 63.8 14.5 During the early stages when growth is proceeding rapidly and chem- ical activities may be considered to be at their height, the proportion of 432 Chemical Differentiation of the Brain non-colloidal, relatively smaller, neutral sulphur molecules is at a maxi- mum. This is what we should expect when we consider living matter not as a collection of highly organized molecules, but rather as a hetero- geneous substratum in which relatively smaller molecules or their dissoci- ated products are engaged in chemical transformations. As the tissues grow and become more highly differentiated and mature, more and more protein is laid down as structural material, and the proportion is shifted . in the direction of the protein sulphur. A comparison of the cortex of the human at two years and at maturity illustrates this point.16 2 years' cortex Protein S, 63; Neutral S, 22. 19 years' cortex Protein S, 73; Neutral S, 12. The change suggests, therefore, a decrease in chemically active mate- rial associated with the increasing complexity of the tissue. Such data as we have at hand indicate that we have in the protein sulphur and neu- tral sulphur ratio a valuable means of measuring the relative growth intensity of the nervous system at different periods during its development after the state of cell division has practically ceased. There might be another way of measuring this intensity of chemical activity, namely, by means of the inorganic^ sulphates, which represent the end products and the final state of oxidation of the compounds involved in these reactions, but they are eliminated rather easily from the cell, and it is therefore difficult to attach any significance to their variations. Distribution of phosphorus. Chemistry. PROTEIN P. This group represents phosphorus largely in combination as nucleic acid. In the nervous system this nucleic acid is combined with such a very large amount of protein17 that the per cent of phosphorus in the resulting nucleoprotein drops to 0.57 per cent as com- pared with 3 to 4 per cent in such a tissue as the pancreas. LIPOID P. Already discussed under phosphatides. WATER-SOLUBLE P. This group includes non-colloidal, water-soluble organic combinations of phosphoric acid and inorganic phosphates. On account of the relative ease with which the organic extractive combina- tions of this form of phosphoric acid break down, it is difficult to estimate the proportion which is in organic combination. The results which are so far recorded represent, therefore, rather the possible maximum, than a very close approach to the actual value. (See articles of Grindley,18 Trow- bridge,19 and Forbes.20) 16 Koch, W. and Mann, S. A.: Journ. of PhysioL, xxxvi, p. 2, 1907. 17 Also accounts for poor staining reaction of neurone nucleus. 18 Grindley: Journ. of Amer. Chem. Soc., xxviii, pp. 25-63, 1906. 19 Trowbridge, P. F. and Francis, C. K.: This Journal, vii, pp. 481-501, 1910. 20 Forbes, E. B. : Ohio Agric. Exp. Sta. Bulletin 215, 1910, pp. 459-489. W. Koch and M. L. Koch 433 Anatomical distribution and physiological significance. The protein phos- phorus is largely associated with the nucleic acid of the nucleus. In the nervous system it is also supposed to be associated with the Nissl sub- stance, but this is still a doubtful matter. The accuracy with which we can estimate nuclear material in the anatomical sense from such a figure as the protein phosphorus is difficult to determine, as there are three complicating factors. 1. The possibility that the nucleus, as an anatomical unit, contains other compounds besides nucleoproteins. 2. The fact that nucleic acid itself may be associated with very widely varying quantities of protein. 3. The possibility that substances yielding a protein phosphorus frac- tion may occur in the cytoplasm. Observations by Miescher21 on the sperm, however, very strongly sug- gest that protein phosphorus is largely associated with the nucleus, while lipoid phosphorus is largely associated with the cytoplasm. As regards the water-soluble phosphorus, the principal point of interest, just as in the case of the neutral sulphur, is its ratio to the protein or col- loidal forms of phosphorus. Thus in a study on a lower plant form (Asper- gillus niger] Koch and Reed22 could demonstrate that under extreme con- ditions, such as can only be realized with plant material, it is possible to carry the growth processes to such a point that all the non-colloidal, water-soluble phosphorus is converted into colloidal combinations. At such a point the growth of the plant comes to a stop. The function of the inorganic phosphates in maintaining the neutrality of protoplasm as suggested by Henderson23 is also a point of interest, although of less importance to the nervous system than to muscle tissue. Inorganic constituents. Chemistry. The inorganic constituents found in the nervous system are the cations Na, K, Ca, Mg, Fe, and the anions Cl, SO4, P04. In the method devised for this study the usual method of estimating these constituents by the ash was abandoned on account of the fact that by the process of ashing, the relation of cations to anions is profoundly altered. As a result of this precaution, the interesting fact has been clearly demonstrated that a large proportion of the cations, especially sodium and potassium, occur combined with complex anions, sometimes colloidal in nature. The work of Pike*4 has brought to light the interesting point that in the nervous system the sodium and potassium, more especially the latter, 21 Miescher, F. : Hoppe-Seyler's Med.-chem. Unters., p. 452. 22 Koch, W. and Reed, H. S.: this Journal, iii, p. 49, 1907. 23 Henderson, L. J. : this Journal, vii, pp. 29-35, 1910. 24 Koch, W. and Pike, F. H.: Journ. of Pharm. and Exp. Ther., ii, pp. 245-248, 1910. THE JOURNAL OF BIOLOGICAL CHEMISTRY, VOL. XV, NO. 3 434 Chemical Differentiation of the Brain are combined with such lipoids as the sulphatides and kephalines (a sub- group of phosphatides), while the Ca and Mg have more tendency to remain combined with the proteins. Anatomical distribution and physiological significance. Very little is known of the anatomical distribution of the salts except as shown by the work of Macallum25 which demonstrates that chlorides and potassium are associated with the nerve fiber. According to Alcock,26 potassium is sup- posed to play an important role in the propagation of the nerve impulse. With this introductory statement of the anatomical distribu- tion and physiological significance of the substances quantitatively determined, we may now present the results of a study of their variation during the growth of the brain. The brain of the albino rat was selected for this study, for the reasons already presented in the first paper of this series.27 From a comparison of the brain of the albino rat at birth and the brain of the fetal pig, it was found that the brain of the new born rat is as young nervous material as can conveniently be analyzed at present. It forms therefore a suitable starting point for this study of chemical differentiation during growth. The analyses reported in this paper are those of the brains of rats aged respectively, 1, 10, 20, 40, 120, and 210 days. The results show that it was possible to follow closely the various structural changes which occur during the differentiation of the growing nervous system The material was furnished by the Wistar Institute of Anatomy; the brains being collected and analyzed in the manner already detailed.28 Koch's quantitative methods were used.29 RESULTS OF ANALYSES. The results of analyses are embodied in Table II. Duplicate analyses have been carried on throughout, and are summarized in Table III. This table gives the averages of the analyses, except 25 Macallum, A. B. : Journ. ofPhysioL, xxxii, pp. 95-128, 1905; Macallum, A. B. and Menten, M. L. : Report 75th Meeting British Assoc. Adv. Sti., p. 555, 1906. 26 Alcock, N. H.: Journ. of Physiol., xxxix, pp. 402-410, 1911. 27 Koch, Mathilde L. : this Journal, xiv, pp. 267-279, 1913. 28 Koch, Mathilde L.: loc. cit. 29 Koch, W.: Journ. of Amer. Chem. Soc., xxxi, pp. 1335-1364, 1909. W. Koch and M. L. Koch 435 in two instances where the value from one analysis only is pre- ferred: Table IV gives the absolute weights of these constituents, as found in one brain; while Table V gives the ratio of increase of the different constituents, taking the amount of each constitu- ent in the brain of the rat at birth to be unity and determining the number of times each constituent had increased at successive ages from birth to maturity. For comparison there is given in Table II one analysis of the spinal cord at 120 days. DISCUSSION OF RESULTS. The growth of the nervous system from the first laying down of the neural canal to maturity may be divided into four periods. The first period, during which cell division is the most charac- teristic feature, lasts to about birth. A short time before birth cell division begins to decrease. The chemical changes during this first period were not studied directly in the albino rat for the reasons stated in the first paper30 but the composition of the nervous system in this primitive, undifferentiated state may be seen in the analysis of the fetal pig brain reported in the first paper of the series. At this time phosphatides are present, sul- phatides are relatively less important and cerebrosides are en- tirely lacking; proteins, phosphatides, extractives, salts and water are the predominant constituents of the tissue. The second period (see Table VI) lasts from birth for about ten days, when the third period begins. The second period is characterized structurally by the development of fibers from the cells and the increase in their size. Donaldson has estimated that the number of nerve cells does not increase more than 3 to 6 per cent during this period, but the cells do add to the number and size of their branching processes. This period, as may be seen from Table VI, is one of intense growth of all the solid con- stituents. The proteins continue throughout this period to be formed at a very rapid rate, 4-5 mgms. being laid down per day. Cerebrosides are either absent entirely, or present in very small quantities. In the third period, that of most rapid growth, from the tenth to the twentieth day, medullation begins. There is a wonderful 30 Koch, M. L. : this Journal, xiv, p. 279, 1913. 436 Chemical Differentiation of the Brain s § V. g *J 1 CO to os IO TjH 'e 8 1 CM CO rH O O • OS k OOCOtOO CO OOTtHT^ | s C^ to C^ !>• t^» ^^ CO rH CO » ^H i— 1 00 OCO-^CO 00 OSlOTt* . . co r^ OOr-ioOCO OS 00 O rH rH rH O •^ CM § 3 tv-. rH §»O O t^ CMOSCOtO l>- rHlO^t1 CO CO CO ^ l>-rHCOCO OS rHOrH rH rH O "^ CNJ rH i OO -tOCOI>- 00 OiOtO £! ^ CO rH (M t^ ^ * 1 GOOCOCNJ CO Oodr-i fe 1-1 ^ ° **-* -* CM rH •€ « -8 9 -3 0 o OS CO -COCO * *c° pD "2 rHt^ 6 ^ ^ _s co'rHco'cM co iodi-i 10 CM rH e 5 ^ OO tO rH CO • •HH O CM O g 42 1 *-S IOCO*CO rH tOOOT^ CO CM CM to CO O rH tO rH rH rH a 0 rH CO £: P 1 »o 00 t^ rH -^ CO o CO CM OO •^CO*l^ CO O1>-CM 1 O TJH O rH O O O OS CO O rH •s? 0 N | g J*j Tt< CM COCO*Tfl CO "*COOS •«* 10 0 0 0 ^ rH ^ 00 IO rH OS tO rH rH tO rH rH '1 to CM Tt< O O «D CMOO*tO to OOSOO 1 d o d ^ ^ GO -^ r-i CO OS O rH o 2 °2 ' ^— ' S P3 • f N Ml B g •1 1 * . . ^ i •2 c -S ; ; *3 : 6 .2 ;s a : : .2 : OQ ?H *C3 M CQ ^ S "2 -^ .2 : : m S-« '• h S +"* r^ ^ ': '• > 5 r§ " bC .2 'o 1 £ '3 1 j : 1 11 t}| +5 "tn ^* ri^ B ™ cS m h ° 2 ft [3 bC Qj rl * _. "S bO a §3r§ * 1 ° ft § ^ 'o ^i 02 03 2R • J5 ... 'In <^ 'd ^3 Jj >> 5 '« |J |S >s^2^-|g^^a S ft4243 fl Sj^rsrs I 1 lill 1 -SoQJ^S,S'o^^ ^43 QJ ''3 ™243 O O PHPHO^OKHUHH W. Koch and M. L. Koch 437 t*- OS CO i-H CO ^ O 00 CO tO rH CO 10 ^ CO CO i— 1 i— 1 00 CO CO CO 1>- to CO C^ "* 1O CO 00 » C^ ^^ CO CO . to co CO 1 O5 i-l O | i 10 o^ ts* oo CO i-H 2 °>%3> ! ,, Si in & CO »O »O t» 00 to t^ _g io ^ ^ a § 0 £§ § oj I »O t>» 1>- i—t o O C. CO O5 CO & CO (M i-t to ^t1 5 CO S-^ C^ i— 1 Tfi CO "S OS i-i O fl o' g ^j CO JO TH ^ 2c€ § I "S -0 CO (M i-H i-H o '•2 O 00 O CO o _J _ " ^* ^^ i^X /"VS tJ CO ^ ^H CO to 3 i-H 2. We may call M the mass of cohesion of a molecule, and K is a constant of proportion. This same value may be obtained directly by supposing that molecules attract each other inversely as the fourth power of the distance between their centers, directly as the product of their cohesive masses 156 Albert P. Mathews and that each molecule attracts only the six surrounding molecules. In another paper I shall show that the value M2K is proportional to the two- thirds power of the product of the molecular weight and the number of valences in the molecule. In this paper I wish to show how M2K may be derived from the surface tension. The computations are made in absolute units. The formula, S == fK/3, states that the surface tension of a liquid is a function of the cohesive pressure of the liquid alone. It is clear, then, that this formula can hold only at very low temperatures, since only at such temperatures can the cohesive pressure in the vapor be neglected. The surface tension, strictly speaking, can not represent the cohesive pressure of the liquid alone, since it is in the very nature of things an expression of the difference in cohesive energy of the liquid and vapor. I shall, then, take the formula as holding at absolute zero, since, if we are going to a tempera- ture in which the vapor may be entirely neglected, it is more convenient to go to the end. The radius of action of the cohesive attraction, or r, has been found, both by calculation and by direct measurement, to be, at higher temperatures, very nearly equal to the distance between the molecular centers; and at absolute zero, with the molecules in contact, we may safely assume that "r" is equal to vl/'\ K is Laplace's constant and is equal to a/V2, or M2K/i;2. The formula becomes then: S = DQ 1/3M2K/>0 2 = M2K/3i;0 5/3. By multiplying both sides of the equation by v02/3 we have, $v0 2/3 == M2K/3?;0 ; v0, the volume at the disposal of one molecule at absolute zero, is, for sub- stances of medium complexity such as ether, very nearly equal to TJC, the volume of a molecule at the critical tempera- ture, divided by 4. For simpler substances, such as O2 or CO2, the volume v0 is equal to Vc/3.63; and for more complex substances, such as octane, it is 1^/4.04, or for some even i;c/4.io. The volume of a molecule is obtained by dividing the volume of a gram mol by 6.21 X io23, which is the most probable number of molecules in a gram mol. There is a Determining "a" of van der Waals' Equation 157 difference of opinion as to the value of the coefficient which is given as 1/3 by Young,1 but as 3/20 by Rayleigh.2 As will be shown presently, Young's value is to be preferred. The value of Si^02/3, the molecular surface tension energy at absolute zero, may be obtained from Eotvos3 and Ramsay and Shields'4 rule, $(mX rt .s s w tS rt vo 1- ^ NO 4) i* cj ininininininioioinioinioininioiominioininininioioi^i* cococococoeococo/cococococococococococococoeocococoro ro bbbbbbbbbbobbbooobbbbbbbbco xxxxxxxxxxxxxxxxxxxxxxxxxxx CO CN CN ON t-H CN CN >O CN CN M t— I OO OO OO CO SO Tf TJ- O CN VO CO CN >-H HH <•<•} OOOT^"OOCNON'^t''^"ONCNi-HCNON'^hOOCv< t^1 O CO "^ ON OO CN ON >-> CN O CN t^OOO ONI-" CNSO iO t^- ONSO -^-lOi-iOO ioO »-" i-" cOcocOt^t^M t~» Q\ 10 sO ^J" CN »O *O sO £"*• OO T}- sO sO SO OO ON rj- iO •^" ^ >O >O >O so sO "^ ininioinin»nioininininininmininininininininin-inin»^io b b b b b b b b b b b b b b b b b b b b b b b b b b b XXXXXXXXXXXX X'X XXXXXXXXXXXXX ON M OO M ON CO sO iO l^ O *O t-n rj- i— i T^- ro O ^ ONOO ON O t~^ r>- co CN I-H ON-^-t-H CNOO ONCNSO M HH rOOO OO ON O O iO ON cOOO T|- M ON t-*- CN r^- rf ^* io ^ ^t" O ON 10 *^ O oo t— < M r^* 10 i^* co O ^"^ CN M t-H 10 co co oo r^1* CN I"** ON sO sO ^O CN iO iO !**• !>• ON O iO SO O CO ininininininininininmininininininminmininininioi'5t-N eococococoeoeocococoeocococococoeococo«ococococococo.-5 '2 *2 '2 '2 k '2 2 '2 '2 2 '2 '2 k '2 2 2 '2 2 2 2 2 2 2 2 2 2 2 xxxxxxxxxxxxxxxxxxxxxxxxxxx CN V. CN O iO t-i ONSO SO voO CN Tj-cocot-< coO HH ^\ Qs HH io ON ON ON t~^ TJ-O ONiO^ONl^O ^ O O t^ ON u->vO OOcO-^-Ot^^OO I>-OO ONOO CN CO COOO rj- O t^ t^ >-< iO t^. lOSO CNOt^CNWMiOCO COOO t^ ON io\o SO oo O vO *£ a S o ' . I-H 'o O 00 CO ^ T3 ^ OJ en II +•» -4-> II oJ a3 $ 0 Is 1 £> *^v^^ V5 C/) r^ O QJ CJ m > >> - co ^ ^ co co II M S C M M II II oJ a3 II H E 0 HH aa X §CN 10 coOOO T|- co fN ^i" O *— ' iOC ON O GO ON >- ONONC O O ONONONQ ON O ON 00 O ON 00 ON <• OONOOOOO COCOO H ON ON ON t^1* t~^ O OC ; oo ^ ON ro (vjfMr^romrofvjrM O fM rO fM fM CO fM PM CM •OCMCMfMCMfMmfN o . CN CO ^§ (N (NOO CN (NO OOOOOOOOOOOO O ^h ^ ^ M C D 0 ^- ^ 0 0 0 0 R 1 6 -» COCOCOC OM CS M CO^COCOCOI O^OcocOcOcO"-1 i- IT £*« PQ to - -ffi £ Tf 00 OO ^ rr' CMCN^Tt-TJ-OGOTj-O^ ^GOGOOOO O CN'M.CN ' f^(N CN r^ThcoC T3 f-H ° r-^t^i^r^r^ot^GOoooooooocooNOOO~>-i'-i'-'OOioOioo " t— 4 * El 1 8 15 tH^ 5 S5 ff ? 1 ?^5 §1 1 ?£ S- ^ § Ml ffi 3 cf 5* >-iOOOfX^I>.MOOCC 0 3 *? *? *? ^ ^^ °° ***. C D ONOOGOOO ONt^oo a bio »o 10 10 ^o *o 10 10 10 10 10 *o ^o *o ^o ^o 10 *o ^o T ^-lOiOiOtOiOiOiO '-4J o cocococococococococ Ocococococococococ Ocococococococ' 5 .ts h3 1 1 1 11 1 1 1 I rt "^ M *Qooo ,o"o"o"E .«0 » «.C1<-1« w . 2 1 \£jtijtij£j£j£R^^ ^ 0 M CN CO^-^0 ^CSCSCSdCSCNCS Value of "a" of -van der Waals' Equation 185 It will be seen in Table III that C has a mean value of 2.98 X io~37 for the twenty- six non-associating, or nearly non-asso- ciating, substances studied by Young. The greatest devia- tion from, the mean is octane on the one side, which deviates a little more than four percent; and on the other, fluorben- zene, which deviates 6 percent, methyl isobutyrate deviating a little over 3 percent, and brom- and iodobenzene. The deviation of the last two substances, about 7 percent, may, I think, be disregarded since the critical temperature and pres- sure were estimated and not directly determined; the critical data are therefore less certain. The other twenty-one sub- stances do not deviate more than 3 percent from the mean. The cause of the deviation of fluorbenzene is uncertain, but there is a regularity in the other deviations which suggests that the size, shape, or compressibility of the molecule may play a slight role in determining the critical data, or the co- efficients of our formulas. For it appears that M2K, and hence C, is always lower in the iso-compounds than in the cor- responding normal compounds. Diisobutyl has C of the mean value, whereas normal octane has a high value for C; a similar relationship exists between diisopropyl and hexane; and be- tween iso- and normal pentane. Some of this difference would disappear if, instead of taking v0 equal to vc/4, I had used the accurate ratio computed from the law of rectilinear diameter. Thus V0 of isopentane is only Vc/3-92; while that of pentane is Vc/3-96 and octane is Vc/4.°4- The sub- stitution of these values for V0 in computing M2K would have made C of isopentane 2.98; of pentane, 3 . 03 . The difference is not entirely due then to this factor. It seems more proba- ble to me that bc is not always exactly the same fraction of Vc and that consequently the coefficients of the formulas used in computing M2K are not exactly the same in all substances. Notwithstanding x these slight deviations, the constancy of C is certainly remarkably good and shows beyond ques- tion, I think, that a close connection exists between molecu- lar cohesion and. the molecular weight and number of valences in the molecule. 1 86 Albert P. Mathews w used M2K U "o X 0 y 6 O O CO ... _ oo >-i . co^, CN ', Q »-4 ; . M. « • O • :** u ««•»•• ">r^!ll co co O ON ^ ro ^ « n ,i i, P O O O *O /s CL,n .-Pn •-.-•- ^,§ 10 — ^° ^ 10 - 00 >O •-" "-• O'-|t^lO OOO ON ' O f" "-" 00 O ON ONO O^CNi-iQ OO CN COCOCOCNCOCNCNCMCOCOCNCOCO COCN 10 co in t^ uo ON GO -* 00 t^ ON CO co rh^o voooo O \O O ON . CO cococococococococococococo coco cOcOcocOcO^ u u"uuucjcjuuuucruu uo DOcToo0 3 ! 1> .= U — a>'n.;=i'o liiil «L> C C a; (D N t^QO ON O -i HH .-( 1-1 CS CN Value of "a" of van der Waals' Equation 187 vO CO ii «« o°° 0^ ^ 10 rj- o '-' O (NCNCNCNCNfO ON-IOOOOO ON(N«-I>-IOOQNO u-j TJ- h-i CO 'vh I-" ^ to CO ^h ON § , O Q iO « ^B^^^^^^... co O'OCN CN O O >-i -H CO O GO (N IO CO I CO to O 00 r^OO OO rf M to *-O ^^ ^O *-O ^O *-O ^O *O *-O *O *-O ^O ^O ^^ "^~ ^f" to to to coco CO CO .CO CO CO CO cOcOcOcOcO cOcOcOcOcOcOcO vOO1^- 10 ON -i I-H O ON ON 00 00 ON O . , a2s oo ooOooo ooooo ooooooo -S -- 3 (N CO o-- - tilll WXHHO O O >-> ™ CO "*• .ff Sl ON O -H 1 88 Albert P. Mathews Table IV is a summary of the values of C computed for forty-one other substances, of many of which the critical data are not so accurately known as those included in Table III. Since for most of these substances dc and Vc were not given in the lyandolt-Bornstein-Meyerhoffer tables, I com- puted Vc by the formula VCPC/TC == 21.92 and then M2K by the formula M2K •• 3.594 X icT40 VCTC. The values of Vc so computed are of course not so accurate. Wherever the critical volume or density was given it was used. In a few cases, which are specified, M2K was obtained by dividing the value of "a," given by Guye and Mallet, by N2, i. e., (2.77 X io19)2. While the values are thus less certain and the deviations somewhat greater, the mean value of all is what it was for the other substances, i. c., C •• •• 2.98 X io"37. The values are surprisingly uniform. It will be noticed that the more com- plex substances such as diphenyl and diphenyl methane have C a little high. This may possibly be due to slight associa- tion or quasi-association in these substances With this and one or two other exceptions the uniformity of C is marked. All substances known to be associating give a value for C higher than 2.98 X io~37 when M2K is computed by the same formulas as for normal substances, and the weight and valence used in computing C are those of the normal non-associated molecule. This result is to be anticipated, since in these substances the molecular weight and valence do not remain the same throughout the temperature interval. The computation of M2K from the critical data is uncertain in the case of associating substances for the reason that there is no certainty that the coefficients of the formulas are the same for these substances as for normal substances. Although the figures for M2K are thus uncertain, I have nevertheless calculated M2K by the usual formula in order to show that these substances deviate in the direction we should expect from the value of C found in normal substances. The re- sults are given in Table V : Value of "a" of van der Waals' Equation 189 Cn 4^. Co to HH O^O CO^J ONCn 4^ Co to H< - i ^ D- d» ft) i-t O 2. 2- PPP 2 Q W 2 p op ppp QP tn K ffi w « t/) B B B B.B Bal" B -0 as o °°b i i i CoCoCoCoCoCoCoCoCoCoCoCoCoCoCo CnCnCnCnCnCnCnCnCnCnCnCnCnCnCn •> M O OJ ON to O to CO CO I-H to 4^. 00 ON to toCoCotO^Mt-itO ONOOONO ONONO Ki s O O ? n ^J 00 ^ •-• OO^O OOON-^JCn OJ N) O 00 004*. HH Cn 00--J O ON 4s- Co 4* 4^ Cn * M 1-1 i-i to Co Ki Cn O Co •^j Cn to • Cn O O • • -• 00-- -• -• co ^r - • .. ., ., || 4^ -^J VO >-- Cn ON K) O •-" Cn 4^. Cn to CO Cn xxxxxxx D O O O( o o o o 190 Albert P. Mathews I have also calculated the value of M2K and the quotient C of a number of other substances from the surface tension measurements of Ramsay and Shields. From the surface tension, the critical temperature may be calculated approxi- mately by the rule of Botvos, revised by Ramsay and Shields, namely, K(TC — T — 6) =- S(M^)2/3, where S is the surface tension in dynes at the absolute temperature T and Mi; the molecular volume. Having thus found T0 the fraction T/TC may be calculated for any temperature and by interpolation from the curve expressing the relation between T/TC and V/VC, V/VC may be found and from this Vc may be calculated if V is known. We thus obtain the theoretical critical volume. From Vc and Tc the value of M2K may be calculated either by the surface tension formula or the formula M2K * • VCTC 3.594 X io~40. I shall, in the first place, show how exactly the law correlating M2K with weight and valence holds for some substances which do not contain chlorine but contain elements of which the valence is fairly certain. Some of these substances associate slightly. The results are summar- ized in Table VlX From Table VI it may be seen that the values obtained for C, although less reliable than those from the other tables, nevertheless agree well with the mean of about 3 X io~37. The substances marked associating in the above table were found to be so by Dutoit and Mojoiu,1 who determined the association and calculated the mean molecular weight at various temperatures. There are two sources of uncertainty in computing "a" and M2K for the simplest gases; the first is the uncertainty of the critical data of some of them; and the second, the un- certainty that the coefficients of the formulas hitherto used for calculating these values remain the same for the gases. If the usual formula for "a" be used, by which "a" is calcu- lated from the pressure and temperature, the formula being derived from certain assumptions as to the value of the crit- 1 Dutoit and Mojoiu: Constante de capillarite et poids moleculaire. Jour. Chim. Phys., 7, 169 (1909). Value of "a" of van der Waals' Equation •-i O ss: 11 - PPPPPPPPPPPP ONCn ON O 00\O :*« >£. 191 \ OO O 4^ vO OOO O CoCoCoCoGJCoCoCo Co CO CnCnCnCnCnCnCnCnCnCnCnCn O NO H- OO^O OO^J J wOJ K) 5 < OJ COCOCOOOJ K) K) O O ONCnCn KJ £L2 4i>CoCoCotoCn-f^4^CoCs)CoK) rnO O K) 4^ -f^ 004^ O ON ON ON4^ OO j 5 ' QCn H^4i. 004^. ONO ONQOON4;>- 52 C O R| p- 192 Albert P. Mathews ical coefficient, the value of C is, for fairly complex substances, about 2.80 X io"37; but in the simple gases, such as N2O, H2, O2, N2, etc., it is uniformly lower and about 2.25. The question is, therefore, whether this formula does not give an exact value for M2K, or whether the relation of M2K to the weight and valence is only approximate and does not remain the same for the simple substances. It is possible that the ratio 27/64 in the "a" formula is not the same for all sub- stances, but may vary a little with the complexity, or com- pressibility, of the molecule. To solve this uncertainty, T have computed M2K directly from the surface tension of the liquid gases by the half empirical formula which holds pretty well for more complex substances: M2K(i/7;1 — i/^) = 3SV*[(TC — 6)/(Tc -- T— 6)]2/» I have used the data of Baly and Donnan in this calculation. I also calculated M2K from the latent heat of vaporization from figures given by Dewar in his article on Liquid Gases in the Encyclopedia Britannica, by the formula: L — E = a(i/V1 — i/VJ. a = N2M2K. This formula neglects the heat used in the expan- sion of the molecules in passing from the liquid to the vapor. It generally gives, therefore, a value of M2K somewhat too high. I have also computed M2K by the usual formulas in- volving Vc and Tc instead of Pc such as the surface tension formula. In Table VII I have included the values thus cal- culated for hydrogen, using successively the critical data of Olszewski and Dewar. It is evident that a considerable uncertainty in the case of hydrogen arises from the uncertainty of the critical data. I think the PCV2C formula gives too high results. If we take the mean of all it would be —37 .88161. With the valence 2 and the weight 2 this would give for "c" a value of 3.02 X io~37, which is close to the value obtained before. I have made similar calculations in the case of oxygen, nitrogen, carbon dioxide and nitrous oxide. For the sake of brevity I omit these and give in Table VIII only the mean value for M2K for each gas. It is clear from Table VIII that with the exception of car- Value of "a" of van der Waals' Equation 193 CO CO CO CO CO CO CO ^J - to i 8 e. 194 Albert P. Mathews X JCL_ o c Vl li8 - -HH gS«5 ON GN O OOO ON O (N rj- CM CO 00 iO 00 to CO to 00 to 8 u> u S S C CJ 1 5 £ Value of "a" of van der Waals' Equation 195 bon monoxide, which is hopelessly aberrant if carbon and oxy- gen have a higher valence than one, the value of C comes close to the value found in other substances, namely, 2.98 X io~37. It is true that in making the calculation I have, in one or two instances, used other valence numbers than those generally attributed to the elements. Thus chlorine is trivalent. But I shall show in a subsequent paper that chlorine is always trivalent in its organic compounds; nitrogen in the gaseous form is univalent, while it is bivalent in nitrous oxide;1 but both these valences have already been ascribed to nitrogen. Oxygen cannot have more than two valences in the molecule. The value I have taken for M2K is certainly as high as the facts warrant and the critical data seem reliable. I believe that the conclusion is justified that oxygen is monovalent in its gaseous form. The surprising fact is the value for car- bon dioxide. The carbon cannot be more than bivalent here, giving a kind of peroxide formula, if carbon dioxide is to follow the rule. On the whole, the gases agree well with the results already obtained for C. In a subsequent paper I shall show that all chlorine com- pounds also follow the rule if the chlorine be taken trivalent; and that there is good reason, from quite independent sources, to attribute a valence of three to chlorine. The valences of sulphur, nitrogen and the argon group will also be considered separately. In closing it may be pointed out that the law just estab- lished can be used to determine the number of valences in a molecule if M2K of the substance is known, or if its critical data are known, as follows: If the computation is made from the approximate values of "a" given in the Landolt- Bornstein-Meyerhoffer tables, which were computed by the formula: a 27^7(64 X 273' X Pc), proceed by the following formula: (i) Valence number •• •• (a)3/2 X 3.2 X io5/(Mol. Wt.). Or, if the calculation is made from the critical data, the following formula will serve: 1 Possibly nitrogen is univalent in this gas, but the oxygen is quadriva- lent, having two free valences. 196 Albert P. Mathews (2) Val. number = = 0.0043 Tsc/Pcs/*(Mol. Wt.); or, if the calculation is made from the temperature and volume: (3) Val. number = 4.21 -X iQ-5 X (VcTc)8/*/(Mol. Wt.). -37.56 Fig. i —Showing the relationship between the logarithm of M2K and the logarithm of the product of the molecular weight by the number of valences In these formulas Vc is the critical volume of a gram mol. in cubic centimeters; Tc, the absolute critical temperature; and Pc the critical pressure in atmospheres. Value of "a" of van der Waals' Equation 197 How accurately the number of valences can be calculated by the first formula is shown by Table IX, which is, of course, little more than a rearrangement of the results already stated in the preceding tables, except that in this instance I have used the approximate values of "a" given in the Landolt tables computed by the formula a = 27Tc2/(64 X 273* X Pc). An inspection of Table IX will, I think, convince all that molecular cohesion, as expressed by the value "a" of van der Waals' equation, is a function of the weight and valence of the molecule. If "a" is calculated for all these substances from the critical temperature and pressure by the formula : a = = 27T2C/64PC x 2732, the number of valences is calculated, with a remarkable degree of approximation, from the as- sumption that a --= CN2Wt2/3Val2/3, by the formula given at the beginning of the table. It will be noticed that all asso- ciating substances give by this formula a larger number of valences than that calculated for the normal substance, and that the degree of excess of trie number of valences is more or less proportional to the degree of association. Thus the largest excess is in the case of water, where the cohesion calls for 22 valences, and in some of the nitriles; whereas sub- stances like the esters, which associate very little, have only a few valences in excess. Putting aside associating substances, of which the deviation from the rule is to be expected, there are certain exceptions characteristic of certain observers. The coefficient 3.2 X io5 was taken from Young's data for ether. It will be seen that always Nadejdine's critical con- stants give a value for "a" lower than Young's, where both observers have worked on the same substances. So the sub- stances computed from Nadejdine's critical data generally show a deficit of two or three valences to the molecule. I think in these cases Young's values are to be accepted. Vin- cent and Chappuis' determinations, also, generally come a little low, though they are very close, The main constant deviation from the law is to be observed in the case of sub- stances like methane and hydrogen of very low molecular weight and great molecular simplicity; and those substances 198 Albert P. Mathews o ^ ^ a o o O j_ S * « *S"b \> A &x | 3 3 O. a r % 0*0*0 o3 ^ o* S" o 60006606000000000600000 oo uaucju O 4-J <1> ijSIl ^ -^ ^j ^ aa 5 o o SSSS<^<<^SSSwSwSSSwSSS Value of "a" of van der Waals' Equation 199 o ^ ft •£ G G G v* c .2 .2.2 .25 g ^ I g '5*3 " *5 8 '-MO +1 o> c/5 c/3 G .2 ~ $ ro^O^O^O^.2 O3o3 ^^j 's » i r| all| tf> ±i ^^ ^H ^^ ^^ — , ^ rj- 10 iOvC ^uoro-^-rDrOrO^- 00 rf iO <-> 00 •-" T!- SO CO O ^ *-• -- -1- - ,J W M O ?! -M § £ s « sa_S" -^ . a ^ i -a S^a'sf | I S 1^1 3 C fe _, , °£3 p>^ o3 o3 O C g Theoretica valences 3 « ^L. c*- r>-. 00 Tt-^C< C^OO C4O -i CM t-i CO ON ONOO Th Tf ONOO I>> CM OO CM cooo oo ir> 10 t*-)\O •-" 10 Ti-l>-O CM t^OO CO CM O ONTJ-OO COCN t^VOOO COCO ^i-O cOONrOi-<00 IN. CM OO o'oooooooooooo'ooooooooooooooo 66666 6- 6 660066666666600000006 W,ffl2W K K W W.d 0 0 W»W ,, ffi ffi ffi , , o o o o o o u u oaooouooouoooouooouo -s Ji* -s- il i .-si §1 § i lliiiitil s si'&i a a ^ O -5 -8 - - ' II Value of "a" of -van der Waals' Equation 201 1 es 5 55 -8 " u 5 £ ^3 i> > <3 o < >< O O ^ a oc r^- vo O *o M :' CN oo oo r^ fi 3 M ! ON ON ON < , OO O CO O 00 00 CN O Tt-00 rh CN c^O CN r^-Tt-cs csvO CM CO ^ CN Tj- M -tf-oo GO lO ON r^ 0s. CO t-i CS CO CO CO CN CN s ON 00 O ON VO ON ^O OO CO^O^O^O O f^ O f^* O >O lO ^ ON < M ON CN t~-» i— i OO O CN CN CO OO CO O CN r^ CN i— > t~^ ^t" lO M 0 ON CO ,00000000000000000000 ooooooooo i'6666 6000666666000000 060666000 'ffiffiffiffiffiWWffiffi 03 n \n ic n m o " IK ocr-^ MMiOM9eoAAn^i =» *->vy> j, \_/ OOf^OOOOOOOOOOOOffic/} 'c7>u"crO*afaTcJb 2 • . .a a fl s 202 Albert P. Mathews studied by Guye and Mallet1 of very high molecular weight and complexity, such as durol, cymol, diphenyl and diphenyl methane. So far as I can find, the evidence is that these sub- stances associate but little; so the deviationc annot be attrib- uted to that. It may be that Guye and Mallet's determina- tions have some constant source of error which brings them higher than determinations on similar substances by Altschul, but I do not think this is the explanation of the facts. The fact that the deviation is in the opposite direction in the sim- plest from what it is in the most complex substances leads me to believe, as stated on page 192, that the calculation of "a" by this formula may be at fault. While there can be little doubt that this formula gives a value for " a " approximately correct for all normal substances, as van der Waals has recently reaffirmed, and accordingly that the ratio 27/64 is at least approximately true for all substances, yet it is not certain that this ratio is exactly con- stant for all substances. It can only be strictly justified for the simplest substances in which "6," the volume of the mole- cule, is constant. I believe it to be far more likely that this ratio changes a little with progressive increase in molecular complexity, and a resulting greater compressibility of the molecule, as indeed van der Waals has suggested, than that the law, which I have here attempted to establish, of the de- pendence of cohesion on molecular weight and valence holds strictly only for substances of medium complexity.- This re- lationship of cohesion to these two molecular properties seems so probable and so fundamental that I believe we may, with some confidence, anticipate that if it holds at all, it holds everywhere. It seems not worth while to consider this question further until it is decided whether "a" is in all instances determined by the formula just mentioned; or whether the ratio 27/64 is true only for the simplest substances with constant molec- ular volume, that is a constant " 6; " and that for more complex 1 Guye and Mallet: Comptes rendus, 133, 1288 (1901); 134, 168 (1902). Value of "a" of van der Waals' Equation 203 and compressible molecules it must be slightly diminished as the complexity and compressibility increases. If the lat- ter shall prove to be the case, then most of the exceptions noted in the preceding pages would disappear. The substitu- tion of the value for "M2K" computed by my formula from the surface tension would have given a much closer agreement of the calculated and observed valence numbers, but I wished to show that the law holds at least approximately even though we use the approximate values of "a" computed by the usual formula. The constant 2.98 X io~37 discovered in the foregoing pages, is evidently the factor M2K of a substance of unit molecular weight and unit valence. No such substance as this is known, but hydrogen with a weight and valence of two, and helium with a weight of four and valence of unity (?) have values of M^K not very different. Thus for hydrogen,1 M2K is between 3.16 and 7.72 X io~37; and M2K of helium lies between the same two values probably. The value of 2 .98 X icT37 is of the order of magnitude of the gravitational attraction of two average molecules. Thus at 20° two mole- cules of ether in the liquid state attract each other gravita- tionally with a force of 3.11 X io~37 dynes. The similarity of these values is, however, probably only a coincidence. Conclusion The facts presented in the foregoing pages enable us to draw the general conclusion: The "mass" of cohesion of a molecule is everywhere proportional to the cube root of the molecular weight multiplied by the cube root of the number of valences in the molecule. Or, to put it in another way, "a" of van der Waals' equation for one cc. of gas under standard conditions is equal to 2.98 X io~37 X Mol. Wt2/3 X Valences27' X (2.77 X io19)2 dynes; or this number divided 1 M2K for H2 computed from the value "a" given by Landolt-Bornstein 27 T 2 is 5-55 X icr37. This was computed by the formula a = 64 X 273 X rc 2O4 Albert P. Mathews by 1.0135 X io6 atmospheres. This formula gives a value for "a" somewhat higher than the ordinary formula and it may be that the coefficient should be taken a little lower. The theoretical significance of this relationship of cohe- sion to the molecular weight and the number of valences is very interesting, but I shall reserve its consideration for a subsequent paper. University of Chicago THE VALENCE OF CHLORINE AS DETERMINED FROM THE MOLECULAR COHESION OF CHLORINE COMPOUNDS BY ALBERT P. MATHEWS Since molecular cohesion is a function of the molecular weight and the number of valences in the molecule,1 we may use the cohesion for the purpose of determining the valence of elements; and in this paper I shall consider chlorine, although something will be said, also, about the other halogens. It is generally believed, at the present time, that the valence of chlorine is not fixed, but varies in different com- pounds from one to seven. In its organic, and some inor- ganic compounds, and in its elemental form it is generally represented as univalent; whereas in the chlorates it is sup- posed to be pentavalent; and in the perchlorates it is hepta- valent. That chlorine even in such compounds as chloroform, where it replaces univalent hydrogen, may not be univalent is indicated by the action of chlorine compounds on light. Drude,2 reasoning that it must be the valence electrons of compounds which would have a period of vibration sufficiently long to respond to light waves, worked out a modification of the Ketteler-Helmholtz dispersion formula which enabled an approximate computation of the number of electrons in- fluencing dispersion in the molecule. He found that in many cases this number was close to the total number of valences in the molecule; but in the case of compounds con- taining chlorine and fluorine, the number of such light-re- fracting valences was always greater than in the correspond- ing hydrogen compounds, and he inferred from this that 1 Mathews: "The Relation of the Constant "a" of van der Waals' Equa- tion to the Molecular Weight and the Number of Valences in the Molecule," Jour. Phys. Chem., 17, 181 (1913). 2 Drude: "Optische Eigenschaften und Elektronen Theorie," Annalen der Physik, [4], 14, 677 (1904). The Valence of Chlorine 253 these elements must be polyvalent, and not monovalent, as they were usually supposed to be. This conclusion of Drude's was confirmed by Pascal1 both by the dispersion method of computing valence and by a study of the diamagnetic proper- ties of halogen compounds, the diamagnetic properties having been shown to be related to the number of valences in the molecule. Pascal concluded that fluorine, in organic com- pounds at any rate, was univalent; but chlorine and the other halogens were polyvalent, and probably chlorine was tri- valent. Traube,2 in a study of the relationship between the molecular refraction of compounds and the number of their valences, found that for most compounds the molecular re- fraction of Bruhl divided by the number of valences in the molecule was a constant, or nearly such, in all saturated com- pounds; but in the case of molecules containing the. halogens it was necessary to ascribe several valences to the halogens to obtain this constant. He attributed seven valences to chlorine, and had to make still other assumptions for bromine and iodine to bring them into line. Several chemists, also, have in the past ascribed several valences to chlorine. Thus Meldola3 wrote the formula of CH3\ methylether hydrochloride, in the form >O = Cl — H, CH 3 with chlorine trivalent; Nef4 represented elemental chlorine as trivalent, but combined chlorine generally as monovalent; and recently Thiele5 has especially emphasized the reserve, or extra, valences of iodine and bromine, although, as a rule, he represents chlorine as univalent. Even in sodium chlor- ide it is not certain that the chlorine is univalent, since it is 1 Pascal: "Recherches magneto-chimiques sur la structure atomique des halogenes," Comptes rendus, 152,862 (1911); "Sur un mode de controle optique des analyses magneto-chimiques," Ibid., 152, 1852 (1911). 2 Traube: "Valency, Lichtbrechung u. volume," Ber. chem. Ges. Berlin, 40, 130 (1907). 3 Meldola : I have mislaid this reference and have not been able to find it again. 4 Nef: Liebig's Ann., 298, 205 (1897). 6 Thiele and Peter: Ber. chem. Ges, Berlin, 38, 2842 (1905). 254 Albert P. Mathews known that sodium chloride will add iodine, presumably by the extra valences of the chlorine.1 There is good ground, therefore, for doubting whether chlorine is ever monovalent. This question can be tested easily by the cohesion method. Before proceeding to the actual computations it must be decided whether the cohesion method detects only valences actually employed in binding atoms together, or stretching between the atoms; or whether it detects in addition the re- serve valences; and also valences which do not extend to atoms, but which are open, in an active form, and ready to combine if the opportunity arises. It is clear from my former paper that concealed, polarized, resting or reserved valences do not play any part in cohesion ; or, at any rate, they are not to be counted in the number of valences affecting the cohesion. Thus oxygen has certainly two reserve valences which are usually in an inactive or resting state. In many compounds examined, not more than two valences could be attributed to the oxygen as affecting its cohesion. These two reserve valences played no role as long as they were inactive. Sim- ilarly, nitrogen has the power of opening up at least seven valences, but it was actually found that only one, two or three valences played a role in the cohesion of the nitrogen compounds, depending on how many active valences the atom had. The reserve, or inactive, valences played no part. Carbon is usually quadrivalent, but it is suspected of having the power of becoming hexavalent; but the number of valences active in carbon compounds was always two, or four. If these reserve valences of carbon exist they do not affect cohesion. Sulphur, too, although it may be hexavalent, has only four of its valences playing a part in the cohesion of sulphur dioxide; the two reserve valences are inactive on the cohesion. It is clear, then, that the cohesion does not detect, and consequently it is not affected by, those reserve valences which are polarized, or resting, or, which are, as it were, like antennae, withdrawn or folded, within the atom. 1 See Friend: "The Theory of Valency," London, 1909, pp. 58 et seq. The Valence of Chlorine 255 But valences may conceivably exist in an active state not stretching between atoms, but extending outward from the atom and in a condition to unite with other atoms. These are active valences. For example, we may expect the va- lences on the atoms of a dissociated, monovalent gas to be in this condition. Such atoms would naturally be very ac- tive chemically and we should expect the cohesion of such particles to be affected by this condition. There are many evidences that this is actually the case and that valences of this kind are detected by cohesion and will be included in the number of valences computed from the cohesion as ex- isting in the molecule. This is well shown in the argon group, which I shall discuss later, in which it appears that there are two such active valences in argon, krypton, and probably xenon. It appears to be the case, too, in unoxidized sulphur compounds such as ethyl sulphide, as I shall show in a subse- quent paper. And there is evidence elsewhere that these open- or active valences affect cohesion, although they do not stretch between the atoms of the same molecule. It is, then, active vajences, and valences actually employed in binding together the atoms of the molecule, which affect molecular cohesion. It is only the number of such valences which the cohesion enables us to compute, and it is, of course, exactly for this reason that the method has so great a value. We may then be certain that if we find the valence of chlorine to be three and the compounds are not associating, those three valences are not free, but are actually extending between atoms in the molecule, and if we wish an accurate graphic formula of the compound we must represent these ties. The method of measuring the number of valences is to compute the number from "a" of van der Waals' equation, or from what I have called the square of the cohesive mass, or M2K, a factor which is equal to "a", divided by the square of the number of molecules in the volume of fluid for which "a" has been taken. The method of computing M2K is given in the previous paper. The formula employed is M2K .- 2.98 X io~37 (Mol. Wt. X Valences)273. Or: Number of Valences - (M2K)3/2 X 6.147 X io54/(Mol. Wt.). 256 Albert P. Mathews ^2 T\/1 w " Yx ? a ^_^ i _i fj ]? VO VO VO ^- o D *Q VO 'I M M M CO H 3 Iks F* O O •^- CO CO w iO t^ 1/5 B* o <* ^o *o ^t- cs CO ON gJZTg MM CO 4-i 00 00 CO >O to to £ rf CO Tj- co O ^h ^H tO VO •-( CM ON 6 O S M M M I** o x r^ *o c^ to oo vo 00 HH CN .1-1 HH 00 <4 ^ t^ ON <^- r^ oo t^ oo O CM be iO IO tO to to to o CO CO CO CO CO CO M 1 1 13 ^ ^ D^ CO ,— I H f\ ) O .D Mffi it , / W CO § t/3 . ci O -• d aj ON ON »ovo (N O VO g -P O vO 00 Tf ONVO t^ S 2 M 00 ONVO "- &<• cococococococo W 1) Td •e'S -a a) "^ O' ll 21 fl T^^-3 S»-i '~H -PO03 H .Si y L**C -28 ^ o o C = C1— H, H/ the reason why it decomposes into methylene and hydrochloric acid appears at a glance, since such a double bond is always a source of weakness; and similarly with ethyl chloride, which would be in reality ethylidene chlorhydrate, decomposing into ethylidene and hydrochloric acid. One can also more easily understand in this way the decomposition of chloroform into dichloro- methylene and hydrochloric acid, as the for- mula C1\ j | >C = C1— H CK shows. Phosgen will arise from the dichlormethylene uniting with oxygen. The evidence, then, from such various sources as the be- havior, toward light, the diamagnetic properties and cohesion is unanimous that chlorine is polyvalent and not monovalent; many of the chemical and physiological properties of chlorine compounds are also more easily understood on the hypothesis of its trivalency. We may, therefore, conclude that in all these compounds chlorine is trivalent. The question which must now be settled is no longer whether chlorine is trivalent, but whether it is ever mono- valent. It is certainly trivalent in most of these compounds in which it was supposed to be monovalent; it is trivalent even in its elemental state. It remains to be seen whether it is ever monovalent. In hydrochloric acid it would appear to be monovalent; but it is exactly here that association takes place. Is it without significance that exactly that compound associates which has but one of the valences of the chlorine The Valence of Chlorine 261 satisfied by another monovalent atom? Is it not rather more probable that this is the cause of its association, the other two valences being not closed, but out and active? The com- putation actually shows that the chlorine is here also tri- valent. I know of no means of telling whether it is mono- valent or trivalent in sodium chloride. But it is not impos- sible that sodium chloride itself is a highly associated sub- stance. Furthermore, its power of adding iodine indicates that the chlorine may be trivalent. Friend also states that sodium chloride may be Na — Cl =C1 -- Na. Concerning the valence of the other halogens, the facts are too scanty and the data too unreliable to draw a conclu- sion from the cohesion, except perhaps in the case of elemental bromine, which appears to be univalent. The critical data of brombenzene and iodobenzene were not directly deter- mined by Young, but computed from the temperature, pres- sure and density curves. I do not believe that they are en- tirely trustworthy, since the number of valences found in the molecule is too small even if these halogens are considered monovalent, unless the carbon be here trivalent, and this does not seem possible. Tc and Vc of ethyl iodide, I com- puted from Ramsay and Shields' surface-tension determina- tions, and this computation is not very accurate. Hence I do not attach much weight to the cohesional evidence of the valence of any of these compounds of bromine and iodine. There are no indications, however, that they are polyvalent. That they are polyvalent is, however, indicated from their action on light, their diamagnetic properties and many of their chemical properties. Inasmuch, however, as the re- fraction method is not very satisfactory for determining valence, the question of the valence of these substances must be left open, with the probability that they will be found to be polyvalent like chlorine.1 1 The fact that bromine is monovalent in its elemental state may account for its relative inertness and is confirmed by its dissociation at high tempera- tures, when the atoms have been shown to have but one active valence. See Friend: "The Theory of Valence," 1909, p. 18. 262 Albert P. Mathews Fluorine is apparently monovalent in fluorbenzene, since even with fluorine monovalent the number of valences computed from the cohesion is still too small. For this I can give no reason since the critical data of this substance seem to be accurately known. In methyl fluoride the total valences are computed as 9, whereas there should be 10 if fluorine is trivalent and 8 if it is monovalent. Pascal found fluorine, to be monovalent by the magnetic and optical method; but Drude, from the optical behavior of calcium fluoride, be- lieved it to be polyvalent. The critical data of more fluorine compounds must be accurately determined before the cohe- sional method can determine the valence of fluorine. The chemical behavior of hydrogen fluoride leaves no doubt that in it fluorine is polyvalent. There is still another interesting conclusion from this study : it appears that all substances, and only those substances, associate, which are found by this method to contain active, free valences. I believe we may here have the explanation of the cause of association; and possibly the reason why as- sociating substances dissolve in other associating liquids and are there normal, but as this is a separate problem in itself, I shall hope to return to it later. Summary and Conclusion 1. If the valence of chlorine be determined by the co- hesional method it is found to be trivalent in its elemental state and in nearly all the compounds examined. The three valences of the chlorine in these compounds are not reserve valences, but are all in action and extending between the atoms of the molecule. Graphic formulae have been suggested based on this fact. 2. This result is in harmony with the determination of the valence of chlorine by the diamagnetic and refraction method. 3. The valence of fluorine is more doubtful, but appears to be unity in fluorbenzene. Bromine has unity valence in ts elemental form. The valence of iodine and bromine in The Valence of Chlorine 263 their compounds cannot be definitely determined from their cohesion on account of the inadequacy of the critical data. 4. The cohesional method detects two kinds of valences, namely, valences actually extending between the atoms and active in binding the atoms together; and valences active or open, which are in a position to unite, but to which no atoms are attached. Reserved, or resting, valences play no part as valences in molecular cohesion. University of Chicago THE VALENCE OF OXYGEN, SULPHUR, NITROGEN AND PHOSPHORUS DETERMINED FROM THE MOLECULAR COHESION BY ALBERT P. MATHEWS i. The valence of oxygen In my paper on the relation of molecular cohesion to molecular weight and valence,1 in which the number of valences were computed from the value "a" of van der Waals' equa- tion, I considered oxygen to be bivalent except in the case of oxygen gas. The question of the quadri valence of oxygen was not taken up, because I wished to get the main point of the connection of valence and cohesion well established be- fore discussing particulars. But as it is believed by many that oxygen is at times quadrivalent and as this possibility has such an important bearing on theories of association and solubility and, indeed, on the ionizing powers of water, and is also full of importance in physiology, it is interesting to examine the oxygen compounds of that table for evidence in favor of tetravalence. One atom of oxygen is supposed by Stieglitz, Arm- strong, and Goldschmidt2 to be quadrivalent in the esters. In the first table, therefore, I have incorporated the results of a determination from the molecular cohesion by the usual formula :n == a3/2 X 3.2 X io5/(mol. wt.), of the total number of valences per molecule, and compared it with the number of valences computed if one oxygen is quadrivalent. I have taken only those esters whose critical data were recently very carefully determined by Young. All of these esters associate a little at low temperatures, but their vapors are normal for some degrees below the critical temperature, except possibly that of methyl formate. 1 Mathews: Jour. Phys. Chemv 17, 181 (1913). 2 Zeit. Elektrochemie, 10, 221 (1904). 332 Albert P. Mathews TABLE i Theoretical •KJ £ Number of Substance Formula "a" JNO. OI valences. valences det. from One oxygen quadrivalent cohesion Ethyl acetate C4H802 o . 04076 30 29.9 Ethyl formate C3H602 0.03122 24 23-9 Ethyl propionate C5H1002 0.05088 36 36.0 Methyl acetate C3H602 0.03206 24 24.8 Methyl butyrate C5H10O2 0.05082 36 35-9 Methyl formate C2H402 0.02371 18 19-5 Methyl isobutyrate C5H1002 0.04882 36 33-8 Methyl propionate C4H802 0.03968 30 28.8 Propyl acetate C5H1002 0.05149 36 36.7 Propyl formate C4H802 o . 04086 30 29.7 From Table i it is clear that the number of valences computed from the molecular cohesion is very close to the theoretical number, if one of the oxygen atoms is quadri- valent. The only cases in which fewer valences were found were methyl isobutyrate and methyl propionate. In the former, the computation of "a" may not be exactly right as elsewhere pointed out, the isocompounds always giving a value for "a" a little low as compared with the normal. It is, of course, possible that in them the oxygen is bivalent. Table 2 contains some other oxygen compounds which associate and may be supposed on that account to have quadrivalent oxygen. In Table 2 the alcohols probably associate somewhat at the critical temperature and acetic acid also; the number of valences, given in the table as computed, were computed with the normal molecular weight and they are, accordingly, too many. It is impossible to determine the total number of valences per molecule by this method for these substances until the average molecular weight has been independently determined at the critical temperature. Of the other sub- stances: acetone, anisol, phenetol, nitrobenzene, acetic anhy- dride, ethyl diacetate and methyl propyl ketone appear to have tetravalent oxygen. The method, however, is not The Valence of Oxygen, Etc. 333 TABLE 2 Number of Substance Formula "a" valences computed ; No. of valences found one oxygen quadrivalent Acetone C3H60 0.02459 22 21.3 Ethyl alcohol C2H60 0.02395 18 25.8 (Assoc.) Anisol C7H80 0.05645 40 39-8 Methyl alcohol CH4O 0.01895 12 26.0 (Assoc.) Phenetol C8H10O 0.07009 46 48.7 Propyl alcohol Acetic acid C3H80 C2H402 0.03250 24 0.03504 I 8 31.3 (Assoc.) 34.9 (Assoc.) Nitrobenzene C6H5N02 0.05968 38 37-9 Acetic anhydride C4H603 o . 04442 30 29-3 Ethyl diacetate C6H1003 0.06668* 42 42.4 Methyl propyl ketonei C5H10O 0-04437 34 34-8 m-Cresol ! C7H8O 0.06254 4° 46.0 (Assoc.) sufficiently accurate to enable this conclusion to be drawn without reservation. The critical data of some of these sub- stances have not been determined with entire accuracy and it is possible, in some, that a very small amount of association may occur at the critical temperature and such association would have the effect of making uncertain the valence de- termination. With these reservations, however, this method undoubtedly supports the view that oxygen may be tetra- valent, and particularly that one atom is tetravalent in the esters. The method shows oxygen to be bivalent in ether, sulfur dioxide, carbon dioxide and one of the oxygen atoms of the esters. In oxygen gas, carbon monoxide and nitric oxide, NO, the oxygen is monovalent, if computed from the cohesion. The maximum number of valences in a molecule of oxygen found by this method was two. It must be confessed that it seems unlikely that oxygen is monovalent in the gaseous form, but the critical data are accurately determined and if this method of determining valence is reliable, as it appears to be, there is no escaping the conclusion. The critical data of carbon monoxide should be redetermined, but from those 334 Albert P. Mathews accepted only two valences can be found in the molecule. Nitric oxide has always been a puzzle, since the oxygen is monovalent, or the nitrogen bivalent. The cohesion shows that there are only two valences, which again means that the oxygen and nitrogen are monovalent.1 From its cohesion, then, oxygen appears to be either monovalent, bivalent, or tetravalent. 2. Sulfur Sulfur is generally supposed to be bivalent, but at times to open up two or four residual valences. So far I have not found any compounds with bivalent sulfur, except probably carbonyl sulfide, when the valence is computed from the cohesion. All the sulfur compounds have either quadrivalent or hexavalent sulfur, if the valence is determined by this method. Even sulfuretted hydrogen is no exception to this statement. Table 3 contains the results. TABLE 3 — THE VALENCE OF SUBSTANCES CONTAINING SULFUR Valences permol. Valences per Substance Formula. " n " S = 6 S = 4 cohesion Hydrogen sulfide H2S o . 00890 8 6 7.9 Mercaptane C2H6S 0.02497 20 18 20.4 Thiophene C4H4S 6.04130 26 24 32 .o(Assoc.) Sulfur dioxide S02 0.01349 10 8 7.8 Carbon bisulfide CS2 0.02316 16 12 14.8 Sulfuryl chloride S02C12 0.04382 1 6 14 14.1 Thionyl chloride SOC12 0.03110 14 12 14.8 1 A very interesting fact which may be cited in support of the view that oxygen in these three gases is in a state different from the ordinary, and, hence, possibly monovalent, is the following: Oxygen gas and nitric oxide are strongly paramagnetic, and carbon monoxide is far more paramagnetic, or rather far less diamagnetic, than carbon dioxide which contains more oxygen. In all other compounds oxygen is diamagnetic. It is clear that the oxygen in the three gases which this method shows to contain monovalent oxygen has magnetic properties different from oxygen in other oxygen compounds. This fact has not hitherto been explicable. The following figures showing the para- or diamag- netism of different gases I have taken from Auerbach's article on Magnetism in the Handbuch der Physik, 5, 274 (1908). O2 NO CO Air C2H4 CH4 CO2 N2O N2 H2 + 4.83; +1.60; -0.009; *> -0.068; -0.063; ~°-°33! -0.018; -0.015; -o.oo2(?) The Valence of Oxygen, Etc. 335 From Table 2 it appears that sulfur is quadrivalent in sulfur dioxide and sulfuryl chloride; hexavalent in sulfuretted hydrogen, carbon bisulphide, mercaptane, thionyl chloride and probably in thiophene. It is probably bivalent in car- bonyl sulphide, but the critical data are uncertain. These results are in agreement with the general idea of the valence of sulfur except in the case of sulfuretted hydrogen, mer- captane, and carbon bisulfide. Thiophene probably associates a little at the critical temperature so that the valence com- putation is uncertain. j. Nitrogen Nitrogen is generally supposed to be either mono-, tri- or pentavalent. The cohesion method supports this con- clusion. Nitric oxide has been a stumbling block. The re- sults are given in Table 4. TABLE 4 — NUMBER OF VALENCES PER MOLECULE OF NITROGEN COMPOUNDS Substance Formula "a" Number of valences found from cohesion Number computed N = i N=3 N = 5 Nitrogen N2 0.0032808 2 .0 2 6 10 Nitrous oxide N2O 0.009465 6.2 4 8 Nitric oxide NO 0.002570 1.4 2 4 Nitric dioxide N02 O.OIIIQ 7.6 7 9 Ammonia NH3 o . 00844 13-5 6 8 (Assoc.) Methyl amine CH5N 0.01521 17.9 12 14 (Assoc.) Dimethyl amine CyHjN 0.01922 17-5 18 20 Trimethyl amine C3H9N 0.02594 22 . 7 24 26 Diethyl amine C4HUN 0.03625 27.9 30 Triethyl amine C6H15N 0.05415 39 9 42 Propyl amine C3H9N 0.02729 24-5 24 Dipropyl amine C6H15N 0.05835 5i-2 52 Aniline C6H7N 0.05157 37-2 34 36 Dimethyl-0-tolui- dine C9H13N 0.08187 5i-2 52 54 From the table it appears that nitrogen in nitrogen gas is monovalent. It appears to be also monovalent in nitrous and nitric oxide. Two formulas may be written for nitrous \/ oxide with a total valence of six, i. e., N — O — N or N — N =O. 336 Albert P. Mathews I think the former, with some of the molecule having quadri- valent oxygen, is perhaps the more probable. I have taken the highest possible value of "a" for nitrous oxide. In nitric oxide two valences per molecule are the most that can be assigned by this method. This would mean that both elements were monovalent, but as there is other evidence that oxygen is univalent in its elemental form, this formula is not entirely improbable. The valence of nitrogen in nitric dioxide is quite uncertain. In the amines there can be hardly a doubt that it is trivalent, or at times pentavalent with two free valences on the nitrogen. This is the case in ammonia and methyl amine, both of which associate. All of the critical data of the amines were determined by Vincent and Chappuis and I believe their figures are uniformly a little too low. In aniline the nitrogen appears to be pentavalent, but some association occurs. The results are not, then, very satisfactory for these compounds, but they indicate very clearly, however, that nitrogen is either monovalent, tri- valent, or pentavalent, as it is supposed to be. 4. Phosphorus I have found but a single phosphorus compound in which the critical data have been directly determined, although one or two other cases were reported in my paper on the valence of chlorine, in which the critical data had been com- puted from the surface tension. In phosphoretted hydrogen the phosphorus is apparently pentavalent, having two free valences, "a" is given by Leduc and Sacerdote as 0.00939 and from this the valence of 8.6 is computed. Were the phosphorus pentavalent the total number of valences would be eight. It might be, however, that the phosphorus was heptavalent, but only a few of the molecules had the two pairs of reserve valences open at the same instant. In the chlorine compounds it ap- peared that the phosphorus was probably heptavalent. I think the only certain conclusion is that the valence is greater than three. University of Chicago THE VALENCE OF THE ARGON GROUP AS DETER- MINED FROM THE MOLECULAR COHESION BY ALBERT P. MATHEWS The valence of the argon group of elements is one of the most interesting problems in chemistry. They are very generally regarded as zero valent, chiefly owing to the posi- tion they take in the periodic system between strongly electro- positive and electro-negative, univalent elements. That they are monatomic is undoubted, but they might be monatomic, like mercury vapor, and still have valence. Ramsay1 made the suggestion, indeed, that they combine into molecules at other than ordinary temperatures. To account for the atomic weight of argon, which computed from the density is 39.9 if the gas is monatomic, he suggested that argon is a mixture of many monatomic molecules with a few diatomic molecules. The ratios of the specific heats as determined is 1.659; whereas if there were 5 percent of diatomic molecules it would be 1.648. The theoretical number, if the gas is entirely monatomic, is 1.667. After discussing this possi- bility, however, Ramsay says: ''But on the whole the pre- sumption is against the hypothesis that argon is a mixture of monatomic and diatomic molecules." There is some evidence that argon is not entirely lacking in chemical affinity. Berthelot,2 by the action of the electric discharge on a mixture of argon and benzene vapor, or of argon and carbon bisulphide, produced a brownish deposit on the glass from which argon could be reobtained. Ramsay,3 in commenting on the absence of the argon lines in the sun's spectrum, suggests, as a reason, that it enters into combina- tion only at high temperatures, these compounds being endothermic; and he cites4 several observations indicating 1 Ramsay: "Gases of the Atmosphere," London, p. 231 (1902). - Berthelot: Comptes rendus, 120, 581, 660, 1316 (1895); 124, 113 (1897). 3 Ramsay: Loc. cit., p. 261. 4 Ramsay: See footnote, p. 538, to article by C. Trenton Cooke: Zeit. phys. Chem., 55, 537 (1906). 338 Albert P. Mathews a union of argon with zinc, mercury and some other elements. Thus in a Pliicker tube the cathode metal disintegrates more rapidly when argon under low pressure is in the tube than when nitrogen is there, and Ramsay interprets this to mean that a volatile compound is formed under the influence of the intense energy at the surface of the electrode and this com- pound dissociates again setting free the metal, which deposits on the glass. Under his direction C. Trenton Cooke1 measured the vapor tension of zinc, cadmium, sulfur, mercury and some other metals at high temperatures in the presence of various gases and concluded that the tension of zinc in argon was 12 percent above its tension in nitrogen. Cadmium behaved similarly in helium. Helium seems to be in some kind of a union in fergusonite, and to be capable of feebly uniting with platinum. It may be recalled, also, that the solubility of argon in water is greater than that of helium and nitrogen; and this may be urged as indicating some kind of affinity between water and argon. Chemically, then, these gases, though inert, are not entirely indifferent. From their action on light,2 also, a certain argument may be made for their possessing valence. Thus, according to Drude, the dispersion of light in the blue end of the spectrum is due to the valence electrons, and in the red end to the vibrations of the electrically charged atomic groups. If only the valence electrons affect blue light, these gases must also have valence electrons, since they refract light like other gases. The question whether these elements have valence, or not can be put to a decisive test through their molecular cohesions.3 There is no question that they possess cohesion, since they can all be liquefied. They behave like all other gases in this respect. Molecular cohesion in all other sub- stances examined is a function of the product of the molecular weight by the number of valences. It has been shown for a 1 C. Trenton Cooke: Loc. cit., p. 537. 2 Cuthbertson, C.: "Refractive Indices of the Elements," Phil. Trans., 204, 323 (1904). 3 Mathews: Jour. Phys. Chem., 17, 181 (1913). The Valence of the Argon Group, Etc. 339 great number of substances that M2K, a factor proportional to "a" of van der Waals' equation, is equal, when expressed in dynes, to 2.98 X io~37 (mol. wt. X No. of valences)2/3. A substance having cohesion cannot, therefore, have zero valence. If it had no valence it could have only gravita- tional attraction between its monatomic molecules, no cohe- sional attraction. These gases have cohesion and they must, therefore, have valence. Their valence, n, can be calculated from their critical data by the formula: (i) n = 0.0043 TC3/PC3/2 (mol. wt.); or, n '= a3/2 X 3.2 X io5/(mol. wt.); or (2) n ••-- 4.21 X io~5 (VcTc)3/2/(rnol. wt.). The first formula, which is derived from the ordinary formula for computing "a" from the critical temperature and pressure, gives values for "a, " and hence for n, a little lower than the second formula in the case of simple gases. The second formula computes "a" from the surface tension, as shown in my former paper. In Table i, I have given both values and I regard the second as the more correct; but as the critical density of not all the gases is known, I have had to rely on the first formula for a comparison. The results given by the two formulas are not widely different. The valence, n, together with the critical data1 used in the calculation are given in Table i. TABLE i — COMPUTATION OF THE AVERAGE NUMBER OF VALENCES PER MOLECULE FROM THE MOLECULAR COHESION Substances Tc (Abs.) PC dc Number of valences by for- Number of valences by formula 2 > mula i Helium 5-5° 2.75 0.065 0.04 0.07 Helium 8.0 2.75 0.065 0. 12 O. 12 Neon 61.1 29 — 0.32 — Argon 150-56 48 0.509 I . 12 i-35 Krypton 210.5.3 54-3 — 1.23 — Xenon 289.6 58.2 I-I55 i-8° i-95 1 In my preliminary paper, Science, N. S., 36, 6 (1912), in Table I a mistake occurred in the computation of helium. The value 2.90 is wrong. The critical density was taken as 0.015 instead of 0.065. 340 Albert P. Mathews Before discussing these results a word may be said about the reliability of the critical data. Those of argon and xenon are perhaps the most certain; krypton, neon and helium follow in the order named, helium being least certain. Onnes1 gives 5.5° absolute as the critical temperature of helium, but as this makes helium quite aberrant in several particulars,2 I have also computed the valence assuming the critical tem- perature to be 8° as suggested by Dewar.3 The critical data of neon are somewhat uncertain, due in part to the very low critical temperature and in part to the great difficulty in separating the gas completely from helium. A little im- purity of the latter gas would have the effect of making Pc too high. It is clear, from the table that all of these gases possess valence. They are not zero valent as they are supposed by many to be. Furthermore, the average number of valences per molecule is in no case an exact integer, although in argon and xenon it is not far from a whole number. Since these gases have their critical data most accurately determined I at first supposed, as I published in my preliminary paper, that argon was univalent, but slightly associated into di- atomic molecules, thus bringing the average number of valences per molecule a little high; and that xenon was di- valent. I attributed the deviation of the other gases from uni valency, to the inaccuracy of the data. A careful ex- amination of all the facts, however, has led me to abandon this explanation for what seems to me to be a better one, since it explains all the facts. In the first place, I have not been willing to abandon the idea that valences are indivisible. If we assume, as Lodge4 suggests, that some of the lines of force from each valence attach themselves to several atoms, or even wander outside 1 Onnes: Proc. Amsterdam Acad. Sci., 13, noo (1-911). 2 Rankin: "On a Relation between Viscosity and Atomic Weight of Inert Gases," Phil. Mag., [6] 21, 45 (1911). 3 Dewar: Article "Liquid Gases," Encyclopaedia Britannica, 16, 749, nthed. 4 Lodge: Nature, 70, 176 (1904). The Valence of the Argon Group, Etc. 341 the molecule, and thus split the valence up; or that there are partial valences in the sense of Kauffmann,1 it seems to me that we might as well abandon the whole valence idea. It entirely loses its usefulness. Again our structural formulas become so indefinite, if the valences are regarded as split up, as to be nearly useless. The explanation which I sought was one which would explain why we appear to have frac- tional valences in this case, but really do not have them. That my first explanation was wrong, was indicated by the uniformity with which the valence increases from helium to xenon. The deviation, too, from uni valency in the case of neon is so great that it lies outside the limits of error. The critical pressure would need to be 14 atmospheres instead of that recorded of 29 atmospheres, in order that neon should have one valence to each molecule. The uncertainty of the critical pressure is far less than this. I believe the reason that the average number of valences is a fraction in these gases is as follows: All of them are in reality zero valent, so fas ar their chief valences are concerned, but like many, if not all, other elements, they have the power of opening up two residual valences. By their residual valences, therefore, they are all bivalent. These two valences, like most, or all, other residual valences, are of opposite electrical sign, one being positive, the other negative. Not all the atoms have these residual valences open at the same instant, but always some of the atoms have them closed. The molecular cohesion, as I have already pointed out, is not influenced by these valences when they are in a closed or reserve state, or, we might say, withdrawn within the atom; it is only affected by the valences actually extending between atoms, or open and in a state in which they may combine. In xenon nearly all the atoms, or at any rate, 90 percent of them, have the valences open, and the average valence per atom, or molecule, is, therefore, 1.80-1.95; in krypton about 65 percent of the atom have open valences, and 35 percent are closed, so that the average valence is about 1.30; in argon, 1 Kauffmann: Ber. chem. Ges. Berlin, 41, 4404 (1908). 342 Albert P. Mathews about 60 percent are open and 40 percent are closed, the average valence being about 1.20; in neon, 16 percent are open and 84 percent are closed, the average valence being about 0.32; and in helium only about 5 percent of the valences are open, 95 percent being closed, giving an average valence of o.io. This explains, then, why the elements appear zero valent in the periodic table, since they are zero valent as far as their chief valences are concerned; why, nevertheless, they appear to have some weak chemical affinity, and cohesion; why they refract and disperse light; and also why the average valence is fractional rather than being a whole number. It also explains more than this. It enables us to under- stand the easy dissociation of the molecules into atoms. Unlike atoms that are bound together into molecules by their chief valences, no electrical stresses are set up in the argon elements when dissociation into atoms occurs, because each atom having a positive and a negative valence becomes at once electrically neutral. It is well known that compounds formed from residual valences partake of the nature of molec- ular compounds and break up very easily. Neither their union, nor their dissociation, involves much, if any, energy exchange. Such compounds are often called, indeed, molec- ular compounds. A double bond of this kind is always a weak bond in any molecule, which easily breaks where the double bond is. Were they univalent their dissociation into atoms would be very hard to understand. I suppose we may picture the opening of these residual valences in the manner suggested by Sir J. J. Thomson, as being due to a rearrangement of the electrons within the atom so that an excess of negative electricity is temporarily produced in one spot, and of positive at another spot on the surface of the atom. These excesses are the valences. In closing, it is not without interest to compare the valence numbers computed above from the cohesion, with the refractivity as determined by Cuthbertson.1 Their re- fractivities are in the proportion i, 2, 8, 12 and 19. 1 Cuthbertson, C.: Phil. Trans., 204, 323 (1905). The Valence of the Argon Group, Etc. 343 Refract! vity (/* — i)io6 Ratio Valence Ratio Helium Neon Argon Krypton Xenon 36-3 68.7 284 425 689 I 1.9 7-82 ii. 7 18.98 0. I 0.32 I . 12 I. 80 I 3-2 II. 2 I2.3 18.0 There is a general similarity, but not an identity. The principal facts and conclusions of the paper are: The molecular cohesion, confirming other properties, shows that the argon group of elements have valence. A com- putation of the average number of valences per molecule from the molecular cohesion gave the following results: He, o.i; Ne, 0.32; Ar, 1.12; Kr, 1.23; Xe, 1.80. The valences are apparently fractional, and not whole numbers. The conclusion is that these elements are all zero valent, as far as their chief valences go, but each is divalent in its residual valences. At any one instant of time only a certain pro- portion of the atoms, varying in the different gases, have their residual valences open, consequently the average number of valences actually open, or active, per molecule is less than two. One residual valence is positive, the other negative; and hence the combining power of the atoms is very weak, since on dissociation an electrically neutral atom is formed, by the saturation within the atom of the oppositely charged valences. This explanation enables us to understand, also, why there is a progressive increase in solubility of these gases in water from helium to xenon if solution be a process involving the union of solvent and solute through their residual valences. University of Chicago A NOTE ON THE STRUCTURE OF ACETYLENE BY ALBERT P. MATHEWS From a study of the volume of liquid acetylene, Macintosh1 concluded that acetylene was in reality acetylidene since one of the carbon atoms seemed to have the volume of bivalent carbon. He supported this contention also by various chem- ical arguments. On the other hand, Nef,2 while showing that the halogen substitution products were in reality acetyl- idene compounds, believed that acetylene was acetylene and not acetylidene, because it was chemically and physiolog- ically so inert. Nef's pupil, Lawrie,3 confirmed the acetylidene nature of the bromine and iodine substitution products. Although it is improbable, for the reasons stated by Nef, that acetylene is acetylidene, the matter may be definitely settled by my method of4 determining the number of valences in the molecule from the molecular cohesion. If it is acetylene there should be ten valences; if acetylidene, there should be eight, since acetylene does not associate and one carbon atom would be bivalent. The most recent determination of the critical data of acetylene by Cardoso and Baume gives Tc, 35.5° C; and Pc, 61.6 atmospheres. From these figures the value of "a" of van der Waals' equation calculated by the formula: a = 27Tc2/64 X 2732 X Pc, is 0.008745. Computing the number of valences, n from "a" by the formula: n = a3/2 X 3.2 X 1 Macintosh: "The Physical Properties of Liquid and Solid Acetylene," Jour. Phys. Chem., n, 315 (1907). 2 Nef: "Ueber das zweiwertige Kohlenstoffatom," Liebig's Ann., 298, 332 (1897)- 3 Lawrie: "Constitution of Acetylidene Compounds," Am. Chem. Jour., 36, 487-510 (1906). 4 Mathews: "The Relation of the Value 'a of van der Waals' Equation to Molecular Weight and the Number of Valences of the Molecule," Jour. Phys. Chem., 17, 181 (1913)- A Note on the Structure of Acetylene 321 io5/(Mol. Wt), we obtain the value 10.06 for n. There are, therefore, ten valences in the molecule of acetylene; accordingly each carbon has four, each hydrogen one. Acetylene is, therefore, acetylene, as it is ordinarily written, and not acetylidene. University of Chicago DO MOLECULES ATTRACT COHESIVELY INVERSELY AS THE SQUARE OF THE DISTANCE? BY ALBERT p. MATHEWS In a very interesting and valuable recent paper in this journal by Mills,1 the conclusion was drawn that the cohesional attraction of molecules varied inversely as the square of the distance. Besides this conclusion, which was founded on the interesting discovery that the internal latent heat of vaporiza- tion divided by the difference of the cube roots of the densities of the liquid and vapor was a constant, a most valuable part of the paper was the reopening of the question whether the field of molecular attraction is delimited by the surrounding molecules, or whether it owes its small size to the very rapid decrease of the attraction with the distance. This question raised a century ago by Laplace,2 was answered by him in the latter sense without any convincing reason for his conclusion, and has not been reopened since, the opinion being almost universal that the shortness of the radius of action is due to the attraction diminishing with the distance at a rate far more rapid than the square; the fourth, fifth, seventh and even higher powers having been suggested. In thus reopening the question Mills has rendered a valuable service. That his conclusion is correct, that the molecular field is delimited by the surrounding molecules, is clearly indicated by Einstein's3 calculation of the radius of action, showing that the radius is proportional to the distance be- tween the molecular centers. The conclusion that the at- traction is inversely as the square of the distance, however, I believe to be erroneous for the reasons which will be pre- sented in this paper. 1 Mills: Jour. Phys. Chem., 15, 417 (1911). a Laplace: "Traite" de me"canique celeste," Supp. an Livre 10, p. 351. 3 Einstein: Drude's Ann., 34, 165 (1911). Do Molecules Attract, Etc. 521 Mills1 discovered the empirical relationship that the quotient of the internal latent heat of vaporization divided by the difference of the cube roots of the densities of the liquid and vapor was a constant, except in the neighborhood of the critical temperature. If I/ is the total latent heat, and E is the part of it used in doing external work, then IY - - E would be the part of the heat used in doing internal work. He found that (L - - E)/(^/3 - - D1/3) = /*' '. He assumed that this internal heat was all used in overcoming molecular cohesion, and he ascribed the fall of the^constant near the critical temperature to the inaccuracy of^the data. He then reasoned that since for u//3 — D^/3 the expression i/V|/3 -- i/V^/3 might be substituted, the molecules must attract each other inversely as the square of the distance; since it is only on such a supposition that the difference in potential energy of the molecules in the liquid and the vapor can be given by an expression of this kind. The similarity of his expression: I, -- E -= /(i/V;/3 -- i/V^/3) to Helm- hoi tz's formula for the heat given off by the contraction of the sun seemed significant, the Helmholtz formula being W == 3/5M2K2(i/R-- i/CR). From this similarity Mills reasoned that molecular attraction, like gravitational, must follow the inverse square law. Since it is impossible that molecules should attract each other cohesively according to this law, if the cohesional attraction penetrated matter, he concluded that Cohesion did not penetrate matter, but was delimited by the surrounding molecules. There is, however, another relationship expressing the latent heat consumed in overcoming molecular cohesion or the internal pressure, which has been given by van der Waals and is derived from his expression for cohesive pressure of a/V2. This relationship is: Iv — E = a(i/V, — i/VJ,2 where 1 Mills: Jour. Phys. Chem., 15, 417 (1911); Comptes rendus, 153, 193 (1911); Phil. Mag., [6] 22, 84 (191;!); [6] 23, 484 (1912). 2 This formula should, in my opinion, be written: L -— E — X = a (i/Vj- i/Vy) where X represents heat used in any other internal work than the separa- tion of the molecules. See van der Waals: "Condensation of Gases," Encyclo. Britannica, xi edition. Also .Sutherland: Phil. Mag., [5] 22, 83 (1886). 522 Albert P. Mathews g 6 X *£T M « CO CO i^i "o a > 8 ffl o X ^ M CO O M Th ON O GO Tj- ON\O M (N • rt O ^O iO Tj- rO CO.TH CO CO CO CO CO coO 1 w 13 ON ^O OO GO 1^* CM ^j IO M O *O M \O ' J3 CM CM CN M O OO . ",U CO CO CO CO CO CM O H ^ 2 £ lu | 0 8Q O O r^ ON O O CM *t* ^t* "V *O 00 0 O O O O O co CO CM O 00 ON ON ON 3|& 1 to *b lIPARISON -E)/(V« 6 X rf CO 1 CO 0 >J 1 |1 a) M 1-4 O M 3 ^ o ON iO COVO a 1 C/l 00 TTGO >0 CO 0 • O cs *o LO ^ DCO 00 O ^"00 O (N 00 00 M M M ri CN C4 vN I M 00 CO ,r! o M M CN| o ... ^^ O QO O^ O^ O^ O^ 524 Albert P. Mathews "a" is van der Waals' constant. According to Sutherland1 the latter expression indicates that the attraction of the mole- cules is inversely as the fourth power; whereas Mills has interpreted the former as meaning that it is inversely as the square. To show how constant Mills' constant is, I have given, in Table i, the results of the calculations by his formula of a number of substances from Young's data. The figures re- present ergs for gram molecular quantities. It will be seen that the constancy is good for a considerable range of tem- perature, but that in all cases there is a more or less pro- nounced drop close to the critical temperature, and in some, as in ether and ethyl acetate, there is a pretty steady fall in the constant throughout. The fall near the critical tem- perature might be ascribed to errors of observation, or calcula- tion. There is no doubt, therefore, that for most substances the expression (L -- E)/(3v/^1 - 8VDW) closely approximates a constant except near the critical temperature, as Mills has pointed out. The conclusion that this relationship shows that the molecules attract inversely as the square of the distance is, I believe, sound, if the premise is correct. The premise, or assumption, is that the internal latent heat of vaporization, or Iy — E, represents only the work done in separating the molecules against their molecular cohesion. While Mills1 states in a recent paper that not all the internal heat may be used in doing this work, and attempts to show that this is not incompatible with this conclusion, the conclusion never- theless depends on the assumption that it is so used and that there is no change in the internal energy of the molecules on passing from the liquid to the vapor. It is clear that if this premise be not true, then the conclusion does not follow. This premise I believe to be certainly erroneous. It could only be true if the molecules remained of the same 1 Mills: Phil. Mag., [6] 22, 97 (1911); 23, 499 (1912). Do Molecules Attract, Etc. 525 size in liquid and vapor, or do not in other ways gain energy.1 I believe the internal latent heat of vaporization consists of at least three parts, not two as is often stated, these three parts are : (i) the heat consumed in expanding against external pressure, or E; (2), the heat consumed in overcoming molec- ular cohesion, or A; (3), heat consumed in increasing internal molecular potential energy by expanding the molecule, or increasing its energy of rotation, or I. This last factor is often overlooked. If L is the total latent heat of vaporiza- tion the expression should be : L — E — I == A. And if mole- cules attract inversely as the square of the distance we should have (I, — E — I)/(V^— VDJ - /. There are two principal reasons why it cannot be assumed that all the internal latent heat of vaporization goes to in- creasing the potential energy by separating the molecules against the force of their molecular cohesions. The first of these reasons is that the value "b," of van der Waals' equa- tion, has to be taken larger in the vapor than in the liquid for some distance below the critical point. And there are good reasons for thinking that "6" represents the real volume of the molecules. The second reason is the value a/V2 re- presenting cohesion in van der Waals' equation. A third reason has been given by Tyrer. To show that the molecules actually do expand in passing from the liquid to the vapor, I have calculated the value of "6" for pentane, and benzene using Young's data. I have also calculated several others of his substances, but as the result is similar in them to that in these two, I give only the latter in Table 2, which shows the value of b in cc in the liquid and vapor for gram molecular quantities b V - RT/- (P + a/V2).2 1 A similar objection to Mills' conclusion has been raised by Professor Tyrer: Phil. Mag., [6] 23, 112 (1912). 2 Since sending this paper to the publisher I have found that the value of "a" is larger than assumed here. This change requires be and b-v both to be larger. At 150° bv should be about 127 cc and at 40° only — 14. The relation between be and bv is not greatly changed, but the difference between them becomes greater. 526 Albert P. Mathews TABLE 2 Pentane t *, bv Vv P(At) a/MJ 40° 100.6 —140 21,420 I-I5 0.04 100 106.5 - 64. i 4,428.OO 5.80 I .09 150 114.6 79-9 1,512.60 15-54 8.72 180 124.2 140.3 790.05 25.46 31.96 190 130.8 I5I-9 567-36 29.61 6 1 .90 195 137.2 155-2 447-5 3I-9I 99-55 197 H4-5 153-3 359-iQ 32.90 I54-70 197.2 149.4 149.4 309-94 33-03 2°7-5o (critical) Benzene t *, M vv 80° 82.3 70 28,570 1 20 84.6 -170 10,160 1 60 87.4 22 5,428 2OO 90.8 52 2,200 240 96.2 89 1,093 260 100.5 II0.3 751-7 270 103.6 120 606 280 108.2 127 470 288.5 123.8 123.8 256.2 Table 2 shows that, in pentane, bv is larger than b^ from the critical temperature to about 160°. Below this point vv falls rapidly and apparently soon becomes negative. The reason for this apparent fall is undoubtedly the association, or quasi- association, occurring in the vapor as the temperature falls, as van der Waals suggests, the result being that the number of the molecules in the space does not remain con- stant and hence R does not remain constant. The effect of reducing R to its real value, were we able to correct for the association, would be to make bv larger. In benzene bv falls below bl sooner than in pentane, from which we may infer that the association in benzene is a little larger than in pentane. The apparently negative value of bv is found closer to the critical temperature in the esters which are known to associate slightly. Since association produces an Do Molecules Attract, Etc. 527 apparent decrease in bv, it is practically certain that the differences between b^ and bv are actually larger than those indicated. The main fact is then established that bv is actually larger than bl for some degrees below the critical point in spite of the association which tends to mask the actual molecular ex- pansion and which, at lower temperatures, conceals it en- tirely. It will be seen, also, that, as one would expect, bv actually decreases close to the critical temperature, owing to the com- pression of the molecules due to the great increase in internal and external pressure. This increase of pressure (P + a/V2) is indicated in Table 2 in the case of pentane, the pressures being given in atmospheres per sq. cm. It is of interest, in this connection, to compute what is possibly the real value of bv, making van der Waals' assumption that in the vapor, at low temperatures, bv is equal to 2&r Supposing that this is the case at absolute zero we may write the rectilinear diameter formula : 6t + b^ = bc ((3TC + T)/2TC). This assumes that at absolute zero the molecules are so com- pressed that in the solid their volume is one-half of what it would be in the vapor at the same temperature, and that the volume of the vapor molecules at absolute zero is that of bc. + T)/2TC))-&1. bc is very nearly Vc/2 . Table 3 bv TABLE 3 Pentane t 6, bv (cal) bv (taken from Table 2) 40° 100.6 173.2 —140 100 106.5 176.9 -7—64.1 150 114.6 176.7 79-9 1 80 124.3 171.9 140.3 190 130.8 166.8 151-9 195 137.2 161.3 155-2 197 H4-5 154-3 153-3 197.2 149.4 149.4 149.4 The parabola represented by these figures is given in Fig. i. If there is any association in the liquid the effect 528 Albert P. Mathews of. correcting for it would be to reduce the value of bv and increase bl and so to make the parabola flatter. I have computed bl assuming that there is no association in the liquid. The increase of bv with the temperature is seen to be very slight. From absolute zero to the maximum value at 100° the increase is at the rate of 0.074 cc per degree for gram mol quantities. On the other hand "6" changes markedly with the pressure. There are several reasons for believing that "b" is the real volume of the molecules and not four times the volume as was originally suggested. One is that "b" at the critical Fig. i — Volumes of molecules (b for gram mol quantities in cc.) for pentane. bl calculated from formula b1 = V1 — RT/(P + a/Vf ) assuming no asso- ciation; bv calculated by formula: bv = bc((sTc + T)/2TC)) b^, bvf apparent volume of bv by formula: &„' = Vv RT/(P + a/Vj) assuming no association. temperature is very nearly Vc/2 and this is just twice the volume at absolute zero. It is unlikely that the molecules do not expand in passing from absolute zero to the critical temperature, since at the former temperature they are under a pressure of 3572 atmospheres in pentane, whereas at the critical temperature the pressure is only 240 atmospheres. It would take very little separation of the atoms to double the volume. The latent heat shows, also, that heat is absorbed by the molecule roughly proportional to the number of atoms in the molecule. This would mean that the atoms vibrated (or expanded) and they must hence take up more space as the vigor of vibration increases. This would seem to be sufficient Do Molecules Attract, Etc. 529 to account for doubling the volume of b between o° Abs. and Tc. Van der Waals1 himself only assumed the constancy of the molecular volume "6" for simplicity, an'd has now definitely adopted the idea of a change in volume of the molecules. He has obtained, by making certain assumptions, the following expression for the change of ilb, " the molecular volume: (b — b0)/(V - - b) = = i - - (b -- b0Y/(bg — b0}\ In this equation bg and b0 are the limiting values of b ; bg at low vapor pressures, and b0 under high pressure; the actual value under any temperature and pressure is "b." Van der Waals assumes that in liquids at low temperatures, bg == 2b0. Both probability and direct observation lead, therefore, to the conclusion that the molecules expand on passing from the liquid to the vapor state. It is clear that if the molecules do thus expand against- the great force of atomic affinity, or intra-molecular cohesion, some heat must be absorbed. The quantity thus absorbed will probably be greatest at low temperatures, where there is a maximum difference in cohesive pressure between the liquid and the vapor, and will diminish rapidly near the critical temperature, since the volumes of the molecules in the two states approach each other and become equal at the critical temperature. As we near the critical temperature, therefore, the value of I will become very small, and the equation L - E == A will become very nearly true. The second reason why the internal latent heat of vaporization can not be assumed to go altogether to increas- ing the distance between the molecules is the fact that the internal pressure, the cohesive pressure, is inversely pro- portional to the square of the volume. For, assuming as be- fore that all the latent heat goes toward separating the mole- cules, if a/V2 is the cohesive pressure per unit surface then the cohesive energy in the liquid will be a/V^ and in the vapor, a/V,,; and the difference in their cohesive energies will be 1 Van der Waals: "The Liquid State and the Equation of Condition," Proc. Roy. Acad. Sci. Amsterdam (English Translation), 6, 123 ,(1903)- 530 Albert P. Mathews a(i/V1 -- i/VJ. By our assumption this difference in energy must be equal to L - - E. Hence (L - - E)/(i/V1 - - i/Vw) must equal "a." We come, therefore, to an expression different from Mills and one incompatible with it. Since it is certain that the cohesive pressure varies at least ap- proximately inversely as the square of the volume this ex- pression must be the correct expression, if Iy — E represents only heat consumed in overcoming cohesion. As a matter of fact (Iv — E)/(i/Ve — i/Vv) does not equal a constant, hence our assumption must be wrong. But if the assumption is wrong then the fact that (L — E)/(i/V'/3 — i/V^3) happens to equal a constant can not be adduced as evidence that molecules attract inversely as the square of the distance. I think therefore, that Mills' empirical expression, I, — g = ^'(i/V'/8 -- Vj,7*) does not mean, as he supposed, that the work done in overcoming molecular cohesion from the volume V1 to the volume Vv was equal to /(i/V^3 - i/V^3), but that the total internal latent heat, i. e., that used in overcoming molecular cohesion as well as that absorbed in the expansion of the molecules, or in doing other work is equal to this expression. That Mills' expression, /(i/V^1 - i/V^3), does not represent the work done in overcoming molecular cohesion may be shown, also, if the attempt is made to deduce the formula on this basis,*~assuming the attraction to vary in- versely as the square of the distance. A value is obtained for // widely^different from that found. Mills realized this difficulty and tried to avoid it by assuming that the law that matter attracted itself as the product of the masses was incorrect. ^ll will make the simplest possible assumptions. If the molecules are assumed to be cubical in shape, to lie a mean distance apart and the lines of attractive force to run per- pendicularly from*each face of the cube in three directions of space and to end upon the* six surrounding molecules, but not to penetrate them; and if the molecules attract with a force varying inversely as the square of the distance between Do Molecules Attract, Etc. 531 the centers and directly as the product of the cohesive masses M, then the attraction of two molecules would be M2K/V/3. The pressure per square cm would be M2K/7//3. Since the attraction goes but a single molecular diameter we may multiply the numerator and denominator by N4/3, where N is the number of molecules in the mass which will make N4/3M2K/V4/3. It is obvious that this expression cannot be true, for the cohesion varies inversely as the square of the volume, and not as the 4/3d power. Assuming, however, that it is correct we would have, as the difference in the cohesive energies in the liquid and vapor, the expres- sion: N4/3M2K(i/V/3 - i/V^/3). Or changing to density N4/3M2K(c/;/3 — D;/3)/Wt1/3. Hence I, — E should equal this expression, and (I, — E)/(d'/3 — D^3) - N4/3M2K/Wt1/3 = /. This last expression can be tested, since M2K can be easily computed from van der Waals "a" by dividing it by N2, the square of the number of molecules in the volume V, or Wt; and p.' is given by Mills. The two values are not of the same order of magnitude. For example in pentane, // is no, whereas N4/3M2K/Wt1/3 for i gram is 2.214 X io~13 calories. A constant very like // is obtained, however, if the foregoing constant is divided by V2/3/3N2/3. This changes it to the expression 3N2M2K/V2/3Wt'/3. This would give the value 1 02 for pentane. How closely this constant agrees with Mills is shown in Table i. N2M2K is equal to "a." The constant cannot be deduced, therefore, by the assumptions we have made, one of them being that molecules attract each other inversely as the square of the distance, but it is necessary to divide the theoretical constant by Vj/3/3N2/3 to get that found. This however, has the effect of changing the equation, near the critical temperature, to the form: Iy — E = 3a(i/V1 — i/Vv) which is almost identical with van der Waals. This argument will perhaps be still more convincing if it be turned around. Let us suppose Mills' contention is correct and the internal latent heat represents only heat used in overcoming cohesion, then //VJ/8 is the cohesive energy 532 Albert P. Mathews in the liquid; and ^'/V^8 is that in the vapor. If now we divide the first by Vl and the second by Vw we obtain the cohesive pressure per unit surface in the liquid and vapor, respectively, or ///V]/3 and /*'/V*/3. But we know that the cohesive pressure is a/V2, hence the original assumption must be incorrect. If, on the other hand, the difference in the cohesive energy in the liquid and vapor is equal to a(i/V1 - - i/Vv), and all the latent heat is used in overcoming external and cohesive pressure, then L — E should equal this expression. If, however, some of the heat, I, is used in expanding the molecules then the equation should be : L — E — I = a(i/Vl - i/VJ. Near the critical temperature I becomes very small and hence (L — E)/(^ - - DJ, near the critical temperature, must be very nearly equal to a/Wt, where Wt is the weight taken in grams; but at temperatures below the critical, (If - - E)/(^ - - DJ should be progressively greater than a/Wt by the amount l/(dl - - DJ. Table 4 shows that this expectation is realized, since in all cases the value (L -- E)/- (VMol. WtJ/s, seems at first peculiar. It is odd that the valence of an atom should be of as much importance in cohesion as the weight of the atom; it is a relationship which one would not have anticipated. The significance of this fact, if I am not mistaken, is that the electron couples constituting the molecules are of two kinds, namely, those of the atoms themselves, which added together presumably give the molecular weight; and the valence elec- trons, which differ from the others so that they cannot be added to them. Hence the formula is not M2K = (/) (Wt. + Val.), the cohesion being proportional to the sum; but the mass of cohesion is proportional to the cube root of each of 1 Sutherland: Phil. Mag., [6] 4, 632 (1902). Relationship between Molecular Cohesion 483 these kinds of electrons and so is proportional to the cube root of their product. The valence electrons are probably more labile, more easily removed and replaced. They have a different degree of liberty and they cannot be summed with the atomic. The formula thus confirms the correctness of Drude's promise that the electrons of the valences differ in their properties from the electrons of the atoms. He concluded that only the valence electrons would be sufficiently free to vibrate synchronously with light and hence these electrons must be particularly concerned in the refraction and dis- persion of light. Drude's1 suggestion of electrons of different degrees of liberty confirmed, as it was, by experiments show- ing a relation between valence and dispersion, is thus con- firmed also from the wholly different field of cohesion. A still more interesting conclusion may be drawn from this relationship, namely, that a neutral, uncharged atom having no valence will have no cohesion. Since it will have no chemical affinity either, if chemical affinity is, as it ap- pears to be, of an electrical nature, it is thus seen that a close relation must exist between chemical affinity and cohesion. Such neutral atoms will presumably still have gravitational attraction. A free electrical charge on the atom is, therefore, necessary for cohesion, but not for gravitation. Furthermore, the cohesional effect is the same whether the charge be positive or negative; and it is proportional to the number of charges. The formula shows, also, that the effect of a free charge on any atom is proportional to the weight of the atom; that is, the effect of the valence charge is multiplied, as it were, by the number of electron couples in the atom; and the effect of the total number of valence charges in the molecule is multiplied by the whole number of atomic electron couples in the mole- cule. Just how such an effect could be produced, and why the attraction, or cohesive mass, should ultimately prove to be proportional to a linear function (the cube root) of the product Drude: Annalen der Physik., [4] 14, 677 (1904)- 484 Albert P. Mathews of the number of valences by the molecular weight, I do not see. It appears, then, that refraction, dispersion and cohesion all involve the valence electrons, but the connection between cohesion and valence is far closer and simpler than the other relationships appear to be. The relationship of valence to light is necessarily a less direct one, refraction depending on the rate of vibration of the electron. It is said1 that if the natural period of the molecule (electron) is slightly less than the frequency of a light wave the light will be accelerated; if greater, retarded. It is evident that in dispersion other properties of the electrons than number come into play, and, hence, the relationship between dispersion and number is not so simple and direct. Double bonds, neighboring groups, etc., influence the periods of the electrons and so influence the dispersive power; whereas these factors appear to play no important part in cohesion. The relation between the refraction of light of one wave length and the valence number is still less direct than be- tween dispersion and valence, but still a general relation exists which for substances of the same type is rather uniform, as shown by Traube2 for many liquids and by Cuthbertson3 for several gases. Another very interesting fact correlating the refractive and cohesive properties of matter is the resemblance between the constant "K" of the Ketteler dispersion formula and the value M2K of cohesion. Thus with the Ketteler formula n2 = a2 — K P + D X\/(P — Pv) the constant "K," Drude found, could be computed with a fair approximation, in some cases at any rate, from the sum of the valences, the molecular weight and the density, and this result was confirmed by Erfle.4 This constant "K," therefore, contains at least 1 Cotter, J. R.: "Dispersion," Encyclo. Brit., nth edition, 8, 317. 2 Traube: Ber. chem. Ges. Berlin, 40, 130 (1907). 3 Cuthbertson: Proc. Roy. Soc., 8aA (1909-1910); Phil. Mag., [6] 21, 69 (1911); Phil. Trans., 204, 323 (1905); 207, 135 (1907). 4 Erne: "Optische Eigenschaften und Elektronen Theorie." Annalen der Physik, [4] 24 (1907). Relationship between Molecular Cohesion 485 sometimes the same factors as the value a/V2 of van^der Waals' equation. I have not, however, attempted to estab- lish any closer connection between them.1 We conclude, therefore, that while cohesion and re- fractivity are both dependent on a common factor, namely the valence electrons, and possibly upon the molecular weight, the connection between them is not direct, but indirect; and while cohesion and refraction, or dispersion, often parallel each other, they, at other times, diverge considerably since other factors enter into refraction. It is not without interest to recall as an example of the perspicacity of genius, that I^aplace2 long ago foretold a con- nection between these properties. Writing in 1805 of the formula of capillarity which, as will be remembered, con- tained two terms, -one K, representing molecular cohesion, or van der Waals' expression a/V; the other, H, the capillary constant, Laplace says (p. 351): "I saw that this action (pressure) is smaller or larger than if the surface is plane; smaller, if the surface is concave; larger, if it is convex. Its analytical expression is composed of two terms: the first (K), much larger than the second, expresses the action of the mass terminated by a plane surface; and I think from this term depends the suspension of mercury in a barometer tube at a height two to three times greater than that due to atmospheric pressure, the refractive powers of diaphanous bodies, the cohesion, and in general, chemical affinity; the second term expresses the part of the action due to the spheri- city of the surface." And again (p. 362): "The function, K, is analogous to that I have designated by the same letter in the refraction of light." But of even greater interest and more fundamental importance than the relation between the optical and the cohesive properties, which is now understandable since both 1 See also Natanson: Bull, de 1'Acad. des Sci. de Cracovie, 1907, April p. 316 for the relation of refraction and valence. 2 Laplace: Sur 1'action capillaire. Oeuvres. Supp. Liv. X, Trait£ de Mecanique Celeste, p. 351. 486 Albert P. Mathews involve the number of valence electrons, is the relation be- tween the magnetic and cohesive properties, since here we touch, I think, the very kernel of the problem of the nature of cohesion. The connection between the magnetic properties and cohesion is brought out very clearly, in an empirical way, by Pascal's1 investigations on the relation between magnetic susceptibility and the molecular properties. In a series of papers Pascal has shown that there is a remarkable con- nection between the specific susceptibility of diamagnetic elements and the atomic weights and valences. Thus if elements of the same family having the same valence are arranged in their order of increasing specific susceptibility, they are in the order of their atomic weights; and, on the other hand, a close dependence on valence may be observed. In elements of nearly the same weight but of different valence numbers, the atomic magnetic susceptibility, XH, increases with the number of valences. An empirical relationship was found between them, Xa — io~7ett + fta where a is about 2.1, ft about 0.004 and a the atomic weight, a and ft depend only on valence. But here, also, double bonds made their effect felt, though less pronouncedly than in refraction. Double bonded molecules have, in general, a lower molecular susceptibility than that calculated. Other effects of molecular form are seen; for example, the benzene nucleus augments the diamagnetism, although the double bonds it is supposed to contain should have an opposite effect. The factor of molecular structure appears, then, to influence diamagnetism. On the whole, however, calculation of the number of valences in the molecule by this procedure agrees better with the calcu- lation from the cohesion, than does the calculation from the refractivity or dispersion. Thus, double bonds are of less action on the diamagnetism and in the benzene nucleus they 1 Pascal : ' 'Sur un mode de contr61e optique des analyses magpie" tochimiques," Comptes rendus, 152, 1852 (1911). Recherches magneto-chimiques sur la struc- ture atomique des halogenes," Comptes rendus, 152, 826 (1911). Ann. Chim. Phys., [8] 19, 1-80 (1910). Relationship between Molecular Cohesion 487 appear to exert no effect. By this method, as by the cohesional, all organic chlorine compounds examined were found to have trivalent chlorine and fluorine was monovalent. The agree- ment was good in regard to other elements also. The connection between cohesion and diamagnetism is, therefore, again an indirect one. Both involve the molecular weight and the number of valences, but it is clear that cohesion is independent of molecular form, or very largely so; whereas whether a substance is magnetic, or diamagnetic, may depend, in part upon this very factor. Oxygen in an elemental form, is paramagnetic, not diamagnetic, and is quite anomalous in Pascal's scheme; whereas the cohesion of oxygen is not anomal- ous. In other words, whether a body is, as a whole, para- magnetic or diamagnetic, and to what degree, depends, prob- ably, on the possibility of the orientation of the molecules, their polarity, etc., factors which do not seem to affect their cohesion or gravitation. Nevertheless cohesion and magnetic properties are, no doubt, closely related, since both depend on the same molecular properties, only magnetism involves still other properties, (form) not involved in cohesion. The fact that cohesion is thus determined by the number of electron couples (atomic and valence) in the molecule plainly points toward the conclusion that cohesion is either electro-static or electro-magnetic in nature. Both of these possibilities have already been suggested. Sutherland, * from his discovery of the relation of cohesion to valence in salts, inferred at first that the cohesion must be of an electromagnetic nature. He supposed these rotating electron couples acted like little magnets, and he attempted to show, though whether successfully, or not, I am unable to judge, that small magnets, at sufficient distances apart, would attract inversely as the fourth power of the distance between them, and this he sup- posed to be the law of molecular attraction. This conclusion was attacked by van der Waals, Jr.,2 who concluded, also 1 Sutherland: Phil. Mag., [6] 19, i (1910); 4, 625 (1902). 2 Van der Waals, Jr. : Kon. Akad. v. Wetensch. te Amsterdam, Pro- ceedings, n, 132 (1908-1909). 488 Albert P. Mathews from mathematical reasoning, that the attraction between such magnets would be inversely as the yth or 9th power of the distance, and thus agree with the assumptions of his father, that molecular attraction diminished at such a rate that it was effective only when the molecules were in contact. Later, Sutherland1 concluded that cohesional attraction was due to electrostatic affinity of these electron couples. Lodge2 made a similar suggestion. He thought some of the lines of force between the atoms wandered outside the molecule to atoms of other molecules and thus produced molecular cohesion. This would make molecular cohesion of the same nature as chemical affinity. While the relation between the two is close, both being zero in the absence of an electric charge, or valence, one is not causally dependent on the other, although both depend on the valences. The attraction between the atoms is probably of an electrostatic kind and, if so, should vary inversely as the square of the distance. The atomic weight does not appear to play a part in chemical affinity, for very light elements may enter into very firm union. In cohesion, molecular weight does play a large part; and while it is not impossible that the cohesional attraction may be inversely as the square of the distance, it is not probable, or at any rate it has not yet been proved to general satisfaction. It seems much more probable to me that cohesion is more closely related to magnetism than to electrostatic affinity and I would raise the question whether magnetism is anything else than molecular cohesion made apparent at distances more than molecular. Is it not possible that molecular cohe- sion, involving as it does both atomic and valence electrons (atomic weight and valence) is due, perhaps, to the magnetic effects produced by the movements of these electron couples? In this view the atoms would be united by their electro- static affinities and these same valences and the other atomic electrons by their magnetic effects produce the molecular 1 Sutherland: Phil. Mag., [6] 17, 667 (1909). 2 Lodge: Nature, 70, 176 (1904). Relationship between Molecular Cohesion 489 cohesion. I may state briefly some of the reasons which ap- pear to lead to such a conclusion. In the first place we have the surprising fact that the field of cohesion of a molecule is apparently delimited by the surrounding molecules. The evidence for this, while perhaps not conclusive, is both direct and indirect. The reason for the shortness of the radius of action of cohesion is one of the most interesting questions of molecular physics. It is of interest to see how this question came to be generally considered closed and settled in favor of the view, now gener- ally accepted, that cohesional attraction diminishes with the distance at a rate far greater than gravitational attraction. It is chiefly due to Laplace. Laplace,1 in his beautiful memoir on capillarity, first raised the question whether the short radius of attraction of the cohesive forces was due to the fact that matter shut off the attraction, or was due to the attraction diminishing with the distance at a rate far more rapid than gravitation. He says, when discussing Hawksbee's well-known ex- periments proving that the height to which water rises in a glass tube is independent of the thickness of the wall of the tube: "Hawksbee a observe que dans les tubes de verre, ou tres minces ou tres epais, 1'eau s'elevait a la meme hauteur toutes les fois que les diametres interieurs etaient les memes. Les couches cylindriques du verre qui sont a une distance sensible de la surface interieure ne contribuent done point a 1' ascension de 1'eau, quoique dans chacune d'elles, prise separement, ce fluide doive s'elever au-dessus du niveau. Ce n'est point 1'interposition des couches qu'elles embrassent qui arrete leur action sur 1'eau, car il est naturel de penser que les attractions capillaires se transmettent a travers les corps, ainsi que la pesanteur; cette action ne disparait done qu'a raison de la distance du fluide a ces couches, d'ou il suit que V 'attraction du verre sur Veau n'est sensible qu'a des distances insensibles." I have italicized the end of Laplace's statement to bring out 1 Laplace: "Sur 1'action capillaire. Oeuvres. Supp. au Livre X," Traite de Mecanique Celeste, p. 351; see also p. 487. 490 Albert P. Mathews clearly the reason which led him to the conclusion that molecu- lar cohesion penetrated matter like gravitation, and that the attraction must, hence, decrease very rapidly with the dis- tance. It will be seen that the sole reason for his decision was the possible analogy between gravitation and molecular cohesion. The very important result of the rejection by Laplace of the possibility of cohesive attraction not penetrating matter was that it forced him to the conclusion that the cohesive force must diminish far more rapidly than gravitation as the distance increases. Laplace did not make any assumption as to the rate at which the cohesional attraction diminished with the distance, except that it was at so rapid a rate that the cohesion became negligible within all measurable dis- tances. The great English philosopher, Thomas Young,1 who a year before Laplace had shown the true nature of surface tension and practically anticipated all of Laplace's main conclusions, does not appear to have raised the question in a concrete form. His papers on capillarity are so condensed that the reasoning is very difficult to follow.2 But while Young nowhere specifically puts the question whether the cohesional attraction penetrates matter, he made an assumption which might be taken to indicate that it does not. "We may suppose," he says (p. 43), "the parti- cles of liquids, and probably those of solids also, to possess that power of repulsion which has been demonstratively shown by Newton to exist in aeriform fluids, and which varies as the simple inverse ratio of the distance of the particles from each other. In air and vapors this force appears to act 1 Young: "An Essay on the Cohesion of Fluids," Phil. Trans., 1805 (collected works, edited by G. Peacock, i, 418 (1855), London). 2 A propos of this paper of Young's, Clerk Maxwell makes an interesting comment. He says: "His [Young's] essay contains the solution of a great number of cases including most of those afterwards solved by Laplace; but his methods of demonstration, though always correct and often extremely elegant, are sometimes rendered obscure by his scrupulous avoidance of mathematical symbols." Ency. Brit., Article, "Capillarity." Relationship between Molecular Cohesion 491 uncontrolled; but in liquids it is overcome by a cohesive force, while the particles still retain a power of moving freely in all directions; and in solids the same cohesion is accompanied by a stronger or weaker resistance to all lateral motion, which is perfectly independent of the cohesive force and which must be cautiously distinguished from it." "// is sufficient to suppose the force of cohesion nearly or perfectly constant in its magnitude throughout the minute distance to which it extends, and owing its apparent diversity to the contrary action of the repulsive force, which varies with the distance. Now, in the internal parts of a liquid, these forces hold each other in a perfect equilibrium, the particles being brought so near that the repulsion becomes precisely equal to the cohesive force that urges them together," etc. Young thus assumed that the cohesion extended but a short distance, with slight variation in intensity and that it then ended abruptly. So far as I can find, he made no suggestion how it came to end abruptly; but if it be assumed that it does not penetrate matter, it is seen that it must end abruptly at the next layer of molecules. Young tried to esti- mate how far the cohesive force really extended, and found a value surprisingly near the order of magnitude of that now known to be the distance apart of the centers of two molecules. His reasoning on this point is extremely ingenious, and is of interest as the first estimate of molecular dimensions. lyord Rayleigh1 says anent this computation of Young's: "One of the most remarkable features of Young's treatise is his estimate of the range "a" of the attractive force on the basis of the relation T = V3aK. Never once have I seen it alluded to, and it is, I believe, generally supposed that the first attempt of this kind is not more than twenty years old. It detracts nothing from the merit of this wonderful specu- lation that a more precise calculation does not verify the numerical coefficient in Young's equation. The point is 1 Rayleigh: "On the Theory of Surface Forces," Phil. Mag., [5] 30, 285-298, 456-475 (1890). Collected papers, 3, 396. 492 Albert P. Mathews that the range of the cohesive forces is necessarily of the order T/K." T is the surface tension and K the internal pressure. Lord Rayleigh, in his revision of Maxwell's classical account of capillarity in the new edition of the Encyclopaedia Britannica and in his many splendid writings on this subject, does not seem to have considered this question. All other writers whom I have consulted seem to have followed Laplace's lead and assumed, without evidence, that cohesion does penetrate matter like gravitation. Thus, Gauss,1 who introduced clear ideas of surface energy and the potential energy of fluids, writes as follows in 1830: "The ordinary attraction which is proportional to the square of the distance, and which permits the repre- sentation of all motions in the heavens with such good agree- ment, can be used in the explanation neither of capillary phenomena nor of adhesion and cohesion; a correctly carried out computation shows ' dass eine nach diesem Gesetze wirk- ende Anziehung eines beliebigen Korpers der zur Ausfuhrung von Experimenten geeignet ist, d. h., dessen Masse im Ver- gleich mit der der Erde vernachlassigt werden kann, auf einem beliebig gelegenen, sogar den Korper beriihrenden Punkt, im Vergleich mit der Schwere verschwinden muss. Wir schliessen hieraus dass jenes Anziehungsgesetz in den kleinsten Abstanden mit der Wahrheit nicht mehr iiberein- stimmt, sondern dass es eine Modification erfordert. ' In other words, the particles of the body exert, besides the at- tractive force of gravitations, still another force which is notice- able only , in the smallest distances. All appearances show uniformly that the second part of the attractive force (the molecular attraction) is not noticeable in the smallest meas- urable distances. On the other hand, in unmeasurably small distances it may greatly surpass the first, which is propor- tioned to the square of the distance. 1 Gauss: "Allgemeine Grundlagen einer Theorie der Gestalt von Fliis- sigkeiten," Ostwald's "Klassiker der exakten Wissenschaften," No. 135, p. i. "Commentationes societatis Regiae Scientiarium Gottingensis Recentiores," vii, 1830. Relationship between Molecular Cohesion 493 It was necessary for him (p. 21), to make some assump- tions regarding the cohesional attraction which he represented as "/" (r)> r being the radius of molecular action, and he accordingly adopted Laplace's view that / (r) decreases far more rapidly than i/r2, which is the law of gravitational attraction (p. 22). He says, speaking of cohesional attrac- tion, or / (r) : "Da dieser Ausdruck etwas unbestimmtes hat so lange wir nicht eine Einheit zu Grunde legen, wollen wir vor allem darauf aufmerksam machen, dass wir die anzie- hende Kraft / (r), ausgedriickt als eine Function des Abstandes r, mit einer Masse multipliziert denken mussen, damit sie mit der Gravitation g in den Dimensionen ubereinstimmt. Der Sinn unserer Voraussetzung ist dann der folgende : Bezeich- net M irgend eine Masse derart, wie sie uns in Experimenten vorkommt, namlich eine, die im Vergleich mit der ganzen Erde als verschwindend angesehen werden kann, dann muss M / (r) immer merklich sein im Vergleich mit der Schwere, so lange r einen unseren Messungen zuganglichen, wenn auch noch so kleinen Wert hat." Van der Waals, in all his earlier writings, including his famous essay on the continuity of the gaseous and liquid states published in 1869, assumed with Laplace that molec- ular cohesion was appreciable only very close to the molecule ; indeed, the radius of action was less than the mean distance apart of the molecular centers. I have not been able to find in his essays any specific discussion of the question whether cohesional attraction penetrates matter, although he does discuss the radius of attraction of a molecule.1 His general assumption was that the cohesion diminished very rapidly as the molecules separated. In one brief communication to the Amsterdam Academy of Sciences (1893-94, pp. 20-21) he states that he had derived a potential function, that is, a function expressing the potential energy of attraction of two 1 Van der Waals: "Bijdrage tot de Kennis van de Wet der Overeenstem- mende Toestanden," Verhandelingen Konik. Akad. van Wettenschappen, 21, 5 (1881). 494 Albert P. Mathews molecules, of the form — / e~ _J", as the potential of two mass points; and he goes on to say that this formula was based on two assumptions, namely that the attraction of two molecules is inversely as the square of the distance, and second, that the universal medium absorbs the lines of force. He thus assumes an absorption by the ether of the attraction, rather than by a molecule at a distance "r." In his paper published in 1903, in which the real molecular volume, the value of "6," is no longer considered constant and in which he has revised his formula for the isotherm in so important a manner, he does not specifically reraise the question of the penetration of matter by the cohesive attraction. In a succession of papers like those of van der Waals' which represent a progressive succession of ideas, mutually conflicting ideas may, not unnaturally, be found. Thus, in his paper on the thermodynamic potential and capillarity,1 a limiting intermediate layer of rapidly changing density is supposed to exist between the saturated vapor and the liquid, and the existence of such a layer would seem to the writer to presuppose that the radius of attraction at least in this layer must be several molecular diameters. On the other hand, the following statement2 (p. 121) is not entirely reconcilable with this view, and would be so only if the layers of which he speaks are at least equal in thickness to the distance the cohesive force extends. He supposes the cohesive pressure to be exerted only by the surface layer, but he states that exactly the same formulas are obtained if the fluid be considered to be made up of a series of layers of molecular dimensions. "If we consider the gas in a cylindrical vessel of constant area and divided into horizontal layers, the lowest attracts the next higher," etc. The sum of all partial amounts of work will be the same as if one considered 1 Van der Waals: "Theorie thermodynamique de la Capillarite," Archives Neerlandaises des Sciences exactes et naturelles, 28, 121 (1895). 2 Van der Waals : "Die Continuitat des gasformigen u. fliissigen Zustandes," Leipzig, p. 126 (1899). Relationship between Molecular Cohesion 495 only the attraction of the upper layer and the distance as that through which this layer would have been moved. It seems to me that, for this reasoning to be correct, we must assume the layers at least as thick as the radius of action of each layer of molecules. If these layers are of molecular diameters, it would seem that the cohesive force does not extend farther than a molecular diameter. This is not apparently con- sistent with the assumption elsewhere made to explain the transitional layer between vapor and liquid. Plateau also followed Laplace. Sutherland, in his many papers on molecular cohesion, assumes that the cohesional attraction is inversely as the fourth power of the distance, but he does not, so far as I can find, discuss the question of the penetration of matter by cohesional attraction. But it is impossible for cohesion to vary inversely as the fourth power, if cohesion penetrates matter. Kleeman1 supposes that the attraction must be in- versely as the 5th or some higher power of the distance. He has pointed out that cohesion cannot possibly be assumed to vary like gravitation inversely with the square of the distance, because the cohesional attraction is so much greater than gravitation. But Mills weakened the force of this objection, which he had overlooked in his early papers, by making the additional postulate that cohesion does not penetrate matter. For if it be assumed that the cohesion does not penetrate matter then only the layer of superficial molecules of two masses would attract each other; and the number of these is so small, compared to the whole number of molecules in the mass, that the cohesional attraction would be less perceptible at sensible distances than gravitational attraction. I have been unable, then, to find any evidence for the assumption so generally made that cohesive attraction pene- trates matter like gravitation. There is no direct evidence, therefore, so far as I can find, against the inference necessitated 1 Kleeman: "An Investigation of the Determination of the Law of Chemical Attraction between Atoms from Physical Data," Phil. Mag., [6] 21, 83 (1911). 496 Albert P. Mathews by the square or fourth power law of attraction, that cohesion does not penetrate matter. There is, on the other hand, some evidence of a direct kind that cohesional attraction extends only as far as the nearest molecules. This evidence is the length of the radius of action as determined by direct measurement, and Einstein's proof that the radius of action varies with the distance apart of the molecular centers. Laplace believed that the radius of action, although short, nevertheless extended many molecular diameters and this opinion prevailed until recently, but as means of measure- ment have improved the radius has shrunk. Quincke gave an estimate of about 6 X io~6 cm, but the most recent deter- minations of Johannot,1 and Chamberlain2 show it to be about 1.6-2 X io~7 cm in a soap film and in the case of glass. The diameter of a molecule of trioleate of glycerine, according to Perrin,3 is i.i X io~7 cm. The radius of action is certainly not more than two molecular diameters and indeed is hardly more than one. The average distance between the centers of two molecules of ether in the liquid state at 20° is about 5.5 X io-8cm. But not only has direct measurement shown the radius of action to be one or two molecular diameters, but com- putations of it by Kleeman make it very close to this. For example, in ether Kleeman4 computed the radius to be about 3.4 X io~8 cm which is about the distance when the mole- cules are in contact. Van der Waals supposed it, indeed, to be only as long as this. Recently Einstein5 has made a very interesting computation starting from the law of Eotvos, by which he shows, from thermodynamic reasoning, that the 1 Johannot: Phil. Mag., [5] 47, 501 (1899). 2 Chamberlain: Phys. Rev., 31, 170 (1910). 3 Devaux: Journal de Physique, [5] 2, 699 (1912). 4 Kleeman: "On the Radius of the Sphere of Action," Phil. Mag., [6] 19, 840 (1910). 5 Einstein: "Bemerkung zudemGesetz von Eotvos," Annalen der Physik., [4l 34, 165 (1911). Relationship between Molecular Cohesion 497 range of cohesive action must be of the general value of, and proportional to, the distance between the molecular centers. This distance is of the order of magnitude in most liquids of io~8 cm. This result is so surprising that Einstein says of it: "This result appears at first very unlikely, for what should the radius of action of a molecule have to do with the distance between neighboring molecules? The supposi- tion is only reasonable in case the neighboring molecules alone attract each other, but not those farther removed. " Sutherland1 also came to the conclusion that the radius of action was about equal to 7//3. We see then, that the evi- dence points to the conclusion that the radius of action of the molecular forces agrees very closely with -z//3, the distance between the molecular centers, and varies directly with T//3. The only explanation of this fact appears to me to be that the attraction does not extend beyond neighboring molecules and, hence, must, in some way, be stopped by them. Mills2 alone, so far as I can find, has reopened the question whether the field is delimited by the surrounding molecules. Concluding, I believe erroneously, from an empirical law, that the attraction between molecules must vary inversely as the square of the distance, he was driven to the second conclusion that if this were the case the cohesion could not penetrate matter. He assumed, hence, that the surrounding molecules absorbed, or neutralized, the lines of cohesive force. The direct evidence and Einstein's reasoning leaves little doubt that the radius varies with the distance apart of the molecules and the only possible conclusion from this is that the surrounding molecules delimit the field as Mills supposes. If it is a fact that the surrounding molecules delimit the field as the evidence indicates, cohesion is allied at once with magnetism, foi" this is the very supposition which Ewing made to explain some of the phenomena of magnetism. Each molecule of a ferromagnetic substance is supposed to be a 1 Sutherland: Phil. Mag., [6] 4, 632, 636 (1902). 2 Mills: Jour. Phys. Chem., 15, 417 (1911). 498 Albert P. Mathews magnet. In the non-magnetic state the magnetism of each molecule is supposed to be neutralized by the surrounding molecules, which have their magnetic axes variously directed. The magnetic field is thus limited to a single molecular diam- eter and will vary with the distance apart of the molecules. The magnetic field of each molecule is delimited by the sur- rounding molecules. If, however, these molecules are oriented, either by acting on each ether, or by external forces, then the magnetic fields coincide and the magnetism may be perceived extending outward from the mass. In other words, to explain why magnetism does not persist in soft iron and to account for magnetic hysteresis, Ewing has made exactly the same assumption which has been made to explain why cohesion does not extend beyond molecular distances. It seems to me not impossible that magnetism is the cohesive attraction of a molecule. If the molecule is of such a nature, or shape, that the total effect of the little magnets, its electron couples, coincide more or less completely so that the molecule has a polarity, and the molecules can be oriented in any way and held in position, we have the ferro and para- magnetic substances. If the molecules have many poles, so that there is no polarity of the molecules as a whole, there is a diamagnetic substance. The magnetic field of each molecule would be its cohesive field. The recent work of Cotton and Mouton1 on the orienta- tion of molecules in the magnetic field seems to me to bear out such an interpretation. The work of Weiss2 and Langevin3 appears to point in the same direction, but Weiss, who has considered the possibility of the identity of cohesion and magnetism, states that he will shortly show that they can not be identical. Nevertheless it appears to me not impossible that magnetism is simply a special case of cohesion, and if this is true the rotation of the plane of polarized light by op- tically active substances would be easily understood. 1 Cotton and Mouton: Journal de Physique, [5] I, 40 (1911). 2 Weiss: Journal de Physique, [5] I, 900 (1911); [4] 6, 66 1 (1907). 3 Langevin: Ann. Chim. Phys., 5, 70 (1905). Relationship between Molecular Cohesion 499 Finally, if it is true that the surrounding molecules de- limit the field of cohesion in any way whatever, and that only six molecules really take part in this delimitation, we can derive the value a/V2 of van der Waals' at once and very simply. Suppose that each molecule has a certain mass of cohes- ion, M, and that two molecules attract each other directly as the product of their cohesive masses and inversely as the fourth power of the distance between their centers, then the attraction between two molecules would be M2K/V/3, where v is the space at the disposal of a single molecule. Since each molecule attracts only the one above, below and to each side, the pressure per square centimeter of a double layer of molecules will be M2K/V/3 multiplied by the number of mole- cules in i sq. cm or i/z>2/3, making M2K/7;2. Since the at- traction extends only a single molecular diameter, we may multiply both numerator and denominator by N2, where N is the number of molecules in the volume, V, and we obtain, N2M2K/NV -= M2N2K/V2 == a/V2. M2K has already been shown to be 2.98 X io~37 (Mol. Wt. X Valences)2/3. We have, as yet, no proof that only the six surrounding molecules are attracted, although Sutherland1 has made a similar supposition. But such may be the case nevertheless. Einstein, in his calculation, computed that each molecule could be considered as lying at the center of a cube and that it attracted the 26 other molecules of the cube. Of course the fact that the value a/V2 may be so easily derived in this way does not furnish any proof that the attraction is inversely as the fourth power of the distance. But I know of no other derivation of a/V2 which involves so few, or less radical, assumptions. The general conclusion of the paper is then, that cohesion, 1 Sutherland : Phil. Mag., [6] 17, 667 (1909). "The total potential energy of a number of like molecules is the same as if each caused its own domain to be uniformly electrized with an electric moment proportional to the linear dimensions of the domain, the direction of electrization being such that in general any molecule attracts its six immediate neighbors." 500 Albert P. Mathews being a function of molecular weight and valence, is a function of the number of electron couples of the valences and atoms, and is, hence, probably of a magnetic nature. Magnetic substances may be supposed to be substances in which, owing to the orientation of the molecules, or to their polarity, or both these causes, the cohesive fields of the molecules are not delimited or neutralized by the surrounding molecules so that the cohesional attraction becomes apparent at more than molecular distances, and under such circumstances the sub- stance is said to be magnetic. University of Chicago THE SIGNIFICANCE OF THE RELATIONSHIP BE- TWEEN MOLECULAR COHESION AND THE PROD- UCT OF THE MOLECULAR WEIGHT AND THE NUMBER OF VALENCES BY ALBERT p. MATHEWS In the preceding papers1 of this series I have shown that the value of "a" of van der Waals, representing molecular cohesion, or the value M2K which is the factor "a" for a single molecule, is proportional to the two-thirds power of the product of the molecular weight by the number of valences of the molecule. M2K was found to be equal, when expressed in absolute units, to 2.98 X io~37 (Mol. Wt. X No. of Val.)2/3. In this paper I shall discuss the theoretical bearing of the relationship of cohesion to these molecular properties. Attempts have been made by others to correlate cohesion, or ''a," with molecular weight and the number of valences, but with very partial success. Sutherland2 at first supposed the molecular attraction to be proportional to the product of the gravitational masses of the molecules. This he found would not do, and in his later papers he stated that the gravita- tional mass of a molecule did not enter into the expression "a." Amagat,3 also, recently revived the idea that gravita- tional mass plays a role in cohesion and suggested that a/V2 ought to be proportional to the square of the molecular mass. This, however, he did not find to be the case. Leduc4 has recently confirmed, in part, this view of Amagat's for gases of similar molecular composition, when taken under the same volume and at corresponding temperatures. Kleeman,5 also has tried to find a relationship between "a" and gravita- 1 Mathews: Jour. Phys. Chem., 17, 154 (1913)- 2 Sutherland: Phil. Mag., [5] 27, 305 (1889); [6] 4, 632 (1902). 3 Amagat: "Pression interne des fluides," Journal de Physique, [4] 8, 617 (1909). 4 Leduc: Comptes rendus, 153, i?9 (I911)- 6 Kleeman: Phil. Mag., [6] 19, 783, 840-847 (1910). 482 Albert P. Mathews tional mass, and states that the cohesive attraction of two molecules is proportional to the product of the two sums of the square roots of the atomic weights of the atoms of the molecules. This relationship, however, is of very limited applicability, ifjindeed, it correctly expresses the cohesion of any. As regards valence, I can find but one other suggestion, that of Sutherland.1 He showed that the number of equiva- lents, or valences, in simple substances, such as sodium chloride, influenced the value of their cohesion. He was unable to establish this relationship for more complex bodies. Nevertheless he assumed that it existed in them and correctly surmised from it the relationship between cohesion and chemical affinity, and adduced it as evidence of the electro- static or magnetic nature of cohesion, "a" was made pro- portional to the square root of the valence. The relationship between cohesion and the properties of molecular weight and the number of valences can be inter- preted best by Sir J. J. Thomson's theory of the electrical constitution of matter and valence, and, so far as I can see, on no other hypothesis. It speaks, therefore, for the electro- static, "or electro-magnetic theory of cohesion, and, in my opinion, for the latter. The relation, M2K = (/) Val2/3/Mol. Wt.8/3, seems at first peculiar. It is odd that the valence of an atom should be of as much importance in cohesion as the weight of the atom; it is a relationship which one would not have anticipated. The significance of this fact, if I am not mistaken, is that the electron couples constituting the molecules are of two kinds, namely, those of the atoms themselves, which added together presumably give the molecular weight; and the valence elec- trons, which differ from the others so that they cannot be added to them. Hence the formula is not M2K = (/) (Wt. + Val.), the cohesion being proportional to the sum; but the mass of cohesion is proportional to the cube root of each of 1 Sutherland: Phil. Mag., [6] 4, 632 (1902). Relationship between Molecular Cohesion 483 these kinds of electrons and so is proportional to the cube root of their product. The valence electrons are probably more labile, more easily removed arid replaced. They have a different degree of liberty and they cannot be summed with the atomic. The formula thus confirms the correctness of Drude's promise that the electrons of the valences differ in their properties from the electrons of the atoms. He concluded that only the valence electrons would be sufficiently free to vibrate synchronously with light and hence these electrons must be particularly concerned in the refraction and dis- persion of light. Drude's1 suggestion of electrons of different degrees of liberty confirmed, as it was, by experiments show- ing a relation between valence and dispersion, is thus con- firmed also from the wholly different field of cohesion. A still more interesting conclusion may be drawn from this relationship, namely, that a neutral, uncharged atom having no valence will have no cohesion. Since it will have no chemical affinity either, if chemical affinity is, as it ap- pears to be, of an electrical nature, it is thus seen that a close relation must exist between chemical affinity and cohesion. Such neutral atoms will presumably still have gravitational attraction. A free electrical charge on the atom is, therefore, necessary for cohesion, but not for gravitation. Furthermore, the cohesional effect is the same whether the charge be positive or negative; and it is proportional to the number of charges. The formula shows, also, that the effect of a free charge on any atom is proportional to the weight of the atom; that is, the effect of the valence charge is multiplied, as it were, by the number of electron couples in the atom; and the effect of the total number of valence charges in the molecule is multiplied by the whole number of atomic electron couples in the mole- cule. Just how such an effect could be produced, and why the attraction, or cohesive mass, should ultimately prove to be proportional to a linear function (the cube root) of the product 1 Drude: Annalen der Physik., [4] 14, 677 (1904). 484 Albert P. Mathews of the number of valences by the molecular weight, I do not see. It appears, then, that refraction, dispersion and cohesion all involve the valence electrons, but the connection between cohesion and valence is far closer and simpler than the other relationships appear to be. The relationship of valence to light is necessarily a less direct one, refraction depending on the rate of vibration of the electron. It is said1 that if the natural period of the molecule (electron) is slightly less than the frequency of a light wave the light will be accelerated; if greater, retarded. It is evident that in dispersion other properties of the electrons than number come into play, and, hence, the relationship between dispersion and number is not so simple and direct. Double bonds, neighboring groups, etc., influence the periods of the electrons and so influence the dispersive power; whereas these factors appear to play no important part in cohesion. The relation between the refraction of light of one wave length and the valence number is still less direct than be- tween dispersion and valence, but still a general relation exists which for substances of the same type is rather uniform, as shown by Traube2 for many liquids and by Cuthbertson3 for several gases. Another very interesting fact correlating the refractive and cohesive properties of matter is the resemblance between the constant "K" of the Ketteler dispersion formula and the value M2K of cohesion. Thus with the Ketteler formula n* = a2 -- K P + D X\/(P — X\) the constant "K," Drude found, could be computed with a fair approximation, in some cases at any rate, from the sum of the valences, the molecular weight and the density, and this result was confirmed by Erfle.4 This constant "K," therefore, contains at least 1 Cotter, J. R.: "Dispersion," Encyclo. Brit., nth edition, 8, 317. 2 Traube: Ber. chem. Ges. Berlin, 40, 130 (1907). 3 Cuthbertson: Proc. Roy. Soc., 8aA (1909-1910); Phil. Mag., [6] 21, 69 (1911); Phil. Trans., 204, 323 (1905); 207, 135 (1907). 4 Erfle: "Optische Eigenschaften und Elektronen Theprie." Annalen der Physik, [4] 24 (1907). Relationship between Molecular Cohesion 485 sometimes the same factors as the value a/V2 of van der Waals' equation. I have not, however, attempted to estab- lish any closer connection between them.1 We conclude, therefore, that while cohesion and re- fractivity are both dependent on a common factor, namely the valence electrons, and possibly upon the molecular weight, the connection between them is not direct, but indirect; and while cohesion and refraction, or dispersion, often parallel each other, they, at other times, diverge considerably since other factors enter into refraction. It is not without interest to recall as an example of the perspicacity of genius, that Laplace2 long ago foretold a con- nection between these properties. Writing in 1805 of the formula of capillarity which, as will be remembered, con- tained two terms, one K, representing molecular cohesion, or van der Waals' expression a/V; the other, H, the capillary constant, Laplace says (p. 351): "I saw that this action (pressure) is smaller or larger than if the surface is plane; smaller, if the surface is concave; larger, if it is convex. Its analytical expression is composed of two terms: the first (K), much larger than the second, expresses the action of the mass terminated by a plane surface; and I think from this term depends the suspension of mercury in a barometer tube at a height two to three times greater than that due to atmospheric pressure, the refractive powers of diaphanous bodies, the cohesion, and in general, chemical affinity; the second term expresses the part of the action due to the spheri- city of the surface." And again (p. 362) : "The function, K, is analogous to that I have designated by the same letter in the refraction of light." But of even greater interest and more fundamental importance than the relation between the optical and the cohesive properties, which is now understandable since both 1 See also Natanson: Bull, de 1'Acad. des Sci. de Cracovie, 1907, April p. 316 for the relation of refraction and valence. 2 Laplace: Sur 1'action capillaire. Oeuvres. Supp. Liv. X, Traite de Mecanique Celeste, p. 351. 486 Albert P. Mathews involve the number of valence electrons, is the relation be- tween the magnetic and cohesive properties, since here we touch, I think, the very kernel of the problem of the nature of cohesion. The connection between the magnetic properties and cohesion is brought out very clearly, in an empirical way, by Pascal's1 investigations on the relation between magnetic susceptibility and the molecular properties. In a series of papers Pascal has shown that there is a remarkable con- nection between the specific susceptibility of diamagnetic elements and the atomic weights and valences. Thus if elements of the same family having the same valence are arranged in their order of increasing specific susceptibility, they are in the order of their atomic weights; and, on the other hand, a close dependence on valence may be observed. In elements of nearly the same weight but of different valence numbers, the atomic magnetic susceptibility, Xa, increases with the number of valences. An empirical relationship was found between them, Xa — io~7ea + l*a where a is about 2.1, /? about 0.004 and a the atomic weight, a and p depend only on valence. But here, also, double bonds made their effect felt, though less pronouncedly than in refraction. Double bonded molecules have, in general, a lower molecular susceptibility than that calculated. Other effects of molecular form are seen; for example, the benzene nucleus augments the diamagnetism, although the double bonds it is supposed to contain should have an opposite effect. The factor of molecular structure appears, then, to influence diamagnetism. On the whole, however, calculation of the number of valences in the molecule by this procedure agrees better with the calcu- lation from the cohesion, than does the calculation from the refractivity or dispersion. Thus, double bonds are of less action on the diamagnetism and in the benzene nucleus they 1 Pascal : "Sur im mode de contr61e optique des analyses magne" tochimiques, ' ' Comptes rendus, 152, 1852 (1911). Recherches magne to-chimiques sur la struc- ture atomique des halogenes," Comptes rendus, 152, 826 (1911). Ann. Chim. Phys., [8] 19, 1-80 (1910). Relationship between Molecular Cohesion 487 appear to exert no effect. By this method, as by the cohesional, all organic chlorine compounds examined were found to have trivalent chlorine and fluorine was monovalent. The agree- ment was good in regard to other elements also. The connection between cohesion and diamagnetism is, therefore, again an indirect one. Both involve the molecular weight and the number of valences, but it is clear that cohesion is independent of molecular form, or very largely so; whereas whether a substance is magnetic, or diamagnetic, may depend, in part upon this very factor. Oxygen in an elemental form, is paramagnetic, not diamagnetic, and is quite anomalous in Pascal's scheme; whereas the cohesion of oxygen is not anomal- ous. In other words, whether a body is, as a whole, para- magnetic or diamagnetic, and to what degree, depends, prob- ably, on the possibility of the orientation of the molecules, their polarity, etc., factors which do not seem to affect their cohesion or gravitation. Nevertheless cohesion and magnetic properties are, no doubt, closely related, since both depend on the same molecular properties, only magnetism involves still other properties, (form) not involved in cohesion. The fact that cohesion is thus determined fry the number of electron couples (atomic and valence) in the molecule plainly points toward the conclusion that cohesion is either electro-static or electro-magnetic in nature. Both of these possibilities have already been suggested. Sutherland,1 from his discovery of the relation of cohesion to valence in salts, inferred at first that the cohesion must be of an electromagnetic nature. He supposed these rotating electron couples acted like little magnets, and he attempted to show, though whether successfully, or not, I am unable to judge, that small magnets, at sufficient distances apart, would attract inversely as the fourth power of the distance between them, and this he sup- posed to be the law of molecular attraction. This conclusion was attacked by van der Waals, Jr.,2 who concluded, also 1 Sutherland: Phil. Mag., [6] 19, i (1910); 4, 625 (1902). 2 Van der Waals, Jr. : Kon. Akad. v. Wetensch. te Amsterdam, Pro- ceedings, n, 132 (1908-1909). 488 Albert P. Mathews from mathematical reasoning, that the attraction between such magnets would be inversely as the yth or 9th power of the distance, and thus agree with the assumptions of his father, that molecular attraction diminished at such a rate that it was effective only when the molecules were in contact. lyater, Sutherland1 concluded that cohesional attraction was due to electrostatic affinity of these electron couples. Lodge2 made a similar suggestion. He thought some of the lines of force between the atoms wandered outside the molecule to atoms of other molecules and thus produced molecular cohesion. This would make molecular cohesion of the same nature as chemical affinity. While the relation between the two is close, both being zero in the absence of an electric charge, or valence, one is not causally dependent on the other, although both depend on the valences. The attraction between the atoms is probably of an electrostatic kind and, if so, should vary inversely as the square of the distance. The atomic weight does not appear to play a part in chemical affinity, for very light elements may enter into very firm union. In cohesion, molecular weight does play a large part; and while it is not impossible that the cohesional attraction may be inversely as the square of the distance, it is not probable, or at any rate it has not yet been proved to general satisfaction. It seems much more probable to me that cohesion is more closely related to magnetism than to electrostatic affinity and I would raise the question whether magnetism is anything else than molecular cohesion made apparent at distances more than molecular. Is it not possible that molecular cohe- sion, involving as it does both atomic and valence electrons (atomic weight and valence) is due, perhaps, to the magnetic effects produced by the movements of these electron couples? In this view the atoms would be united by their electro- static affinities and these same valences and the other atomic electrons by their magnetic effects produce the molecular 1 Sutherland: Phil. Mag., [6] 17, 667 (1909). 2 Lodge: Nature, 70, 176 (1904). Relationship between Molecular Cohesion 489 cohesion. I may state briefly some of the reasons which ap- pear to lead to such a conclusion. In the first place we have the surprising fact that the field of cohesion of a molecule is apparently delimited by the surrounding molecules. The evidence for this, while perhaps not conclusive, is both direct and indirect. The reason for the shortness of the radius of action of cohesion is one of the most interesting questions of molecular physics. It is of interest to see how this question came to be generally considered closed and settled in favor of the view, now gener- ally accepted, that cohesional attraction diminishes with the distance at a rate far greater than gravitational attraction. It is chiefly due to Laplace. Laplace,1 in his beautiful memoir on capillarity, first raised the question whether the short radius of attraction of the cohesive forces was due to the fact that matter shut off the attraction, or was due to the attraction diminishing with the distance at a rate far more rapid than gravitation. He says, when discussing Hawksbee's well-known ex- periments proving that the height to which water rises in a glass tube is independent of the thickness of the wall of the tube: "Hawksbee a observe que dans les tubes de verre, ou tres minces ou tres epais, 1'eau s'elevait a la meme hauteur toutes les fois que les diametres interieurs etaient les memes. Les couches cylindriques du verre qui sont a une distance sensible de la surface interieure ne contribuent done point a 1' ascension de 1'eau, quoique dans chacune d'elles, prise separement, ce fluide doive s'elever au-dessus du niveau. Ce n'est point 1'interposition des couches qu'elles embrassent qui arrete leur action sur 1'eau, car il est naturel de penser que les attractions capillaires se transmettent a tracers les corps, ainsi que la pesanteur; cette action ne disparait done qu'a raison de la distance du fluide a ces couches, d'ou il suit que V attraction du uerre sur Veau n'est sensible qu'a des distances insensibles ." I have italicized the end of Laplace's statement to bring out 1 Laplace: "Sur 1'action capillaire. Oeuvres. Supp. au Livre X," Traite de Mecanique Celeste, p. 351; see also p. 487. 490 Albert P. Mathews clearly the reason which led him to the conclusion that molecu- lar cohesion penetrated matter like gravitation, and that the attraction must, hence, decrease very rapidly with the dis- tance. It will be seen that the sole reason for his decision was the possible analogy between gravitation and molecular cohesion. The very important result of the rejection by" Laplace of the possibility of cohesive attraction not penetrating matter was that it forced him to the conclusion that the cohesive force must diminish far more rapidly than gravitation as the distance increases. Laplace did not make any assumption as to the rate at which the cohesional attraction diminished with the distance, except that it was at so rapid a rate that the cohesion became negligible within all measurable dis- tances. The great English philosopher-, Thomas Young,1 who a year before Laplace had shown the true nature of surface tension and practically anticipated all of Laplace's main conclusions, does not appear to have raised the question in a concrete form. His papers on capillarity are so condensed that the reasoning is very difficult to follow.2 But while Young nowhere specifically puts the question whether the cohesional attraction penetrates matter, he made an assumption which might be taken to indicate that it does not. ''We may suppose," he says (p. 43), "the parti- cles of liquids, and probably those of solids also, to possess that power of repulsion which has been demonstratively shown by Newton to exist in aeriform fluids, and which varies as the simple inverse ratio of the distance of the particles from each other. In air and vapors this force appears to act 1 Young: "An Essay on the Cohesion of Fluids," Phil. Trans., 1805 (collected works, edited by G. Peacock, i, 418 (1855), London). 2 A propos of this paper of Young's, Clerk Maxwell makes an interesting comment. He says: "His [Young's] essay contains the solution of a great number of cases including most of those afterwards solved by Laplace; but his methods of demonstration, though always correct and often extremely elegant, are sometimes rendered obscure by his scrupulous avoidance of mathematical symbols." Ency. Brit., Article, "Capillarity." Relationship between Molecular Cohesion 491 uncontrolled; but in liquids it is overcome by a cohesive force, while the particles still retain a power of moving freely in all directions; and in solids the same cohesion is accompanied by a stronger or weaker resistance to all lateral motion, which is perfectly independent of the cohesive force and which must be cautiously distinguished from it." "It. is sufficient to suppose the force of cohesion nearly or perfectly constant in its magnitude throughout the minute distance to which it extends, and owing its apparent diversity to the contrary action of the repulsive force, which varies with the distance. Now, in the internal parts of a liquid, these forces hold each other in a perfect equilibrium, the particles being brought so near that the repulsion becomes precisely equal to the cohesive force that urges them together," etc. Young thus assumed that the cohesion extended but a short distance, with slight variation in intensity and that it then ended abruptly. So far as I can find, he made no suggestion how it came to end abruptly; but if it be assumed that it does not penetrate matter, it is seen that it must end abruptly at the next layer of molecules. Young tried to esti- mate how far the cohesive force really extended, and found a value surprisingly near the order of magnitude of that now known to be the distance apart of the centers of two molecules. His reasoning on this point is extremely ingenious, and is of interest as the first estimate of molecular dimensions. Lord Rayleigh1 says anent this computation of Young's: "One of the most remarkable features of Young's treatise is his estimate of the range "a" of the attractive force on the basis of the relation T = Y3aK. Never once have I seen it alluded to, and it is, I believe, generally supposed that the first attempt of this kind is not more than twenty years old. It detracts nothing from the merit of this wonderful specu- lation that a more precise calculation does not verify the numerical coefficient in Young's equation. The point is 1 Rayleigh: "On the Theory of Surface Forces," Phil. Mag., [5] 30, 285-298, 456-475 (1890). Collected papers, 3i 396. 492 Albert P. Mathews that the range of the cohesive forces is necessarily of the order T/K." T is the surface tension and K the internal pressure. Lord Rayleigh, in his revision of Maxwell's classical account of capillarity in the new edition of the Encyclopaedia Britannica and in his many splendid writings on this subject, does not seem .to have considered this question. All other writers whom I have consulted seem to have followed Laplace's lead and assumed, without evidence, that cohesion does penetrate matter like gravitation. Thus, Gauss,1 who introduced clear ideas of surface energy and the potential energy of fluids, writes as follows in 1830: "The ordinary attraction which is proportional to the square of the distance, and which permits the repre- sentation of all motions in the heavens with such good agree- ment, can be used in the explanation neither of capillary phenomena nor of adhesion and cohesion; a correctly carried out computation shows 'dass eine nach diesem Gesetze wirk- ende Anziehung eines beliebigen Korpers der zur Ausfiihrung von Experimenten geeignet ist, d. h., dessen Masse im Ver- gleich mit der der Erde vernachlassigt werden kann, auf einem beliebig gelegenen, sogar den Korper beruhrenden Punkt, im Vergleich mit der Schwere verschwinden muss. Wir schliessen hieraus dass jenes Anziehungsgesetz in den kleinsten Abstanden mit der Wahrheit nicht mehr uberein- stimmt, sondern dass es eine Mgdification erfordert. ' In other words, the particles of the body exert, besides the at- tractive force of gravitations, still another force which is notice- able only in the smallest distances. All appearances show uniformly that the second part of the attractive force (the molecular attraction) is not noticeable in the smallest meas- urable distances. On the other hand, in unmeasurably small distances it may greatly surpass the first, which is propor- tioned to the square of the distance. 1 Gauss: "Allgemeine Grundlagen einer Theorie der Gestalt von Fliis- sigkeiten," Ostwald's "Klassiker der exakten Wissenschaften," No. 135, p. i. "Corrrmentationes societatis Regiae Scientiarium Gottingensis Recentiores," vii, 1830. Relationship between Molecular Cohesion 493 It was necessary for him (p. 21), to make some assump- tions regarding the cohesional attraction which he represented as "/" (r), r being the radius of molecular action, and he accordingly adopted Laplace's view that / (r) decreases far more rapidly than i/r2, which is the law of gravitational attraction (p. 22). He says, speaking of cohesional attrac- tion, or / (r) : "Da dieser Ausdruck etwas unbestimmtes hat so lange wir nicht eine Einheit zu Grunde legen, wollen wir vor allem darauf aufmerksam machen, dass wir die anzie- hende Kraft / (r), ausgedruckt als eine Function des Abstandes r, mit einer Masse multipliziert denken mussen, damit sie mit der Gravitation g in den Dimensionen ubereinstimmt. Der Sinn unserer Voraussetzung ist dann der folgende : Bezeich- net M irgend eine Masse derart, wie sie uns in Experimenten vorkommt, namlich eine, die im Vergleich mit der ganzen Erde als verschwindend angesehen werden kann, dann muss M / (r) immer merklich sein im Vergleich mit der Schwere, so lange r einen unseren Messungen zuganglichen, wenn auch noch so kleinen Wert hat." Van der Waals, in all his earlier writings, including his famous essay on the continuity of the gaseous and liquid states published in 1869, assumed with Laplace that molec- ular cohesion was appreciable only very close to the molecule; indeed, the radius of action was less than the mean distance apart of the molecular centers. I have not been able to find in his essays any specific discussion of the question whether cohesional attraction penetrates matter, although he does discuss the radius of attraction of a molecule.1 His general assumption was that the cohesion diminished very rapidly as the molecules separated. In one brief communication to the Amsterdam Academy of Sciences (1893-94, pp. 20-21) he states that he had derived a potential function, that is, a function expressing the potential energy of attraction of two 1 Van der Waals: "Bijdrage tot de Kennis van de Wet der Overeenstem- mende Toestanden," Verhandelingen Konik. Akad. van Wettenschappen, 21, 5 (1881). 494 Albert P. Mathews r molecules, of the form — / e * , as the potential of two mass , r points; and he goes on to say that this formula was based on two assumptions, namely that the attraction of two molecules is inversely as the square of the distance, and second, that the universal medium absorbs the lines of force. He thus assumes an absorption by the ether of the attraction, rather than by a molecule at a distance "r." In his paper published in 1903, in which the real molecular volume, the value of "6," is no longer considered constant and in which he has revised his formula for the isotherm in so important a manner, he does not specifically reraise the question of the penetration of matter by the cohesive attraction. In a succession of papers like those of van der Waals' which represent a progressive succession of ideas, mutually conflicting ideas may, not unnaturally, be found. Thus, in his paper on the thermodynamic potential and capillarity,1 a limiting intermediate layer of rapidly changing density is supposed to exist between the saturated vapor and the liquid, and the existence of such a layer would seem to the writer to presuppose that the radius of attraction at least in this layer must be several molecular diameters. On the other hand, the following statement2 (p. 121) is not entirely reconcilable with this view, and would be so only if the layers of which he speaks are at least equal in thickness to the distance the cohesive force extends. He supposes the cohesive pressure to be exerted only by the surface layer, but he states that exactly the same formulas are obtained if the fluid be considered to be made up of a series of layers of molecular dimensions. "If we consider the gas in a cylindrical vessel of constant area and divided into horizontal layers, the lowest attracts the next higher," etc. The sum of all partial amounts of work will be the same as if one considered 1 Van der Waals: "Theorie thermodynamique de la Capillarite," Archives Neerlandaises des Sciences exactes et naturelles, 28, 121 (1895). 2 Van der Waals: "Die Continuitat des gasformigen u. flussigen Zustandes," Leipzig, p. 126 (1899). Relationship between Molecular Cohesion 495 only the attraction of the upper layer and the distance as that through which this layer would have been moved. It seems to me that, for this reasoning to be correct, we must assume the layers at least as thick as the radius of action of each layer of molecules. If these layers are of molecular diameters, it would seem that the cohesive force does not extend farther than a molecular diameter. This is not apparently con- sistent with the assumption elsewhere made to explain the transitional layer between vapor and liquid. Plateau also followed Laplace. Sutherland, in his many papers on molecular cohesion, assumes that the cohesional attraction is inversely as the fourth power of the distance, but he does not, so far as I can find, discuss the question of the penetration of matter by cohesional attraction. But it is impossible for cohesion to vary inversely as the fourth power, if cohesion penetrates matter. Kleeman1 supposes that the attraction must be in- versely as the 5th or some higher power of the distance. He has pointed out that cohesion cannot possibly be assumed to vary like gravitation inversely with the square of the distance, because the cohesional attraction is so much greater than gravitation. But Mills weakened the force of this objection, which he had overlooked in his early papers, by making the additional postulate that cohesion does not penetrate matter. For if it be assumed that the cohesion does not penetrate matter then only the layer of superficial molecules of two masses would attract each other; and the number of these is so small, compared to the whole number of molecules in the mass, that the cohesional attraction would be less perceptible at sensible distances than gravitational attraction. I have been unable, then, to find any evidence for the assumption so generally made that cohesive attraction pene- trates matter like gravitation. There is no direct evidence, therefore, so far as I can find, against the inference necessitated 1 Kleeman: "An Investigation of the Determination of the Law of Chemical Attraction between Atoms from Physical Data," Phil. Mag., [6] 21, 83 (1911). 496 Albert P. Mathews by the square or fourth power law of attraction, that cohesion does not penetrate matter. There is, on the other hand, some evidence of a direct kind that cohesional attraction extends only as far as the nearest molecules. This evidence is the length of the radius of action as determined by direct measurement, and Einstein's proof that the radius of action varies with the distance apart of the molecular centers. Laplace believed that the radius of action, although short, nevertheless extended many molecular diameters and this opinion prevailed until recently, but as means of measure- ment have improved the radius has shrunk. Quincke gave an estimate of about 6 X io~6 cm, but the most recent deter- minations of Johannot,1 and Chamberlain2 show it to be about 1.6-2 X io~7 cm in a soap film and in the case of glass. The diameter of a molecule of trioleate of glycerine, according to Perrin,3 is i.i X io~7 cm. The radius of action is certainly not more than two molecular diameters and indeed is hardly more than one. The average distance between the centers of two molecules of ether in the liquid state at 20° is about 5.5 X io-8cm. But not only has direct measurement shown the radius of action to be one or two molecular diameters, but com- putations of it by Kleeman make it very close to this. For example, in ether Kleeman4 computed the radius to be about 3.4 X io~8 cm which is about the distance when the mole- cules are in contact. Van der Waals supposed it, indeed, to be only as long as this. Recently Einstein5 has made a very interesting computation starting from the law of Eotvos, by which he shows, from thermodynamic reasoning, that -the 1 Johannot: Phil. Mag., [5] 47, 501 (1899). 2 Chamberlain: Phys. Rev., 31, 170 (1910). 3 Devaux: Journal de Physique, [5] 2, 699 (1912). 4 Kleeman: "On the Radius of the Sphere of Action," Phil. Mag., [6] 19, 840 (1910). 8 Einstein: "Bemerkung zu dem Gesetz von Eotvos," Annalen der Physik., [4] 34, 165 (19-11). Relationship between Molecular Cohesion 497 range of cohesive action must be of the general value of, and proportional to, the distance between the molecular centers. This distance is of the order of magnitude in most liquids of io~8 cm. This result is so surprising that Einstein says of it: "This result appears at first very unlikely, for what should the radius of action of a molecule have to do with the distance between neighboring molecules? The supposi- tion is only reasonable in case the neighboring molecules alone attract each other, but not those farther removed. " Sutherland1 also came to the conclusion that the radius of action was about equal to i//3. We see then, that the evi- dence points to the conclusion that the radius of action of the molecular forces agrees very closely with v1/8, the distance between the molecular centers, and varies directly with -z//3. The only explanation of this fact appears to me to be that the attraction does not extend beyond neighboring molecules and, hence, must, in some way, be stopped by them. Mills2 alone, so far as I can find, has reopened the question whether the field is delimited by the surrounding molecules. Concluding, I believe erroneously, from an empirical law, that the attraction between molecules must vary inversely as the square of the distance, he was driven to the second conclusion that if this were the case the cohesion could not penetrate matter. He assumed, hence, that the surrounding molecules absorbed, or neutralized, the lines of cohesive force. The direct evidence and Einstein's reasoning leaves little doubt that the radius varies with the distance apart of the molecules and the only possible conclusion from this is that the surrounding molecules delimit the field as Mills supposes. If it is a fact that the surrounding molecules delimit the field as the evidence indicates, cohesion is allied at once with magnetism, for this is the very supposition which Ewing made to explain some of the phenomena of magnetism. Each molecule of a ferromagnetic substance is supposed to be a 1 Sutherland: Phil. Mag., [6] 4, 632, 636 (1902). 2 Mills: Jour. Phys. Chem., 15, 417 (19* 0- 498 Albert P. Mathcivs magnet. In the non-magnetic state the magnetism of each molecule is supposed to be neutralized by the surrounding molecules, which have their magnetic axes variously directed. The magnetic field is thus limited to a single molecular diam- eter and will vary with the distance apart of the molecules. The magnetic field of each molecule is delimited by the sur- rounding molecules. If, however, these molecules are oriented, either by acting on each ether, or by external forces, then the magnetic fields coincide and the magnetism may be perceived extending outward from the mass. In other wrords, to explain why magnetism does not persist in soft iron and to account for magnetic hysteresis, Ewing has made exactly the same assumption which has been made to explain why cohesion does not extend beyond molecular distances. It seems to me not impossible that magnetism is the cohesive attraction of a molecule. If the molecule is of such a nature, or shape, that the total effect of the little magnets, its electron couples, coincide more or less completely so that the molecule has a polarity, and the molecules can be oriented in any way and held in position, we have the ferro and para- magnetic substances. If the molecules have many poles, so that there is no polarity of the molecules as a whole, there is a diamagnetic substance. The magnetic field of each molecule would be its cohesive field. The recent work of Cotton and Mouton1 on the orienta- tion of molecules in the magnetic field seems to me to bear out such an interpretation. The work of Weiss2 and Langevin3 appears to point in the same direction, but Weiss, who has considered the possibility of the identity of cohesion and magnetism, states that he will shortly show that they can not be identical. Nevertheless it appears to me not impossible that magnetism is simply a special case of cohesion, and if this is true the rotation of the plane of polarized light by op- tically active substances would be easily understood. 1 Cotton and Mouton: Journal de Physique, [5] I, 40 (1911). 2 Weiss: Journal de Physique, [5] i, 900 (1911); [4] 6, 661 (1907). 3 Langevin: Ann. Chim. Phys., 5, 70 (1905). Relationship between Molecular Cohesion 499 Finally, if it is true that the surrounding molecules de- limit the field of cohesion in any way whatever, and that only six molecules really take part in this delimitation, we can derive the value a/V2 of van der Waals' at once and very simply. Suppose that each molecule has a certain mass of cohes- ion, M, and that two molecules attract each other directly as the product of their cohesive masses and inversely as the fourth power of the distance between their centers, then the attraction between two molecules would be M2K/i>4/3, where D is the space at the disposal of a single molecule. Since each molecule attracts only the one above, below and to each side, the pressure per square centimeter of a double layer of molecules will be M2K/V/3 multiplied by the number of mole- cules in i sq. cm or i/V'/3, making M2K/V. Since the at- traction extends only a single molecular diameter, we may multiply both numerator and denominator by N2, where N is the number of molecules in the volume, V, and we obtain, N2M2K/NV == M2N2K/V2 =<* a/V2. M2K has already been shown to be 2.98 X io~37 (Mol. Wt. X Valences)2/3. We have, as yet, no proof that only the six surrounding molecules are attracted, although Sutherland1 has made a similar supposition. But such may be the case nevertheless. Einstein, in his calculation, computed that each molecule could be considered as lying at the center of a cube and that it attracted the 26 other molecules of the cube. Of course the fact that the value a/V2 may be so easily derived in this way does not furnish any proof that the attraction is inversely as the fourth power of the distance. But I know of no other derivation of a/V2 which involves so few, or less radical, assumptions. The general conclusion of the paper is then, that cohesion, 1 Sutherland: Phil. Mag., [6] 17, 667 (1909). "The total potential energy of a number of like molecules is the same as if each caused its own domain to be uniformly electrized with an electric moment proportional to the linear dimensions of the domain, the direction of electrization being such that in general any molecule attracts its six immediate neighbors." 500 Albert P. Mathews being a function of molecular weight and valence, is a function of the number of electron couples of the valences and atoms, and is, hence, probably of a magnetic nature. Magnetic substances may be supposed to be substances in which, owing to the orientation of the molecules, or to their polarity, or both these causes, the cohesive fields of the molecules are not delimited or neutralized by the surrounding molecules so that the cohesional attraction becomes apparent at more than molecular distances, and under such circumstances the sub- stance is said to be magnetic. University of Chicago THE INTERNAL PRESSURES OF UQUIDS BY ALBERT p. MATHEWS The fundamental significance of the constant "a" of van der Waals' equation makes its exact determination important. Various methods have been proposed for the determination of this constant, but none of them are entirely satisfactory. In a recent paper1 attention was drawn to a method by which it could be determined from the surface tension by the use of Thomas Young's formula, T = rK/3, combined with the law of Ramsay and Shields, and the values thus computed were shown to be closely similar for most substances to the values computed by van der Waals' method from the critical temperature and pressure, but in some cases they deviated considerably from his values. I found, also, that the values of the constant "a" obtained by this method were simple functions of the products of the molecular weight and the number of valences in the molecule and that they could be computed from these values. Inasmuch as the method used in computing "a" from the surface tension involved the value of the density at absolute zero, which was computed from Cailletet and Mathias' law of the rectilinear diameter, and involved, therefore, some uncertainty and was certainly too high, it was desirable to find a method of computing "a" directly from the surface-tension measurements. The desire of finding such a method was stimulated by the present great uncertainty of the value of the internal pressures of liquids. Traube2 has within the past few years computed the internal pressure for many liquids, but the results he has obtained are, in my opinion, unreliable because his method of computation involves the use of the value "&" the real molec- ular volume or co-volume, a very doubtful value. The values he finds for the internal pressure at zero degrees are, also, widely different from those computed by the use of ' V found 1 Mathews: Jour. Phys. Chem., 17, 154 (1913). 2 Traube: Zeit. phys. Chem., 68, 291 (1909). 604 Albert P. Mathews at the critical temperature; and in some cases they are not more than half those calculated recently by Lewis1 from the latent heats of expansion of liquids. Walden,2 also, has recently calculated the value of "a" and the internal pressure from the surface tension. His calculation is, however, almost wholly empirical. It is based, first, on Stefan's conclusion3 that it takes one- half the work to move a particle into the surface which is required to carry it all the way to the vapor; and, second, upon an empirical relationship found by Walden between the surface tension and the molecular latent heat at the boiling point. It is, however, by no means certain that Stefan's conclusions are correct and his reasoning does not carry con- viction. There is, also, probably an error in the assumption that the molecules do not change in size on passing from the liquid to the vapor and that the latent heat of vaporization represents only the work done in overcoming molecular co- hesion. Finally Walden's values for "a" resemble Traube's. "a" is always much less than when computed by van der Waals' method from the critical data and much less than the values of Lewis. The values which Walden has obtained are about two-thirds the values given in this paper. Values still smaller have been computed by Davies4 from the latent heat of vaporization. The values he obtains are only about one- third those of Lewis. Winther5 has still other results. Many modifications of van der Waals' equation have been proposed in which "a" was considered variable with the tem- perature and "6" more, or less, constant. These attempts have not been fruitful. It is far more probable that "6," the volume correction, varies with temperature and volume and that "a," which is the ''mass" factor of the cohesion, is 1 Lewis: Phil. Mag., [6] 25, 61 (1912). 2 Walden: Zeit. phys. Chem., 66, 385 (1909). 3 Stefan: Wied. Ann., 29, 655 (1886). 4 Davies: Phil. Mag., [6] 24, 422 (1912). 5 Winther: Zeit. phys. Chem., 60, 603 (1907). The Internal Pressures of Liquids 605 constant, "a," indeed, as van der Waal's has repeatedly shown, should be considered constant unless association, or quasi-association, occurs, "a" contains the factor N2, N being the number of molecules in the volume, V, hence any association will lower "a" by this factor. The wide divergence of these various values proposed for the internal pressure is shown in Table i expressing the internal pressure in atmospheres at zero degrees, except in the case of Walden where the values are for the boiling points and Davies fori50C. i Substance Davies 15° Traube 0° Walden b. p. v. d. Waals 0° Lewis 0° Winther Benzene I 102 1380 1570 2494 2639 1792 Toluene 1188 1180 1340 2228 2847 — Cymene 661 — — — 2718 — Ether 778.9 990 |l2IO( I723 1932 I22O CC14 1076 1305 1490 2205 2518 1680 CS2 1683 1980 2170 3363 2917 Et acetate — 22IO ^1280^ (1340) 2466 I486 In this paper are given the values of "a" obtained in several quite different ways, all of which yield closely agreeing results. 1. The first method is a computation from the surface tension. The assumption involved in this method is the depth of the surface film expressed in the number of molecular layers. That the assumption is correct is proved by the out- come. 2. The second method is a modification of Thomas Young's method combined with the law of Eotvos as developed in my former paper, but with certain corrections. 3. In the third method "a" is computed from van der Waals' equation at the critical temperature, the assumption being made that in all normal substances bc -• = 2V 0 == 2VC/S, S being the critical coefficient and equal to RTC/VCPC. 4. In the fourth method "a" is computed from the internal 606 Albert P. Mathews latent heat of vaporization close to the critical temperature. The only assumption made here is that Mills' or Dieterici's formula for the internal latent heat is more correct close to the critical temperature than the internal latent heat computed by Biot's formula. 5. Finally "a" is computed from the number of valences and the molecular weight by the formula: a = C(M X Val)2/3. 1. Computation of "a" from the Surface Tension The surface film is determined by the difference of cohesive attraction in the vapor and liquid. The surface energy must, hence, be a function of the difference in cohesive energy in the liquid and vapor, or of the expression a(i/V/ — i/VJ. This cohesive energy has been lost in passing the liquid through the surface layer which separates the two states and should be calculable from the surface tension. Suppose we have a sphere of a gram mol of a liquid in con- tact at all points with its saturated vapor. Its density is uniform except in the surface film where it decreases by a series of steps from liquid to vapor density. The surface film may be conceived as a series of concentric shells, each a molecular diameter thick, and each outer one less dense than the inner until the state of uniform vapor density is reached. Furthermore, on passing from one of these molecular shells lying within to the one next beyond it, always the same amount of cohesive energy will be lost, if the density diminishes uni- formly from shell to shell. The surface tension T, is the tension along a line i cm. in length in the surface film and one molecular layer deep. It represents the force necessary to stretch the surface, that is to rupture one of these shells, and thus drag one, or more, molecules from the inner core of uniform density into the first concentric shell; and of course the force necessary to drag molecules from the first shell to the second, and from the second to the third and so on, since the same force is necessary to drag molecules from each inner shell to the next outer. This tension, T, is numerically equal, also, to the surface tension The Internal Pressures of Liquids 607 / energy per square cm. of the surface film. For each con- centric shell of the surface one molecular diameter deep, the surface energy is, then, T times the surface, or TV*73; and the total energy in the surface or a, will be equal to this amount multiplied by the number of shells, which we shall represent by the letter n, thus we have the equation : (1) a = TVl/3n If now a gram mol of liquid passes from liquid to vapor it must pass through this surface in which surface energy is gained at the expense of the cohesive energy which is lost. To determine the total amount of energy which is thus changed in passing the whole gram mol through the surface, it is only necessary to determine how many times we shall have to make a new surface shell until the whole of the gram mol has passed through. Since the total number of molecules in the gram mol is N, and there are in a surface shell N2/3 molecules, we shall have to make a new surface N/N2/3 times, or N1/3 times. The total energy then which will be gained as surface tension energy will be S, or (2) S = TVY*NY*n. This same value may be obtained, also, in the following way: The surface tension measures the force necessary to overcome the surface tension pressure along a line i cm long and a molecular diameter deep. The surface-tension pressure per sq. cm acting in the plane of the surface is hence T/v1/3, i) being the volume of a single molecule. Multiplying numera- tor and denominator by N1/3 we have the pressure per square cm N1/3T1/3/V;/3. If this pressure work through the volume Vt, the volume of one gram mol, we have the surface tension energy if a whole gram mol were present as a surface shell, or N1/3TV*/3. Multiplying this by the number of shells, or n, we have our former expression. Equation (2) contains the unknown factor "n," that is the number of layers one molecular diameter thick constituting the surface film. Since I knew of no way of measuring this, 6o8 Albert P. Mathews the following assumption was made based on van der Waals' conclusion that the surface film is infinitely thick at the critical temperature. At absolute zero, where the molecules are presumably in contact, it may be assumed that the surface film is only a single molecular diameter deep. At the critical temperature, on the other hand, it must be infinitely deep. That is, no matter how many layers of molecules one passes over, one can never, at that temperature, get to a region of differing density. Between absolute zero and the critical temperature the depth of the surface film must lie between these two values, increasing with the temperature, and pre- sumably in all normal substances at corresponding tempera- tures it will be the same number of molecular layers thick. I accordingly made the guess that it would be equal to (TC/(TC — T))2/3 molecular diameters since this fraction is equal to (d0/(di — dj)2 (see page 617). This guess turned out to be correct if Eotvos surface-tension figures are used, but if Ram- say and Shields' are taken the fraction must be raised to the 0.76 power and even then the value of "a" computed by this assumption runs down near the critical temperature. Since Eotvos measured the surface tension by a method which en- tirely avoided any assumption as to the angle of contact and Ramsay and Shields used the capillary method, which involves such an assumption, I believe Eotvos figures and his statement of the law is to be preferred. That his formula of TV*/3 = 2.27 (Tc — T) is to be preferred on other accounts is shown by the calculations which follow: The total surface energy gained by passing a gram mol through the surface is hence : (3) S - TV'/3N1/3(TC/ (Tc — T)) o.76 ergs ; or (4) 2 - TV;/3N1/3(TC/ (Tc — T))2/3 ergs. Formula (4) is to be preferred, when the surface tension is measured by methods which do not involve the angle of contact. We only have left to find the relation between the amount of energy thus lost and the difference in the cohesive energy The Internal Pressures of Liquids 609 in a gram mol of liquid and vapor, respectively. I think that the total surface-tension energy must be one- third of the total difference in cohesive energies in equal weights of the two phases separated by the surface. The energy in the surface is an expression of the difference in cohesive pressure in one direction only, whereas the two phases differ in their cohesion in three dimensions. That the value one- third is correct is shown by the computations which follow. The value one-third was that adopted also by Young more than a century ago, but his reasoning is so condensed that it is hard to follow. His statement is as follows i1 "Upon these grounds we may proceed to determine the actual magnitude of the contractile force derived from a given cohesion extending to a given distance. Supposing the cor- puscular attraction equable throughout the whole sphere of its action, the aggregate cohesion of the successive parts of the stratum will be represented by the ordinates of a parabolic curve; for at any distance x from the surface, the whole in- interval being a, the fluxion of the force will be as dx(a — x), since a number of particles proportional to dx will be drawn downwards by a number proportional to a, and upwards by a number proportional to x, and the whole cohesion at the given point will be expressed by ax — *2/2; and this at last becomes a2/2, which must be equal to the undiminished co- hesion in the direction of the surface. Consequently the dif- ference of the forces acting on the sides of the elementary cube will everywhere be as a2/2 — ax + x2/2 and the fluxion of the whole contractile force will be dx(a2/2 — ax + *2/2)> the fluent of which when x == a becomes a3/6, which is V3 of a xa*/2, the whole undiminished cohesion of the stratum." "We may, therefore, conclude, in general, that the contractile force is one- third of the whole cohesive force of a stratum of particles equal in thickness to the interval to which the primi- tive equable cohesion extends," or T = aK/3. Accepting this coefficient of 1/3 of Young in place of that of 1 Young, T: "Article on Cohesion," collected works, p. 460. 6 io Albert P. Mathews Y, of Stefan, or 3/20 as computed by Lord Rayleigh, we have the complete formula (5) 0(i /V, - i / V.) 1 3 -- TV:/3NI/3(Tc/ (T, — T))v>. Changing to density in place of volume on the left hand side we have (6) a - 3rv;/3N1/3M(Tc/ (Tc— T))2/3/ (^ — ^) ; or if Ramsay and Shields' surface-tension figures are used (7) a = 3^/3N1/3M(Tc/ (Tc — T))°-*V (^ _ i = CRT/wd/D. If the internal latent heat is computed using the vapor pressures computed by Biot's formula close to the critical temperature too low values are obtained as Mills has pointed out. The values for "a" for one gram mol computed from near the critical temperature are given in Table 7. Column 2 of that table shows how many degrees below the critical tempera- ture data were taken for the computation. The nearer the critical temperature the more reliable the data should be. (31) a = M(X — fi)l(dr—dv). It is interesting now to see how large /* is relative to ^ at different temperatures; that is, how much of the heat of vapor- ization is rendered latent by the expansion of the molecules, or by an increase in their rotatory energy. By formula (31) a = M(J — fi)/(dl — dv)\ and by (29) a = (S2 — S + 2)PCV'A (S — 2) . Hence we have (32) v^— ft = (S2 — S + 2)PCVX — ^)/M(S — 2). But it was found by Mills that X = C(d?/3 — — (S2 — S + 2)PcVX — dj /M(S — 2) . The calculation of /£ for one gram mol of pentane by formula (33) resulted as follows : i fi X icr10 ergs -273° 4-25 + 30 4.46 60 3-51 IOO 2.18 150 190 0.62 — 0.0026 197.2 Tc At 30°, therefore, the total latent heat for one gram is 85.76 cals; the total internal latent heat, or A, is 78.80 cals.; 622 Albert P. Mathews §" X H)nOi X = v \ vo M M Th< O N O M 10 M M 04 CM CO iO rj- O ^O i— i M O CO CM CM CM a w g X + s — 1-1 CO ^ CO COCO -< *O CM CM CM CM S/'A'Xs01 X ^o ^ o o ON co t^ O ^ O M O M ""3 CM CM The Internal Pressures of Liquids 623 CO 0 O t^^O Th ON iO t^.co !>• CO O CM GO O CM r>- ri* CO CO IH M O ^" OO CO M M OO CO 1-1 O ^O OO iO CM ^OO VO CM O ONt^ONO ONCOIO ^t- O ^ c^ ONGO T}- rj- M CM CN C4 CS CO ONOO CS CM co CO O 00 ON ON ONVO co CM •< J-J S K ? ^ *^ M OO io^O CM coo"^ • •-...- o *-< tO )/RTc/3(Tc — T)x/3. THE QUANTITY OF RESIDUAL VALENCE POSSESSED BY VARIOUS MOLECULES BY A. P. MATHEWS All, or nearly all molecules .possess some power of com- bining with molecules of the same or different kinds. This combining power is called the residual valence, or affinity, of the molecule. Thus ammonia, NH3, will unite with water or acids, a molecule of hemoglobin with oxygen, glucose and probably all salts with water when they dissolve in it, and other examples of this power of molecular union might be given. The importance of residual valence to the molecule is generally recognized. It is probable that, with the possible exception of ionic reactions, these molecular unions precede, and are a necessary condition for, most chemical interactions between molecules; for molecules do not seem to affect each other by simple contact, but only when they are united into a new molecule by chemical bonds. It would seem that it is only when they are thus united that the atoms of two mole- cules are able to interchange and undergo those rearrange- ments resulting in the birth of new molecular species. The importance of residual valence makes it desirable to know how much of it is possessed by each species of molecule. Since it is possible that this amount may not always be the same even for the same species of molecule, a method must be used in determining the average amount which will examine the molecular system without disturbing it. For example not all the molecules of carbon dioxide may be in a condition to unite with water at the same instant. Many facts indicate that of all the molecules of oxygen in the air only a few, at any instant of time, are in a condition to unite with oxidizable substances. The number possessing residual valence is small. This changing molecular condition seems analogous to, and is very possibly essentially identical with, the varying condi- tion of atoms of radium which only occasionally become Residual Valence of Various Molecules 475 radioactive and decompose. It is not impossible, on the electronic theory of valence, to ascribe the acquiring of addi- tional or residual valence by the atoms of molecules to internal rearrangements of the electrons within the atoms, similar to that rearrangement which, in radioactive substances, leads to an explosion of the atom. The method I have used to determine the quantity of residual valence is a physical one, based on the cohesion of molecules. While the method is not very accurate at present, owing to several doubtful points in the calculations, it probably places the molecules in the order in which they occur when arranged according to the amount of average residual valence they possess; and it will become more precise as the critical data are more accurately known and the cohesion or internal pressure more accurately determined. The method is as follows : The internal or cohesive pressure of a fluid is rep- resented by the value a/V2 of van der Waals' equation. In this expression V2 represents how the cohesional attraction varies with the distance; and "a" is the "mass" factor of the attraction. Now "a" includes the factor N2, N being the number of molecules in the volume V, and if "a" is divided by N2 then the quotient represents the "mass" factor of the cohesional attraction of two molecules and it may be written M2K to correspond with the mass factor of the gravitational attraction, m2k, in which "m" is the gravitational mass and "k" the gravitational constant. The relationship was found1 that M2K is proportional to the two-thirds power of the product of the molecular weight by the number of valences in the molecule, or M2K = C(Wt. X Val.)2/'. How accu- rately this relationship holds is shown in Table I in which the value of "a" which is of course proportional to M2K, and is the mean value computed from various formulas, is com- pared with the value of "a" computed from the molecular weight and the number of valences. The method of com- puting "a" was given in another paper.2 1 Mathews: Jour. Phys. Chem., 17, 181 (1913). 2 Mathews: Ibid., 17, 603 (1913)- A. P. Mathews TABLE I Comparison of "a" computed from the critical data, etc., with "a" computed from the molecular weight and valence. The figures represent dynes for gram mol amounts, multiplied by io~12 Substance Mean value of "a" a = C(Wt. X Val.)Vs (C = 1.259 X ion) Difference Actual Difference Percent H2 0.311 0.317 — O.O06 -1.89 N2 1.836 1.842 (Val. = 2) — 0.006 —0.31 02 1.984 2.014 (Val. = 2) — 0.030 -i-5 w-Pentane 22.07 21 .96 +0.1 1 +0.5 ^-Pentane 21.41 21 .96 —0-55 2 .2 w-Hexane 28.10 27.71 +0.39 + 1-38 Heptane 34-82 33-8o + i .02 + 3.06 Octane 41.94 40.17 -1.77 +4-4 Diisobutyl 40.01 40.17 -^0.16 —0.4 Ether 19.94 20.46 —0.72 —0.32 Benzene 21-95 22. 19 —0.24 — o. ii Chlor-benzene 23.26 22.94 +0.32 +0.14 Toluene 28.12 27.97 +0.15 +0.50 Metaxylene 34.08 34.06 0.02 +0.06 Methyl formate 12.04 12.25 — 0.21 —1-7 Methyl acetate 17. 12 17.42 —0.30 —i . i Methyl propionate 22.49 22.96 —0.47 2.0 Ethyl acetate 22.84 22.96 — 0. 12 —0-5 Propyl formate 23-I3 22.96 + 0.17 +0.72 Methyl butyrate 29 . 53 28.84 +0.69 +2.4 Methyl iso-butyrate 28.27 28.84 --0.60 — 2. I Propyl acetate ; 28.95 28.84 + 0.08 +0.3 Ethyl propionate 28.92 28.84 + 0.05 + 0.2 SnCl4 32.58 32. 58 (Val. = 16) ^O.OO =4=0.00 From the equation a = C(Wt. X Val.)2/8, if M2K, or "a," can be determined, and if C and the molecular weight are known, the total number of valences in the molecule can be calculated. If from this total number of valences there be subtracted the number which is known to exist in the mole- cule on the basis that hydrogen is univalent, oxygen bivalent, carbon quadrivalent and so on, the remainder may be sup- posed to constitute the average amount, or number, of extra or residual valences which the molecule possesses. It is this number which has been determined in this paper. Residual Valence of Various Molecules 477 Granting that this number really represents the residual valence, the accuracy with which it can be determined will depend on the accuracy with which M2K, C, the molecular weight and the number of valences reaching between the atoms can be determined. The molecular weight may be assumed to be normal for non-associating substances at the critical temperature; and we have to assume that the number of valences between the atoms are those which chemists usually assign to these elements. The determination of the constant C, however, is more difficult. It could be deter- mined empirically if M2K was known accurately for any substance of which the valence is fixed and certain. Hydrogen is the only element with an unchanging valence, but un- fortunately the critical pressure of hydrogen is uncertain. Moreover, it cannot be assumed that the valence of hydrogen is exactly unity. There are several indications that a mole- cule of hydrogen has some residual valence although it is certainly small in amount. One such indication is the solu- bility of hydrogen in water. At the same pressure and tem- perature more molecules of hydrogen dissolve in water than of helium, and hydrogen is half as soluble as nitrogen which al- most certainly has residual valence. If solubility involves residual valence, as it may, this means that hydrogen would have some residual valence. Its solubility in platinum may be interpreted in the same sense. The catalytic reducing action of platinum, nickel or other metals or metallic oxides in a hydrogen atmosphere is interpreted by Sabatier1 to mean that a chemical union of the reacting substances occurs. Armstrong,2 too, has expressed the opinion that hydrogen has a small amount of residual valence. For these reasons we cannot accurately determine C from hydrogen. Never- theless I have calculated M2K and C for hydrogen to show what the value of C would be if hydrogen were univalent. 1 Sabatier: Nobel Prize Address, p. 9, et seq.; Les Prix Nobel en 1912; La Methode d'Hydrogenation directe par Catalyse. 2 Armstrong: "Valency," Encyclopaedia Britannica, nth Ed., 27, 848 (1911). 478 A. P. Mathews Using the critical data of Tc = 32.3 and P0 13 atmospheres as found by Olszewski; and dc as 0.033 as determined by Dewar, the critical coefficient S, where S = RTC/PCV0 is equal to 2.860. This value of S is, however, much lower than that of any other substance. Thus for O2, S is 3.5; for N2, 3.6; for N2O, 3.4 and even for helium it is 3.13 according to Onnes. It is probable then that 2.86 is too low. The critical temperature and pressure have recently been determined by Buller1 who finds Tc = 31.95; Pc = n. With these values and dc = 0.033 S computes 3.903, which is higher even than such complex substances as octane and is clearly too high. If S is calculated by my formula which gives a result within 1-2 percent in most cases, namely, S2 = R(Wi - - d^T^3/- (Tc — T)1/3MP0 using Buller's values for Pc and Tc and taking d\ as that of hydrogen at the melting point [ — 258.9° (Travers)] as 0.086 (Dewar) and disregarding the value of dv at that temperature then S is computed as 3.517. If Pc were 11.5 atmospheres, and the uncertainty is at least half an atmosphere, S would be 3.365. I believe S of hydrogen may be taken as approximately that of oxygen or as 3.4. If S is assumed to be 3.4 then with Pc n and dc 0.033, a = (S2 — S + 2)/(S — 2) PCV2C = 0.301 X io12. Dividing this by N2 where N = 6.062 X io23, M2K for hydrogen would be 8.191 X io~37 from which C is found to be 3-23 X io~37. If, however, S = 3.4 and Pc 13, then "a" would be 0.317 X io12, M2K, 8.626 X io- 37 and C would be 3-45 X io~37. Since from Buller's results it is probable that Pc should be lower than 13 atmospheres, the value for M2K is probably not far from correct so that C should be very nearly 3.23 X io~37. Another simple substance from which a calculation of C might be made is methane, since the amount of its residual valence is certainly small and the total number of valences per molecule is very nearly 8. Unfortunately the critical density of methane is unknown and Tc and Pc have not been recently determined. However, if Tc is 191.2 and Pc is 54.9 1 Buller: Phys. Zeit., 14, 860-2 (1913). Residual Valence of Various Molecules 479 (Olszewski) and if S be calculated by the formula already given using the value of the density of the liquid at — 164° as 0.466 and disregarding the vapor density, S computes to be 3.318 which is probably not far wrong. From this "a" is found by the formula given above to be 3.025 X io12, M2K is 8.232 X io~36 and C is found to be 3.24 X io~37 which is al- most the same value as that computed from hydrogen. The critical data of oxygen are known, but the calculation shows clearly that oxygen is monovalent in the elemental form, there being but two valences in the molecule. This un- expected conclusion makes it impossible to use oxygen for the determination of C until it can be shown from independent sources that oxygen in the elemental form is really mono- valent. There is some residual valence also. But if the residual valence be disregarded and two valences only be postulated in the molecule, the value of C would be 3.43 X io-37. The mean value of C determined from all the substances in Table VIII in my former paper was 3-45 X io~37, if Milli- kan's value of the number of molecules in a gram mol., namely 6.062 X io23, is used in the computation. Since in this calculation of C the molecules were not supposed to have residual valence, it is clear that an allowance for the presence of this valence would have the effect of lowering C, so that its true value must be somewhat less than 340 X io~37. A way in which C can be independently determined was suggested by the relation between cohesion and gravitation.1 In the formula M2K = C(Wt. X Val.)2/> it is evident that M2K is proportional to the 2/3ds power of the gravitational mass of a molecule and when weight and valence are unity M2K = C. It occurred to me that under these circumstances C might very possibly be nothing else than the factor (w2&)2/3 of a molecule of unity molecular weight. In this case "m" is the gravitational mass of such a molecule and "k" the gravitational constant. A computation of (m2k)2/1 using Millikan's recent determination of the number of mole- 1 Mathews: Jour. chim. phys., 1914* 480 A. P. Matkews cules in a gram mol, gave the result (w2&)2/3 = 3.20 X io~37 which is very close to the figures already obtained from methane and hydrogen, and somewhat less, as it ought to be, than the mean value of 3.45 X io~37 which was obtained when residual valence was disregarded. In view of these facts I believe we may assume that 3.201 X io~37 is the real value of C, although the theoretical basis of this relationship is still lacking, and proceed with the calculation of the total valence of a molecule on that assumption. The value of M2K is less satisfactory. It is here that the main uncertainty of the calculation lies. As I have already discussed the methods of calculating this value in my paper on the internal pressures of liquids I will not go into the question at this time further than to point out two or three considerations bearing on the probable accuracy of those figures. In all the formulas for "a" which have so far been proposed certain assumptions have been made. The one ordinarily made in van der Waals' method of computing "a" is that bc = Vc/3- In the various methods I have proposed for the computation quite different assumptions have been made in the different formulas, but nevertheless these formulas have all given results which are not widely different if the uncertainty of some of the experimental data are considered. Nevertheless, the formulas do not always give exactly the same values as they should if all the assumptions and data were rigorously correct. The computation of the cohesion from the latent heat of vaporization should give a correct result since the assumptions made here are less radical than in any of the other methods. Now this method generally gives a value for "a" lower, in some cases 5 percent lower, than that com- puted from the critical data. But I have not been able to attach more importance to this deviation for the reason that the computation must be made close to the critical tempera- ture, within a fraction of a degree of it, and a very slight error in the difference of the vapor and liquid densities would make a very large error in "a." That x the formulas proposed for "a" are possibly not entirely accurate may be shown also Residual Valence of Various Molecules 481 by the following circumstance: From the formula a = ((S2 S + 2)/(S 2))P«V* and the formula a = M((S2 — S + 2)/(S — 2))RT2/'(TC - - T)1/'/^ — T2/V(di — d,)Pe. If now we com- pute dc of oxygen by this formula from the densities of liquid and vapor oxygen found by Mathias . and Onnes, we find indeed a constant value for dc, but a density nearly 3 percent higher than that computed by the rectilinear diameter law, as follows: TABLE II Computation of the critical density of oxygen from the densities at different temperatures, t — 118.8° 0.4413 —120.4 0.4428 -140.2 0.4437 —154.51 0.4427 dc computed by Mathias and Onnes by the rectilinear diameter law was 0.4299. The deviation with pentane was in the opposite direction. dc was found by S. Young to be 0.2323. If it is calculated by the foregoing method from Young's density figures we have o 0.2235 160 0.2265 In this case the result was about 3 percent too low. With octane the computed and found values agree very well, as follows, using Young's density figures at various temperatures. / dc 0° 0.2319 60 0.2324 120 0.2336 1 60 0.2348 190 0.2352 230 0.2372 482 A. P. Mathews The density values show a tendency to advance. The mean value of about 0.2330 is very close to that determined by Young of 0.2327. The critical density calculated for various other sub- stances from the liquid and vapor densities resulted as follows when compared with the found values : Substance Temperature and density from dc which calculation made | Calculated dc Found Benzene 0° di = 0.90006; dv = 0.00012 0.3019 0.3045 Fl-benzene 0° di = 1.04653 0-3495 0.3541 Br-benzene 0° di = 1.52182 0.4804 0.4853 CO2 0° d\ = 0.914; dv = 0.096 0.4749 0.46 CCU ou di = 1.63255 0.5569 0.5576 Ethyl ether 0° d\ = 0.7362; dv = 0.000827 0.2605 0.2625 60° di = 0.66580; dj, = 0.006771 0.2622 The foregoing figures show that the formula used for the calculation of dc, which was derived from two of the formulas used in the calculation of ''a," gives results which agree generally within i percent of the values of the critical density determined by experiment, but in some cases there is a devia- tion of about 3 percent. We may, I think estimate the un- certainty in the value of "a" and hence of M2K, as not more than 2-3 percent. The formula which I have chosen for the calculation of the value of ua" is that which is based on the assumption that the value of bc is 2VC/S. This formula is : a = ((S2 — S + 2)/- (S -- 2))PCVJ. This equation involves only the critical data and may be applied to the largest number of substances. While the calculation of the total valence of the molecules is thus subject to these uncertainties, it is probable that the substances are arranged in their proper order of the amount of residual valence and that the error in the total valence of the molecule is not more than 5 percent at the outside. The re- sults are given in Table III. The values of "a" are taken from column 4 of Table VIII of my paper1 on the internal pressures of liquids. 1 Mathews: Loc. cit., p. 622. Residual Valence of Various Molecules 483 TABUS III Amount of residual valence I Substance II Formula III Theoretical No. of val- ences IV Total No. of valences by formula a=C1(Wt.XVal.)2/' V Resid- ual valence Hydrogen H2 2 2.195 O.IQ5 Oxygen 02 2 2.195 0.195 Nitrogen N2 2 2.197 0.197 Nitrous oxide N20 4(?) 6.765 2.765 Kthylene C2H4 12 12.904 0.904 n-Pentane C5Hi2 32 36.59 4-59 z-Pentane C5Hi2 32 35-95 3-95 w-Hexane CeHu 38 43-42 542 Diisopropyl C6H14 38 42.41 4.41 Heptane CrHie 44 50.89 6.89 n-Octane C8Hi8 50 59.08 9.08 Diisobutyl C8H18 50 56.30 6.30 Ether C4H100 28 30.45 2-45 Carbon tetra-chloride CC14 16 (Cl - 3") 20.28 4.28 Benzene C6H6 30 33-70 3.70 Chlor-benzene C6H5C1 32 36.40 4.40 Methyl formate C2H402 16 18.12 2.12 Ethyl formate C3H602 22 24-39 2-39 Methyl acetate C3H602 22 23.52 1.52 Methyl propionate C4H802 28 29-38 1.38 Ethyl acetate C4H802 28 29.20 1. 2O Propyl formate C4H802 28 31.16 3.16 Methyl butyrate C5Hi002 34 36.11 2. II Methyl iso-butyrate C5Hi002 34 35-30 1.30 Propyl acetate CsHioC^ 34 36.05 2.05 rt Ethyl propionate Stannic chloride C6Hi002 SnCl, 16 (Cl = 3) 35-80 17.68 1. 80 1.68 Carbon bisulfide CS2 16 (S = 6) 15-94* — Methane CH4 8 8.15 0.15 * "a" computed from the surface tension. These figures speak for themselves, but a word of com- ment may be made on some of them. It seems probable from the greater solubility in water of oxygen and nitrogen than hydrogen, that the average residual valence of a molecule 1 C = 3.20 X io~37 X N2. 484 A. P. Mathews of hydrogen gas is somewhat lower than the figures indicate. In the series methane, ethylene, pentane, hexane, heptane and octane the residual valence is, respectively, 0.15, 0.9, 4.6, 5.42, 6.89, 9.05. In other words, the residual valence increases with the number of carbon atoms as Armstrong has already inferred it should do. If 0.15 be considered as the average amount of residual valence of a carbon atom when united with hydrogen and if the amount of residual valence increases proportional to the square of the number of carbon atoms, we should have R. Val. = n2 0.15, n being the number of carbon atoms. By this formula the number of residual valences in the series mentioned should be, respectively, 0.6 for ethylene, 3.75 for pentane, 5.40 for hexane, 7.35 for heptane, and 9.60 for octane. These values are not very different from those actually found. This relationship does not hold for the esters. Another fact may be noticed, namely, that the differences between the total number of valences in the various groups of esters is very nearly the theoretical number. Thus between methyl formate and methyl acetate a difference of six valences is required. 5.40 was the difference found. If we take the average of the total number of valences found in the two esters of the formula C3H6O2 it is 23.95. This is 5.75 valences more than methyl formate has and is almost exactly six less than the next higher homologues of the formula C4H8O2, of which the average number of valences found was 29.91. This in its turn is again 5.79 (required 6) valences less than the average of the next higher group of the formula C5Hi0O2. Theoretically, there should be a difference of 32 valences be- tween the molecule of hydrogen and a molecule of the formula C5Hi0O2, whereas the method actually shows a difference of 33.62. It is certainly reasonable to suppose that this, differ- ence from the theoretical is to be ascribed to the larger amount of residual valence possessed on the average by a molecule of the ester as compared with hydrogen. In the case of the chlorine compounds I have assumed that the valence of chlorine is three. The reason for this is Residual Valence of Various Molecules 485 that I have not been able to find any chlorine compounds which show chlorine to have a lower valence than this. It might be assumed that these three valences were composed of one chief and two residual valences. In that case one should, of course, make the residual valence of chlor-benzene 6.40 instead of the value 4.40 which I have indicated. Another reason why I have not counted these two valences of chlorine as residual valences is that these chlorine compounds are non- associating compounds, or at any rate they associate very little. Hence the valences, sixteen in number, found in carbon tetra-chloride are probably not free residual valences, in the sense that they are valences in an active form but not saturated in the molecule, but they must be saturated in the molecule. On the other hand, the excess of 4.28 valences found above the calculated amount may or may not be in part saturated within the molecule. It is evident then that the determination of the residual valence by this method of subtracting the theoretical number from the total number found is open to these serious sources of uncertainty. All that can be claimed for the method at this time is that it gives a method of calculating the total valences and thus estimating the residual valence, and that so far as indications go in the hydrocarbons and the esters the compounds are at least arranged in the order in which they would be placed, judging from their reactions, if arranged according to the amount of residual valence they possess. I hope that methods will be found to differentiate more clearly between the valences extending between the atoms and those additional valences extending outward from the atoms making the residual valence proper. It is still too early to attempt to correlate the amount of residual valence with the solubility of compounds. It is at least possible that in solubility other factors than the number of valences come into play. The attraction between the molecules of solvent and solute may involve the factors which have been shown to influence cohesion, namely, molec- ular weight and number of valences, as well as the amount of 486 A. P. Mathews residual valence; it may also involve, of course, the amount of dissociation of the aggregates formed by cohesion or residual valence. It is probable, since the atomic unions are as a rule far more stable than the cohesive, the chemical attraction between the atoms being of an electro-static kind, that the union between solvent and solute due to the residual valence is of far more importance than that of a cohesional nature, just as the cohesional attraction is of vastly greater importance than the gravitational attraction; and there are not lacking indications that residual valence plays a very important part in solubility. I may mention in this connection the series helium to xenon already discussed elsewhere; the great dis- solving power of associated as contrasted with non- associated liquids, shown by the solvent powers of water; the less asso- ciation of associating substances when dissolved in associa- ting solvents as compared with their state in non-associating solvents; the greater solubility of such gases as hydrogen sulphide, ammonia, sulphur dioxide which have greater residual valence than nitrogen, hydrogen and oxygen, and the fact that they are known to combine with water, and so on. The residual valence is hardly ever found to be a whole number. The probable explanation of this is that the number found represents only the average amount of residual valence possessed by the molecules. It is probable that the residual valences open up in pairs, one positive and one negative, but that at any instant of time only a few molecules have them open, so that the average amount possessed by each mole- cule may appear to be a fraction. Summary The amount of residual valence of a number of non- associating liquids and gases has been computed by sub- tracting from the total number of valences which the mole- cule possesses, as shown by its cohesion, the number which there is reason to believe extend between the atoms of the molecule. The difference is considered to be the residual valence. Residual Valence of Various Molecules 487 The computation of the total number of valences in a molecule from the cohesion is made from van der Waals' factor "a" by the formula :a = C(Mol. Wt. X Val. number) 2/3. a was computed by the formula already given, namely a = ((S2 — S + 2)/(S— 2))PcVc,in which S is the critical coefficient and P6 and Vc the critical pressure and volume. The constant C for a single pair of molecules was assumed to be equal to (m2k)~/3, in which m is the gravitational mass of a molecule of unity molecular weight and k the gravitational constant. From Millikan's recent determination of the number, N, of molecules in a gram mol, the value of this constant was 3.2015 X io"37 expressed in dynes. For a gram mol C is 1.177 X ion. Owing to various uncertainties and assumptions in the calculations, this method of determination can be regarded only as of the nature of an approximation to the actual amount of residual valence of molecules. Reprinted fiom THE JOURNAL OF BIOLOGICAL CHEMISTRY, VOL. XIV, No. 5, 1913, AN IMPORTANT CHEMICAL DIFFERENCE BETWEEN THE EGGS OF THE SEA URCHIN AND THOSE OF THE STAR-FISH. BY A. P. MATHEWS. (From the Marine Biological Laboratory. Woods Hole.) (Received for publication, April 21, 1913.) The eggs of the sea urchin, Arbacia punctulata, differ markedly in their physiological properties from those of the star-fish, Asterias forbesii. The sea-urchin egg is remarkably stable, resistant to oxidation, has a very low rate of respiration and is not easily stimulated to artificial parthenogenesis; the star-fish egg, on the other hand, is, after maturation, very easily oxidized, has a rapid rate of respiration, forms sulphuretted hydrogen when mixed with sulphur, is easily destroyed by oxygen, easily liquefied by heat and is easily cytolyzed by anesthetics. It is readily, even by shock, caused to develop parthenogenetically. Moreover after matura- tion a steady growth of the nucleus takes place, whereas in Arbacia the nucleus after maturation remains of a very small size. Five or six years ago I found an important chemical difference between these eggs to be that cholesterol was lacking in the star- fish egg, but present in some quantity in that of the sea urchin. In view of the relation of cholesterol to hemolysis this observation offers a possible explanation of the great ease of cytolysis of the star-fish egg as compared with the sea-urchin. The eggs "were pressed from the ovary through cheese-cloth to remove the connective tissue, and the mass then extracted three times with a large amount of 95 per cent alcohol, boiling for one hour each time, and then once with boiling ether. The united extracts were evaporated on the water bath, the residue extracted with ether repeatedly, filtered from insoluble substances, and the ether poured into acetone. The fat and cholesterol remain in solution; the lecithin is precipitated. The acetone filtrate was evaporated to dryness, the oily residue saponified with alcoholic 465 THE JOURNAL OF BIOLOGICAL CHEMISTRY, VOL. XIV. NO. 5. 466 Chemical Difference in Echinoderm Eggs sodium hydrate and, after the addition of sodium sulphate and some water, was shaken out with ether repeatedly. The ether was washed several times with sodium carbonate solution and evaporated to dryness. The residue was very small in amount, not crystalline; it looked like oleic acid. It gave no positive tests for cholesterol either by Salkowski's or the Liebermann-Burchard method. I have repeatedly sought for cholesterol in these eggs varying the procedure but I have never been able to find it. On one occasion when the ovaries were not ripe the fatty residue of the ether after repeated saponifications,- both with alkali and acid, gave a very faint, transitory green such as cholesterol gives in the Liebermann test, and there may have been a very small amount of cholesterol present, but no crystals could be obtained. In view of the fact that the color reaction is probably not specific I am doubtful whether there was a trace of cholesterol present or not. It could not be positively identified. It may be mentioned that cholesterol in combination as in lanolin gives the Liebermann-Bur- chard reaction very strongly. The same methods applied to the sea-urchin egg gave, as usual, a crystalline mass on evaporating the ether after saponification; the crystals looked like cholesterol and gave a typical reaction of Salkowski. I may say that the extract of the whole body of the star-fish contains cholesterol in abundance. Another very interesting peculiarity of the star-fish egg is the character of its phosphatide. It resembles the jecorin described by Drechsel. A large quantity of eggs was extracted with hot alcohol and ether; the lecithin (?) precipitated from the ether solution in the usual way, redissolved in ether (not anhydrous) and reprecipi- tated with acetone and the process repeated until it dissolved quite clear in the ether and did not settle out a white substance when standing in the cold. This white substance coming out of the ether had a sweet taste, but had no reducing action on Feh- ling's solution either before or after heating with hydrochloric acid. The phosphatide thus prepared is more hygroscopic than lecithin from the brain or eggs. It makes unusually beautiful, regular mye.in forms when shaken with water, and it seems to be toxic for sea-urchin eggs. It was probably not a pure substance. It contains a large amount of a reducing sugar which, calculated as glucose, amounts to 10.51 per cent by weight. The nature of this A. P. Mathews 467 sugar was not determined; an osazone was prepared; the fermen- tation test was indecisive. The lecithin itself does not reduce Fehling's solution but only after it has been heated with acid. I heated it for ten hours with 3.5 per cent HC1 and determined the sugar by the reduction of Fehling's solution according to the method of Munson and Walker. This phosphatide also contains sulphuric acid in an ester form like Koch's sulphatide. It con- tained in a single analysis 1.19 per cent of sulphur in an oxidized form. This is in organic combination. The fatty acids are very largely oleic, or a similar acid, having an ether-soluble lead salt, The analysis of this impure phosphatide resulted as follows: Glucose (?) 10.51 per cent. Fatty acids 46.16 per cent. Phosphorus 3.57 per cent. Sulphur 1.19 per cent. Of the fatty acid approximately 71.35 per cent was recovered as oleic (?) acid. This phosphatide also contains a considerable amount of magnesium, but I did not determine it quantitatively. I may mention that, of the total ether-soluble portion of the alcohol-ether extract of these eggs, the lecithin in one case weighed 0.6105 gram; the fat, the part not precipitated by acetone, 0.6055 gram; so that there are about equal quantities of fat and lecithin. SUMMARY. Cholesterol is either absent altogether or present in very small amount in the star-fish egg. It could not be positively found in the eggs of Asterias forbesii. It is present in considerable quan- tities in the sea-urchin egg. This difference possibly is correlated with the greater sensitiveness to cytolysis of the star-fish egg. The phosphatide of the star-fish contains about 10 per cent of a re- ducing sugar in firm combination and also sulphuric acid. Reprinted from the American Journal of Physiology Vol. XXXII — June 2, 1913 — No. II CARBON DIOXIDE PRODUCTION FROM NERVE FIBRES WHEN RESTING AND WHEN STIMULATED; A CONTRIBUTION TO THE CHEMICAL BASIS OF IRRITABILITY.1 BY SHIRO TASHIRO [From the Department of Biochemistry and Pharmacology, the University of Chicago, and the Marine Biological Laboratory, Woods Hole, Mass.] INTRODUCTION THERE have been two theories of the nature of conduction — one upheld among others by Hermann, that it was a prop- agated chemical change; the other, at present the dominant view, that it is a propagated physical change. In 1901 Professor Mathews suggested 2 that it was in the nature of a coagulative wave propagated along the fibre; this coagulation of the nerve colloids leading either directly or indirectly to the electrical disturbance accompanying the impulse. At the time, there was no evidence of chemical change in the nerve fibre, and its indefatiga- bility seemed to point to an absence of metabolism. Certain facts were known, however, which were difficult to reconcile with this phys- ical theory. Darwin had observed that in Drosera,3 conduction occurred only if the protoplasm had oxygen; and Mathews 4 observed that salts would not stimulate a nerve, or, at any rate, their power of stimulation was much reduced if the nerve remained in the body for a time after death, or if the nerve were brought into the salt solution in an atmosphere of hydrogen. This clearly indicated a dependence of^the irritability on oxygen. 1 The preliminary report of these investigations was given in part in Bio- chemical section of Eighth International Congress for applied chemistry, Sep- tember, 1912. See original communications, Eighth International Congress of applied chemistry, xxvi, p. 163. See also this Journal 1913, xxxi, p. xxii. 2 Mathews: Century Magazine, 1902, pp. 783-792; Science, 1902, xl, p. 492. 3 Insectivorous Plants, p. 57. 4 Unpublished observations. io8 Shiro Tashiro This fact lead to a search for evidence of the chemical nature of irritability and in a number of papers 5 it was clearly pointed out that the anaesthetics were probably acting directly in a chemical manner instead of indirectly, by affecting permeability, and that probably the anaesthetics acted by uniting with the protoplasm where 02 usually took hold. This view was strengthened by the temperature coeffi- cient of conduction, which is nearly that of a chemical reaction; by the importance of C>2 for artificial parthenogenesis; and by many other facts some of which have recently been collected by Haberlandt, Buijtendijk and others. Although it has been established by repeated demonstrations, that the nerve does not fatigue under ordinary conditions, as measured by the method used in muscular studies, yet Frohlich 6 observed that the nerve undergoes certain changes by long activity. Gotch and Burch discovered7 in 1889 that if two stimuli are successively set up within -%%-$ of a second, only one negative variation is produced. This critical interval, or refractory period, is found to be altered by temperature changes, by drugs, asphyxiation, and anaesthetics.8 Thus by prolonging the refractory period by partial anaesthesia, Froh- lich easily demonstrated that with a frequency of stimulation less than this normal refractory period, stimulation of the attached muscle no longer occurred. He interprets this as a phenomenon of fatiga- bility of the nerve. Thoner's 9 observation seems to lead to a similar interpretation, for he found recently that fatigability is less effec- tive when the refractory period is shortened by high temperature. There seems, then, to be fatigue in the nerve, but it cannot be measured by an ordinary scale. After the complete failure of the chemical detection of CO2 and 6 A. P. MATHEWS: Biological bulletin, 1904-5, viii, p. 333; this Journal, 1904, xl, p. 455; ibid., 1905, xiv, p. 203; Biological Studies by the pupils of William Sedgwick, 1906, p. 81; Journal of pharmacology and experimental therapeutics, 1911, ii, p. 234. 6 FROHLICH: Zeitschrift fur allgemeine Physiologic, 1903-4, iii, p. 445. Ibid., P- 75- 7 GOTCH and BURCH: Journal of physiology, 1899, xxiv, p. 410. 8 See TAIT and GUNN, Quarterly journal of experimental physiology, 1908, i, p. 191; TAIT, ibid., 1909, ii, p. 157. 9 THONER: Zeitschrift fur allgemeine Physiologic, 1908, viii, p. 530; ibid*, 1912, xiii, pp. 247, 267, 530. Carbon Dioxide From Nerve Fibres 109 acids in the excited nerve, Waller still believes that it must give off CO2 when stimulated. In 1896, he showed, with an electro-physio- logical method, that among other reagents, CO2, in minute quanti- ties, increased the excitability of the isolated nerve of the frog, and that the normal nerve, when excited, also increased its activity.10 From this he ingeniously formed the hypothesis that every activity in the nerve fibre must be associated with C02 production. That there may be CO2 production in the nerve, but too small to be measured by ordinary methods, is shown by the following calcu- lations: A frog (Rana temporaria) gives off 0.355 gram of CO2 per kilogram per hour at 19 — 20° C.11 A small piece of the nerve fibre of the same animal, say i cm. in length, will weigh in the neigh- borhood of 10 milligrams. Now, if the mass of the nerve respires at the rate of the whole animal, it would give off about 0.0000007 grams of CO2 during ten minutes. This calculation at once suggested that the lack of positive evidence of metabolism in the nerve fibre was not at all conclusive that such metabolism did not occur, in view of the limitation of the methods for the estimation of C02. It was evidently necessary to devise methods for the detection of very minute quantities of CO2. Thus at Professor Ma thews' suggestion a new method for CO2 analysis was first devised, and then, under, his direction, I have undertaken to go back once more to the question of CO2 production in the nerve fibre during the passage of a nerve impulse. To study the nature of metabolism involved in a tissue, one should at least determine the oxygen consumption and the carbon dioxide production. Inasmuch, as the present problem, however, is concerned only with direct evidence for the existence of metabolism in the nerve fibre, I have attempted to measure C02 production only, for it is true that the lack of oxygen consumption may not necessarily indicate the absence of chemical changes, while the pro- duction of C02 will surely prove the presence of metabolism. Further- more, as CO2 production is the only sure universal expression of the respiratory activity in anaerobic and aerobic plant and animal tissue in normal condition, the inquiry of CO2 production in an excited nerve will not only concern the problem of the nature of the nerve impulse 10 WALLER: Croonian lecture, Philosophical transactions, London, 1896. 11 Taken from Pott's figures. See figures in Table ix, p. 129. no Shiro Tashiro itself, but may, also, aid in forming a fundamental conception of the tissue respiratory mechanism. In this way, if the protoplasmic irri- tability has a direct connection with the cellular respiration, then our idea of the general nature of the pharmodynamics of many reagents on a living tissue may be essentially modified. METHODS AND MATERIALS Two new apparati were constructed which will detect CO2 in as small quantities as one ten-millionth of a gram and estimate it with quantitative accuracy. The detailed method has been described in a separate article.12 Preliminary experiments with these new apparati showed that the sciatic nerves of dogs gave too large quantities of CO2 for my method so that I was compelled to use a smaller nerve of a cold-blooded animal for quantitative estimation. For exact measurements of CO2 production, I have used only two kinds of nerve, although I have used a large variety of nerves in qualitative experiments. For a non- medullated nerve fibre, Prof. G. H. Parker 1S was so kind as to sug- gest to me that I use the nerve trunk of the claws of the spider crab (Labinia Caniliculata) which is a bundle of mixed sensory and motor fibres. The frog, whose sciatic was used as a representative for medullated nerve, was exclusively Rana pipiens, obtained from Indiana. As my apparati in the present form cannot be used for a muscle nerve preparation nor for the normal nerve in situ, the use of an isolated nerve could not be avoided. Experimental factors thus intro- duced should be carefully considered before we interpret the observa- tion as a normal metabolism. This serious objection, however, can be overlooked, as far as our fundamental question of different metabolic activities before and after a stimulation is concerned, for Waller 14 has demonstrated that the presence of excitability in an isolated nerve persists as long as nineteen hours provided that the electrical changes correctly represent the state of excitability. Although 12 See pp. 137-145- 13 For this and other suggestions, I am under great obligation to Dr. Parker. 14 WALLER: 1896, Brain, xix, p. 53. Carbon Dioxide From Nerve Fibres in Herzen claims that under certain conditions of local narcosis the nerve fibre may give an action current without any muscular con- traction (Wedenshi and Boruttau both deny this), and Ellinson 15 recently demonstrated by the use of cinchonamine hydrochloride the absence of negative variations without abolishing the excitability of the nerve, yet evidences are now abundant to indicate that the action current is a normal physiological phenomenon in uninjured tissue expressing the simultaneous activity resulting in a corre- sponding change in the peripheral organ.16 These facts, therefore, must be taken as showing that as long as a negative variation remains, the nerve is probably excitable; and that the phenomena observed in the isolated nerve could be regarded as identical with that of a nor- mal nerve as far as the passage of a nerve impulse in an isolated nerve fibre is concerned. CO2 PRODUCTION FROM RESTING NERVE In this study of the metabolism of the resting nerve, particular care was taken to select those fibres which were free from nerve cells. The work of several investigators 17 seems to indicate that tissue oxidation is primarily concerned with the cell nucleus. Inasmuch as the respiration in the central nervous system is certain 18 and the blood supply to fibres is seemingly scanty, the notion persists among certain biologists that a nerve fibre should not respire since it has no nucleus. In order to test the correctness of such an idea, I have studied quantitatively the output of CO2 from various lengths of nerve which are known to be free from nerve cells.19 Here is the result: 15 ELLINSON: Journal of physiology, 191 1, xlii, p. i. 16 For further details, see: GOTCH and HORSLEY: Philosophical transactions of the Royal Society, 1891, clxxii, p. 514; BERNSTEIN: Archiv fiir die gesammte Physiologic, 1898, Ixxiii, p. 376; REID and MCDONALD: Journal of physiology, 1898-9, xxiii, p. 100; LEWANDOWSKY: Archiv fiir die gesammte Physiologic, 1898, Ixxiii, p. 288; ALCOCK and SEEMANN, ibid., 1905, cviii, p. 426. 17 See SPITZER: Archiv fiir die gesammte Physiologic, 1897, Ixvii, p. 615; M. NUSSBAUN: Archiv fur mikroskopische Anatomic, 1886, xxvi, p. 485; R. S. LILLIE: This Journal, 1902, vii, p. 412. 18 L. HILL: Quoted from Hulliburton's Chemistry of nerve and muscle, p. 79. 19 In this connection, I wish to express my indebtedness to Prof. H. H. Donald- son for his kind advice. ii2 Shiro Tashiro Non-Medullated Nerve Fibre. — (The nerve of the spider crab, and apparatus 2 for the qualitative, and apparatus i, for the quanti- tative, estimations were used.) When I place the nerve of a spider crab in the right chamber and no nerve in the left, and watch for the deposit of barium carbonate, the drop on the right will soon be coated with the white precipitate, but no precipitate whatever is visible with a lens in the left. ,C02 is thus shown to be produced by this resting nerve. Now, by interchanging the nerve from the right to the left, no nerve being in the right, we can convince ourselves of the correct- ness of this conclusion, by eliminating any technical error which might produce the different results in different chambers. The rate at which the precipitate appears and the quantity of the precipitate, depends on the size of the nerve. In fact, C02 production from the resting nerve of the spider crab is found to be proportional to its weight, other things being equal, and is constant: For 10 milligrams per ten minutes it gives 6.7 X -io-7 grams at 15 — i6°c. The quantitative determination of this amount is made in the following manner : The claws of the crab are carefully removed, and, by gently cracking them, the long fibre of the nerve trunk is easily isolated. After removing the last drops of the water by a filter paper, the nerve, with the aid of glass chop sticks, is carefully placed on the glass plate,20 and quickly weighed. The glass plate with the nerve is now hung on the platinum hooks in the respiratory chamber A, and then the .chamber sealed with mercury. The analytic chamber is now filled with mercury in the manner described elsewhere,21 and then the apparatus is washed by C02 free air as usual. The time when the barium hydroxide is introduced to the cup in chamber B is recorded, and the stop-cock between the two chambers is closed. When at the end of ten minutes the drop at cut F is perfectly clear, having not a single granule of the precipitate visible to a lens, thus insuring that the air is absolutely free from CO2 then a known portion of the gas from the respiratory chamber is introduced into the chamber below in which the clear drop of barium hydroxide has been exposed, and it is determined whether or not the amount of the gas taken contains 20 The weight of this plate is known so that the weight of the nerve can be determined very quickly. See p. 120. 21 See pp. 139. Carbon Dioxide From Nerve Fibres 113 enough CO2 to give the precipitate in ten minutes. If it does, a fresh nerve is prepared and a less volume of the gas is withdrawn; if it does not, a larger volume should be taken till the precipitate appears within ten minutes. (See footnote, page 140.) In this way, by repeated experiments with several fresh nerves, a minimum volume of the gas for a known weight of the nerve which gives a precipitate is determined. This minimum volume should contain exactly a definite quantity of CO2 — namely i.o X io~7 gram.22 In this way, since we know the original volume of the respiratory chamber from which this minimum volume is withdrawn, and since we know the quantity of CO2 contained in this volume, it is easily calculated, how much CO2 is produced by the nerve during the known period. It should be understood that in determining the minimum volume of gas taken from the respiratory chamber, a series of experi- ments were conducted in order to calculate both the minimum volume which just gives the precipitate and the maximum volume which does not give the the precipitate for a known weight of the nerve for a known period of respiration In the tables following, columns 8 and 9 refer to these volumes calculated from experiments. Table I, gives the result for a non-medulla ted nerve. Medullated Nerve Fibre. — For the quantitative estimation of C02 production from the medullated nerve I have taken a frog's sciatic, using apparatus 2. The results given in Table II, obtained by, similiar methods, show that each ten milligrams of the frog's sciatic nerve gives off 5.5 X icr7 grams for the first ten minutes. A large quantity of nerves were tested and it was determined whether or not all resting nerves give off CO2. As a result, I found no exception in any of them. The following varieties of nerves were examined: 1. MOTOR NERVE: Occulo-motor nerve of the skate. (Raia Ocallata.} 2. SENSORY NERVE: Olfactory nerve of the same. (Raia Ocallata.) 3. MEDULLATED NERVE: Sciatic nerve of the dog, frog, turtle, mouse; • optic nerve of the skate. (Both Raia Ocallata and Raia Erinecia.) 4. NON-MEDULLATED NERVES: Nerves of the spider crab; olfactory nerve of the skate. (Raia Ocallata.) 5. NERVE OF INVERTEBRATE: Spider crab's nerves. 22 See p. 140. Shiro Tashiro Carbon Dioxide From Nerve Fibres fi D 2 ,Q s|.§! o1|-S 0 JBy glancing at the columns 8 and 9 it is clear that 2.70 c.c. is the minimum volume, for 2.6 c.c. is maximum volume which does not give the precipitate. Since original volume of respiratory chamber is 15 c.c. we have 1.0 X 10-7 g. X £V = 5-5 X 10-7 g. C02 at 19° - 20° 2 Little high result in these cases is no doubt due to high temperature. -1 •*:8'i!fS 0 ^ +J^H « 3 oo.o.HJ 6 uu o^ •^H CN'^-lt-lCNCN CN CNCN CNCNCN illi . O O O 10 w O . l>- t^ r*> OO >O -^ . VO . . Tj" . . . • Tt< CN CN fO CN CN CN ' CN * ' CN ' ' ' — > 1 + 1 1 1 1 1 ++++++ 1 + 1 1 + 1 1 1 «n 3 h. II II J"3 § ° ju^ooouuooy^o^u^^^yuu II £ •« P CU 0 g j !o">0 0^0.0^001000 000000^. d o 3 en . «j_i CO f-j m u. V ^ <0 ® S O\ . 1-4 T-H OO ^D ^^ O\ ^^ O\ CN **^ *O *^ f*O fO C5 l>» Os *O 00 ^ . CN CN CN CN CN '-I CN TH CN CN CN CN CN T-I CN CN CN CN »-i ft CN-CN-^f'i1*'" S < u6 Shiro Tashiro 6. NERVE OF VERTEBRATE : Nerves of frog, dog, mouse, squiteague (cynoscion Regalis), and skate. (Both Raid Ocallata and Raia Erinecia.) 7. NERVE or WARM-BLOODED ANIMALS: Those of dog, mouse and rabbits. 8. NERVE OF COLD-BLOODED ANIMALS: Frog, squiteague (cynoscion Regalis) and skate. (Both Raia Ocallata and Raia Erinecia.) From this I have concluded that isolated nerves of all animals give off CO2. It remains, now, to consider whether this CO2 is the product of normal respiratory activity or due to disintegration of the dead tissue. IS THE C02 GIVEN OFF PRODUCED BY LIVING PROCESSES? Comparison of Dead and Living Nerves. — In the first place, it was thought that if C02 was due to normal metabolism of a living nerve, its production should be diminished when the nerve was killed. The following result (Table III) is self explanatory. TABLE III COMPARISON BETWEEN NORMAL AND KILLED (BY STEAM) NERVES OF SPIDER CRAB 123 4567 Date Tempera- ture of room Weight of nerve in mg. Stimula- tion c.c. of gas taken from respiratory chamber Duration of respiration : minutes Ppt. of Ba(CO3) after ten minutes Nov. 4 13° 40 (kiUed) no .5 10 - it (C 40 (killed) st'n .5 10 - " 5 16 (normal) no 1. 10 + " 6 15 16 (killed) no 1. 12 - " 7 16 16 (normal) no 1. 10 + Comparison of Anaesthetized and Non-Anaesthetized Nerves. — It is naturally feared, however, that the killing experiment itself may not prove that C02 production is necessarily due to the living mechan- ism, for high temperature may drive off C02 produced already by the process of tissue disintegration, just as the C02 diffused out from a wet thread saturated with the gas, the rate of diffusion being a func- tion of temperature. Thus anaesthesia was tried, although we should Carbon Dioxide From Nerve Fibres 117 expect at the outset that if ether had no direct affect on the respira- tory process, as some physiologists believe, then the negative results would not at all interfere with my contention. The fact is, however, that either an isolated nerve directly treated with ether vapor or urathane, or the nerve isolated from a deeply anaesthetized frog gave a much less quantity of CO% than the normal nerve isolated from a normal frog whose heart has been cut away for a period of time equal to that of etherization. Anaesthetics, then, diminish CO2 production from an isolated nerve fibre. These experiments are being continued quantitatively. CO 2 Production of Isolated Nerve at Successive Time Intervals. — It was also thought that if CO2 production was due to bacterial decomposition, although it is highly improbable for such a fresh tissue, we may expect that either killing by steam or treating with TABLE IV SHOWING DECREASED CO2 PRODUCTION BY LONG-STANDING (FROG'S SCIATIC) 123 4 Temperature of room Time elapsed after isolation Minimum c.c. necessary to give J, calculated for 10 mgs. 10 minutes Total CO2 pro- duced from nerve of 10 mg. for 10 minutes 24° immediately 2.7 c.c. 5.5 X 10-7g. CO2 25 1 hour 7.08 c.c. 2.1 X 10-7g. COa 24 2 hours 10.8 c.c. 1.4 X 10-7g. CO2 24 5.5 hours 12.8 c.c. 1.1 X 10-7g. C02 23.5 7 hours 15.3 c.c. .9 X l(Hg. C02 23.5 10.5 hours 21.0 c.c. .6 X 10-7g. C02 24 26 hours 9. cc.1 1.6XlO-7g.C02 24 27.4 hours 1.8. c.c 8.1 X 10-7g. C02 1 The gradual increase at this point should be noted (after 26 hours, it is clear that bacterial decomposition sets in). ether would check the C02 production, and that the results observed above may not necessarily prove that C02 production from the isolated nerve fibre is due to a respiratory process. Hence a number of the nerves were isolated from several frogs of the same size and sex, and n8 Shiro Tashiro were left in Ringer's solution, and then the rate of the gas production is determined with the different nerves removed at successive inter- vals of time from the Ringer's solution for twenty-five hours. The interesting results given in Table IV not only show that CO 2 from the fresh nerve is not due to bacterial decomposition, but it also indicates that when such abnormal decomposition sets in, the output of gas takes a sudden jump. This Table further shows that the vital process by which CC>2 is produced gradually slows up as the tissue approaches death, indicating that the decrease of CO2 production is parallel to the decrease of irritability of the nerve. Increase of CO2 on Stimulation. — The most convincing evi- dence of all that CC>2 is formed by a vital process is the fact that a stimulated nerve gives off more CO2 (Part II) indicating the presence of normal metabolism in the living nerve which is accelerated when the nerve is stimulated. Thus we may safely conclude here that like any other tissue or organs, the nerve, too, respires whether it has a nucleus or not, and that the rate of C02 production is pro- portionate to its weight, other things being equal. CO2 PRODUCTION FROM STIMULATED NERVE We have now come to our main inquiry, namely, is there any chemical basis for irritability? Just what relation exists between nervous activity and chemical changes is the question that a biologist should consider before he attempts to build any conception of the real dynamics of living matter. For it is the phenomena of excita- bility in the nerve fibre that has stood so long in the path of under- standing protoplasmic irritability in general. As for the brain, it is now established that certain chemical changes are involved during stimulation and that definite chemical changes are associated with pathological cases either in its chemical composition 23 or in the for- mation of abnormal metabolites.24 Aside from the confused facts concerning histological changes in the ganglion cells of fatigued ani- mals, Hill has observed, using Ehrlich's method of methylene blue 23 KOCH and MANN: Archiv of neurology and psychiatry, 1909, iv, p. 44. 24 DIXSON: Journal of physiology, 1899-1900, xxv, p. 63; CROFTAN: American journal of the medical sciences, 1902, p. 150. Carbon Dioxide From Nerve Fibres 119 for the determination of the rate of oxidation, that a spot of cerebral surface, if stimulated, loses its blue color owing to the using up of the oxygen.25 In case of the nerve fibre, however, we have already seen that no direct evidence has ever been presented to show any chemical changes connected with its activity, although there has been some indirect evidence. As considered before, the failure of the direct detection of CO 2 from the stimulated nerve must be due to the lack of a delicate method. Thus using the new method we have already demonstrated that a resting nerve gives off C02, and will now attempt to prove that nerves give off more C02 when stimulated.26 Electrical Stimulation of non-Medullated Nerve. — Owing to the scope of delicacy of the new method, which is sensitive to as small a quantity as i.o X io~7 gram (an amount corresponding to the CO2 contained in -^ cc. of pure air), the utmost caution must be taken to prevent any complication which may result in formation or absorption of minute quantities of C02. After I had found by experiment that there is no appreciable increase of CO2 due to the direct electrical decomposition in the nerve when stimulated by a weak induction current and that several other forms of stimulation qualitatively confirmed the results obtained by the electrical stimulation, I have naturally employed the induction current as a stimulant in all my experiments on the quantitative estimation of CO2 production from the stimulated nerve.27 As Table V shows, the stimulated non-medullated nerve fibre of the spider crab gives off 16. X io~7 grams of CO2 for 10 milligrams of 25 HILL: loc. cit. 26 Professor Carlson has very kindly called my attention to a recent publica- tion from the Physiologisch Laboratorium der Utrechtsche Hoogeschool, in which Buijtendijk reports that certain head nerves of fishes take up more O2 when electrically stimulated. He could not, however, find any increase of O2 con- sumption in the sciatic of the frog. Also see: Koninklijk Akademie van Wetenschappen, Amsterdam, afd, xix, pp. 615-621. Haberlandt also recently reports (Archiv fur Physiologic, 1911, p. 419) that the resting nerve takes up of O2, 41.7 - 33-4 cmm. at 19° - 24° per gram per hour. When this nerve is excited, intake of O2 is increased. Since the respira- tory quotient of the stimulated nerve is equal to that of the resting, he con- cludes that when the nerve is excited, it must give off more CO2. He does not, however, indicate how much CO2 is produced by stimulation. 27 Use of non-polarizable electrodes was impossible for my apparatus, for the presence of foreign liquid in the chamber interferes with C02 estimation. As i2o Shiro Tashiro nerve for ten minutes, while a fresh resting nerve gave only 6.7 by iQ-7 grams for the same units. The details of the methods are as follows: The nerve of the claw of the spider crab is isolated as before. A comparative estimation was made first. Two pieces of the nerve of equal . weights and length were placed separately on the two glass plates, each nerve being laid across the electrodes of the plate, in the manner shown in Figure i . In this way either nerve can be stimulated at will. These glass plates are hung by their wires upon the platinum wires fused into the side of the apparatus, these wires being con- nected in turn with the induction coil. Under this condition, when both nerves are not stimulated, the amounts of the precipitate are equal in both chambers. However, when one of the nerves is elec- 0 FIGURE. 1. Glass weighing plate. A. B. Platinum wire fused in the rear of the glass plate, with hooks. C. The nerve which is stimulated at D. G. The plate proper. I have the other piece of the same glass out of which this plate is made. This piece of glass is weighed exactly equal to this weighing plate, so that any wet tissue can be weighed very quickly. In order to make results more accu- rate, no attempt was made to weigh closer than | milligram. trically stimulated (the distance between the primary and secondary coils was always more than 10 cm. using a red dry battery, the current being barely perceptible on the tongue), not only does the precipitate appear sooner in the chamber in which the excited nerve is placed, but also the quantity of the carbonate is much greater. To test whether the increase of CO2 production from the stimu- lated nerve is due to the direct decomposing influence of the current, or to the increase of metabolism produced by the passage of a nerve long as we are not concerned with the electrical changes in the nerve, the use of platinum electrodes instead, is not a great objection, provided that the current is weak enough not to decompose the tissue directly, and that the duration of stimulation is not very long. Carbon Dioxide From Nerve Fibres 121 impulse, the following experiments were performed. If we assume that the condition under which an electrical decomposition takes place is the same -both in the living and the dead nerve, then if the increased C02 is due to the current itself, we should expect that when a killed nerve is stimulated by a current, it ought to increase C02 production just as much. When I placed two nerves killed by steam in each chamber, and stimulated only one of them, the stimulated nerve did not give any more CO2 than the unstimulated, using the same strength of current employed in the other experiments. In the next place, it was thought that if the increase of CO2 is due to direct electrical decomposition, not limited to the point of contact with the electrodes, we ought to get a proportional increase of C02 by altering the distances through which the current directly passes. The fact was, however, that we could produce an increase of C02 production by stimulating with electrodes 2 mm. apart as well as by 15 mm. apart. Increase of C02, therefore, is due to nervous excita- tion and not to the direct influence of the electric current itself. With this consideration, I have proceeded to make a quantitative estimation of CO2 from the stimulated nerve in the manner described before. The results are shown in Table V. Electrical Stimulation of Medullated Nerve. — With apparatus 2, the output of C02 from the excited sciatic nerve of the frog has been quantitatively estimated. As shown below, 10 mgs. of the sciatic nerve gives off 14.2 X icr7 grams of CO2 during ten minutes stimula- tion while the resting nerve of the same animal gave off 5.5 X io~7 grams for the same units. Mechanical Stimulation. — We have now established the fact that when a nerve is stimulated by an electrical stimulus, it gives off more C02. In order to prove more conclusively that this CO2 production is due to the passage of a nerve impulse, I have employed several other means which are known to have definite influence on excita- bility of the nerve. So far, the use of these methods has been confined to qualitative experiments, but the results are a sufficient confirmation of the observations made by electrical stimulation. I cite them here as a preliminary report. Since the ordinary method for mechanical stimulation cannot be applied directly to the nerve in my apparatus in its present form, I used a different method, namely, crushing the nerve. That, when a 122 Shiro Tashiro s |1|| Q O" C« y " IH M o*S^»«.rt y3 3 . T)J . . r^ . . . ^ 'a' M *z • .y Wp3 8 .S s S a § 00 sl|.|l"B o . W jji en be '~ 6 bb £0*7 > CO S s M fc > ^— 1 i— 1 T-l i— ( '—''—' CS| T-l ^-H »— 1 »— I 1 |ll o CNCNCS'-i'-lCNeNCNCSCN»-lCSCSCNCNCSCNCS 1 p^-^^^""^^^ •g ^5 S < tn cs § ! ^ o > ^ 93 l» .§ » 124 Shiro Tashiro protoplasm is smashed, there occurs vigorous chemical changes, is shown by several investigators. Fletcher 28 reports that injured muscle gives off more C02 than the normal. Later he and Hopkins 29 discovered that muscle, under a similar condition, is richer in lactic acid. Dr. Mathews has observed a similar activity in crushed eggs of Arbacia. Quite accidently, I have discovered that a fresh nerve, too, when crushed with the rough edge of a glass rod gives off more C02. This increase of gas production from the injured nerve, I take to be due to mechanical stimulation. To test this hypothesis, I rendered the nerve unexcitable by means of ether and 0.2 m. solution of KC1, which is known to abolish excitability of a nerve.30 Under these conditions, I observed no increase of gas production when the nerve is crushed. Therefore, the metabolism existing in the living nerve must be accelerated by this stimulation when it is injured. This interpretation, however, is not accordant with that of Fletcher and Hopkins, on muscle. In studies of lactic acid formation in muscle, they found that lactic acid is spontaneously developed, under anae- robic condition, in excised muscle, and that fatigue due to contractions of excised muscle is accompanied by an increase of lactic acid. In an atmosphere of 02, there is no survival development of lactic acid for long periods after excision. From a fatigued muscle, placed in 62, there is a disappearance of lactic acid already formed. But this disappearance of lactic acid, due to oxygen, does not occur, or is masked, at supraphysiological temperature (e. g., at 30 ). Now traumatic injury to an irritable muscle too produces a rapid develop- ment of acid. Since, however, in this case the disappearance of lactic acid due to O2 does not occur, they conclude that one essential condi- tion for this effect of oxygen appears to be the maintenance of the normal architecture of the muscle. Thus they contend that the increase of the lactic acid by mechanical injury is not due to stimula- tion, but must be due to tissue destruction. They, however, did not determine, as 'far as I know, how much the output of CO2 is affected by treating the injured tissue with O2. 28 FLETCHER: Journal of physiology, 1898-9, xxiii, p. 37. 29 FLETCHER and HOPKINS: ibid., 1906-7, xxxv, pp. 261, 288. 30 MATHEWS: This Journal, 1904, xi, p. 463. Carbon Dioxide From Nerve Fibres 125 Unless it is proven that CO2 production from the injured muscle is quantitatively equivalent to lactic acid formed, their interpretation cannot be applied to the injured nerve, for in the case of the " plateau " of the survival muscle respiration, when in complete loss of irritability, the lactic acid yield remains stationary, Hill calculated that the C02 production corresponds to the amount liberated from the carbonate of the tissue by the lactic acid formed.31 Furthermore, if their interpretation is applied to the nerve, the fact that etherized nerves or nerves rendered unexcitable by KC1 do not increase CO2 output when crushed, cannot be explained. The fact that only excitable nerves when injured increase their CO2 pro- duction, is a sufficient proof that some sort of stimulation is applied to the nerve when crushed, the tissue destruction, no doubt, following afterward. The increase of CO2 production on crushing the living nerve and its absence on crushing the anaesthesized nerve is the point that I want to emphasize here in order to confirm my results obtained by electrical stimulation. I may add here that a perfectly parallel increase of CO2 by crushing has been observed hi dry seeds, including wheat, wild oats, Lincoln oats, Swedish select oats, leaves of Japanese ivy, and spinal cords of rabbit.32 Chemical Stimulation. — The study of the nature of chemical stimulation has been so thoroughly made 33 that at first it was thought that chemical reagents would be ideal as stimuli. It was soon discovered, however, that the presence of minute quantities of a foreign liquid is such a disturbing factor that stimula- tion by salt solutions could not be used for quantitative experiments. With a qualitative analysis, however, I found a variety of evidences which show that the nerve* stimulated chemically gives off more CO2, and that the nerve rendered less excitable by reagents decreases CO2 production. When each sciatic nerve of a frog is isolated and one is left in the normal saline in one case, and in the body of the frog in the other, for the same length of time, and then transferred to the two chambers of the apparatus, if the quantities of the precipitate are compared, it is found that the nerve which has been in normal saline gives more CO2. 31 HILL: Journal of physiology, 1912, xliv, p. 481. 32 Fuller discussion of these will appear in a subsequent paper. 33 MATHEWS: This Journal, 1904, xl, p. 4555 ^os, riv» P- 203- 126 Shiro Tashiro It is known that normal saline stimulates frog's sciatic nerves. The different rates at which CO2 is produced from the different nerves treated by various concentrations of KC1 is equally instructive. It is known that when a nerve is placed in a molecular solution of KC1, a stimulation takes place for a considerable time. Then it finally becomes unexci table,34 whereas, .2 m. KC1 solution abolishes nervous excitability in a short time without primary stimulation. The CO2 production follows exactly analogous to this. The nerve treated with the stronger solution gives more CO2 than that of a weaker solution. This was true even after both nerves became unexcitable, showing that the nerve must be giving off more CO2 while being stimulated by the stronger solution. Although my quantitative data are not complete at this stage, this preliminary statement is sufficient to show that the nerve chemically stimulated gives off more C02. It may be added in passing that the different solubility of CO2 in the different concentrations of these salts solutions cannot explain these results solely by a physical interpretation, for there is not enough difference in the solubility of C02 in dilute equimolecular solutions of KC1, and NaCl, whose effect on C02 production is so divergent, the former salt diminishing, the latter increasing it. Heat Stimulation. — It may be recalled in Table I that high tem- perature increases the output of CO2 from the resting nerve. A respiratory process should increase proportionally to the temperature. Raising of temperature, however, not only increases the rate of res- piration, but also (particularly by sudden changes of it) stimulates the nerve. A very interesting fact is observed in connection with the killing of the nerve. When the nerve is killed gradually by a slow increase of temperature, it gives off more CO2 than when killed suddenly, the determination being made after both are killed. CO2 production from the dead nerve under this condition must be due to the diffusion of the gas which was formed previously, just as Fletcher's dead muscle is charged with C02 gas. The different outputs of CO2 between slowly killed and suddenly killed nerves cannot be accounted for unless we assume that in one case, C02 is produced more while being killed than in the other. Whether such increase of CO2 produc- tion, however, was due to the acceleration of normal respiration by the slowly increasing temperature, or due to direct stimulation caused 34 MATHEWS: loc. cit. Carbon Dioxide From Nerve Fibres 127 by heat, or due to both, cannot be decided here unless we consider the relation between excitation and tissue respiration.35 It is hoped that we may have a better understanding of this matter when we study the temperature coefficient of normal respiration of the nerve. At present, we are satisfied to state only that there is a strong evidence to support the conclusion that heat, ,too, increases C02 production from the nerve. DISCUSSION OF THE RESULTS Comparison of Metabolism of Non-Medullated and Medullated Nerve. — Although it appears ridiculous to attach any significance to the marked similarity in the magnitudes of C02 production from non- medullated and medullated nerves, the temptation is irresistible to comment on the high output of C02 from the non-medullated nerve fibre. Let us study the Table following (Table VIII), in which a summarized comparison is given. TABLE VIII Nerve CO2 from resting nerve • CO2 from stimulated nerve Rate of increase of CO2 Non-medullated (spider crab) Medullated (frog) 6.7 X 107 g. (15° - 16° ) 5.5 X 10-7 g. (19° - 20° ) 16. X 107 g. (14° - 16°) 14.2 X 107 g. (20° - 22°) 2.4 times 2.6 " Since I have found that injury increases the C02 production from the nerve, the values I have obtained from cut, or isolated, fresh resting nerves, such as I had to use, may be somewhat greater than the output of normal uninjured nerves would be. But since Alcock36 has shown that a non-medullated nerve gives a higher electrical response, both in the negative variation and the injury current, the C02 increase due to the cut alone will probably be greater in case of the non-medullated nerve than in that of the medullated one. That means that the value of the CO2 production for the resting uninjured, 35 See p. 134. 36 ALCOCK: Proceedings of the Royal Society, 1904, Ixxiii, p. 166. 128 Shiro Tashiro non-medullated nerve should be reduced more from the figures found for. the isolated nerve, than that of the medullated one. In other words, by lowering 6.7 X io~7 gram which is the value for resting, non- medullated, isolated nerves, the rate of increase of CO2 by stimula- tion in the uninjured nerve would become higher than 2.4 times, and probably higher than 2.6 times, which is the rate for the medullated nerve. This greater effect in the non-medullated nerve is what we should expect if our present conception that conduction is in the axis cylinder only, is correct. Before any accurate comparison of the increase of CO2 production on stimulation of non-medullated and medullated nerves can be made it will be necessary, however, to determine how much of the CO2 from the resting nerve is due to injury alone. Before we consider this point seriously, also, we should deter- mine the metabolic activities of greater numbers of nerves of different animals. Such an investigation is at present useless until we deter- mine more quantitatively the relation between CO2 production and the various strengths of stimulation and the degree of excitability. 'If any uniformity of C02 output in respect to anatomical varia- tions is discovered, light may be thrown on the function of the medullary sheath and other differentiations. However insignificant these results may be as far as the similar rates of the gas production of these two nerves is concerned, it should be strongly emphasized that technical error plays no part in these determinations. Inasmuch, as we are dealing with such an extremely small amount of the gas, it is quite natural for those who are not familiar with my apparati to suspect, by a hasty inspection of my results, that the small differences I found under different metabolic conditions may be due to mere experimental variations. For this reason, particular attention is called to a detailed description of the quantitative method I used, especially the footnote on page 144, where I have cited a series of determinations of unknown quantities of CO2 in testing my apparati. I may repeat here that my experi- ments with the spider crab and the winter skate were done at Woods Hole37 during the summer of 1911, while those with the frog were done in Chicago during the winter of 1912. Under these different conditions, I have not only used the different sizes of nerves, but also 37 I take great pleasure in acknowledging my indebtedness for the kind accom- modation offered me by Drs. Lillie and Drew at Woods Hole. Carbon Dioxide From Nerve Fibres 129 experimented with two different apparati, the respiratory chambers of which have had entirely different capacities.38 Comparison between the Metabolism of Resting Nerves and that of Other Tissues. — To compare the rate of metabolism of the nerve with that of other tissues is a matter of no great physiological value on account of great variations which do not affect equally the rate of CO2 production. Simply to give a better picture of the scope of nervous metabolism, however, let us make the following comparison: Since there is no exact determinations made on either the other organs, or the whole animal, in the case of the spider crab, I have quoted those of the nearest Crustacea of which data are available. (Table IX). TABLE IX Animals C02 per Kg. per hour Temperature Determined by l Crustacea (whole animal) Jolyet and Regnaut Cray fish (Astacus) . Crab (Cancer pagurus) 37.7 c.c. 89 9 c.c. 12°.5 16 1C (( " ..obster (Homarus vulgaris) .... 54.4 c.c. 15 « M « ^erve of spider crab (Labinia cani- liculata) 212 c.c. 15° - 16° Tashiro ?rog: (Rana esculenta) (whole animal) . .082 gms. 17 Schultz (Rana temporaria) (whole animal) .355 " 19° - 20° Pott (Rana pipiens) (sciatic nerve) . .33 " 15 Tashiro (Rana temporaria 2) (isolated muscle) .18 " 21 Fletcher Dog . 1.325 " Regnaut and Reiset .41 " Pettenkoffer and Voit .61 " « « n a it .37 " Speck - All the figures are quoted from Schafer's Text Book of Physiology i, pp. 702, 707 and 708, except that of the isolated muscle which I calculated from Fletcher (toe. a/.). Fletcher fails to state the weight of a leg, but gives the value .2 c.c. for one-half hour. Hill believes that if we take each leg 6 g. in average, the value will not be far from the truth. 2 Fletcher fails to state the species of the frog, but it is inferred from Hill s paper. 38 See the last columns of Table I and Table II. 130 Shiro Tashiro Active Nerves. — That the nerve increases its CO2 production approximately 2.5 times when stimulated, is in accordance with our conception of the metabolism of other acting organs. Just how much increase of CO2 takes place during functional activity of an organ or organisms depends on conditions as well as on habits of different organs and animals. Pettenkofer and Voit 39 report that a man (weighing 70 kgs.) gives off when working 0.76 grams per kg. per hour, while resting only .56 gram. Barcroft40 found that the submaxillary gland when stimulated by the chorda tympani gives off 3-7 times more CO2 than the resting gland. In the case of contracting muscle, the results are very contradictory. Hermann 41 found that the contract- ing muscle gave off 9.3 per cent of CO2 (by volume) while the resting one, only 1.4 per cent. Tissot 42 and other workers also found a similar increase of CO2 from contracting muscle. Minot,43 working with Ludwig, maintains that there is no relation whatever between CO2 production and muscle tetanus. L. Hill 44 and Fletcher 45 both con- firmed Minot's work by finding no increase of CO2 production from muscular tetanus. According to Fletcher, the increase he found in C02 production from a contracting muscle in a closed vessel is due to the rigor. Under this condition, he believes, increased formation of lactic acid is responsible for liberating CO2 already produced. In either case, it is understood that functional -activity in the muscle is accompanied by an increase of metabolic activity. It is difficult to compare this increase of metabolic activity of the muscle with that of the nerve unless we determine how much and what ! ind of metabol- ism takes place in contracting muscle. Respiration Quotient of the Nerve Fibre. — As quoted before Haberlandt found that a resting nerve consumes 41.7 to 83.4 cmm. 02 for i gm. for an hour at 19° - 24°. Although he has not deter- mined chemically the production of CO2 he could easily read the respiration quotient by means of the index fluid. Thus he found 39 PETTENKOFER and VOIT: loc. cit. 40 BARCROFT: Ergebnisse der Physiologic, 1908, vii, p. 735. 41 HERMANN: Stoffwechsel der Muskeln, Hirschwald, Berlin, 1867. 42 TISSOT: Archives de physiologic, 1894-5, (5) vii. p. 469. 43 MINOT: Arbeiten aus der physiologischen Anstalt zu Leipzig, 1868, p. i. 44 L. HILL: See Schafer's Text Book of Physiology, 1898, i, p. 911. 45 FLETCHER: Journal of physiology, 1898-9, xxiii, p. 68. Carbon Dioxide From Nerve Fibres 131 that the respiratory quotient of the resting and acting nerve is nearly unity. Since he found that O2 consumption is increased when stimu- lated, and since the respiration quotient remains constant before and after the stimulation, he concluded that it must give off more CO2 when stimulated. It is very interesting to compare the O2 consump- tion in this experment with the CO2 production of mine.46 Taking his lowest figure, because he worked in 19° - 24° and I in 19° - 20°, 41.7 cmm. of O2 amount to .00007 cc- f°r jo milligrams for ten minutes. My figure of 5.5 X io~7 grams for the same units may be translated to .00027 cc. of C02 (ignoring temperature and pressure C02 00027 correction). Therefore-—— = - - = 3. 8, the respiratory quotient. O2 00007 As I have not determined O2 consumption of the nerve of Rana pipiens, this figure has no particular value, but the fact that the CO2 produc- tion is comparatively higher than 02 consumption is a matter of considerable interest. One of the most important observations made by A. V. Hill 47 is the fact that he could not detect any rise of temperature in a frog's nerve as measured by an apparatus which is sensitive to a change of one-millionth of a degree. From this, according to his calculation, he concludes that not more than one single oxygen molecule in every cube of nerve of dimension of 3.7 //, can be used up by a single propa- gated nerve impulse. Therefore, he suggested that an impulse is not of irreversible chemical nature but a purely physical change. Although, I confess, my ignorance makes it impossible to interpret his valuable results from my observations, I may add that these two apparently irreconcilable facts may throw light on the true nature of nervous metabolism. Dr. Mathews has suggested that metabolism in the nerve may be something of the order of alcoholic fermentation, which is not a direct oxidation, and where heat production cannot be so large as CO2 production, since the energy content of glucose is only a trifle higher than that of the alcohol produced. The compara- tively little heat production in the case of working glands is a matter of interest in this connection. At any rate we should not forget the 46 He used Rana esculenta, which, by the way, gives for the whole animal .082 g. CO2 per kg. per hour at 17° according to Schultz. My frog was Rana pipiens. 47 'HILL: Journal of physiology, 1912, xliii, p. 433- 132 Shiro Tashiro anatomical as well as the chemical differences between muscle and nerve. In this respect the ratio between CO2 production and 02 consumption from the nerve is suggestive. The extremely small intake of O2 has another point of interest in relation to the general nature of irritability. It has been repeatedly reported that a nerve can remain excitable several hours in an oxygen- free atmosphere, although there is no doubt its excitability diminishes, yet there is a considerable amount of evidence to show that oxygen is very closely associated with the state of excitability, To har- monize these two facts, the oxygen-storage hypothesis has been suggested, by which the exhaustion is attributed to complete consump- tion of "the stored oxygen and that excitability is restored when atmospheric oxygen is readmitted. Without committing ourselves to this hypothesis, I may add that according to Haberlandt's figure, the resting nerve of 10 milligrams will consume only .0042 cc. O2 in ten hours. If we take our figure and assume that one volume of oxygen was necessary to produce one volume of C02 (this assumption is made without any significance except to give a liberal estimate), the C02 production would require about .015 cc. of O2 for ten hours. And if we assume again that activity will increase O2 consumption in propor- tion of C02 production, then it means that the nerve when stimulated takes up only .03 cc. of O2 during ten hours stimulation. I am not aware, at present, of the existence of any method which will surely remove O2 as completely as this from a large vessel; and this is a very liberal estimate. My experiences in rendering the air free from CO2 encourages me to raise the question, How can one remove every trace of 02 from a nerve fibre? Without having a correct criterion for an oxygen-free medium we cannot at present consider definitely any question of the relation of O2 to irritability. CONCLUSION In spite of all the negative evidence against the presence of meta- bolism in the nerve fibre, we have established three important facts: namely, (i) A resting nerve gives off a definite quantity of carbon dioxide; (2) stimulation increases CO2 production; and (3) CO2 production from the resting nerve proportionally decreases as irri- Carbon Dioxide From Nerve Fibres 133 tability diminishes. These facts prove directly that the nerve con- tinuously undergoes chemical changes, and that nervous excitability is directly connected with a chemical phenomenon. There is still another question left, namely, Is there any direct relation between excitability and tissue respiration? To put this question more directly, we may ask: Does excitability depend on the respiratory process in the protoplasm? To answer these questions we must refer to two facts; namely the direct relat on between the rate of respiratory activity and the decrease of excitability; secondly, the influence of reagents on CO 2 production and their effects on the state of excitability. By the studies of C02 production by Fletcher 48 lactic acid forma- tion by Fletcher and Hopkins,49 and heat evolution by A. V. Hill,50 it has been established that in isolated muscle, respiratory processes decrease when irritability diminishes. In the case of the nerve, as shown in Table 3, C02 production reaches this minimum when excitability approaches zero. These relations, however, do not show conclusively that the protoplasmic irritability depends on respiratory activity, for it is quite probable that the dying nerve may alter its physical condition as well, which according to the physical school, may consequently alter the state of excitability. That irritability is independent of the respiratory processes has been hitherto successfully contended in the case of the dry seed. The works of Horace Brown, Thisel ton-Dyer 51 and others indicate that the dry seed can be kept alive at the conditions where no ordinary gaseous exchanges are possible. It is argued, therefore, that life is possible without any metabolic activity.52 While a definite poten- tiality for irritability may exist without any metabolic activity, yet that the irritability can persist without respiratory activity, or vice versa, is a matter by no means settled. In the case of ordinary air-dry seed, Waller could demonstrate the response of electrical changes when stimulated although the detection of CO2 was impossi- 48 FLETCHER: loc. cit. 49 FLETCHER and HOPKINS: loc. cit. 50 A. V. HILL: loc. cit. 51 THISELTON-DYER: Proceedings of the Royal Society, 1897, Ixii, p. 160; ibid., Ixv, p. 361. 52 I am indebted to Professor Crocker for his kind suggestion as to botanical literature. 134 Shiro Tashiro sible. This failure, however, as he himself expected, was due to the lack of delicacy of the chemical methods for detecting C02. I ob- served, with my apparatus that even a single kernel of a dry seed gives off a definite quantity of CO2 as long as it is alive. In ordinary condition not only a living dry seed gives off more CO2 than the dead one, but also like the nerve, it always gives off more CO2 when stimu- lated by mechanical injury. In the normal condition, therefore, we may safely conclude, there is always metabolic activity as long as the seed is irritable, and that in the different states of irritability, the respiratory activity is proportionately different. At present, therefore, we have no decided evidence which will prevent us from considering excitability as a function of respiration under ordinary conditions. This relation is more directly studied by the use of anaesthetics. I have already demonstrated that an etherized nerve gives off considerably less CO2 than the normal. Such an etherized nerve will not give more CO2 when it is crushed. This may be interpreted by some to mean that the etherized nerve may be already dead. This, however, is not the case. This objection, also, I have considered by studying the nerve treated with KC1. When the nerve is treated with .2 m KC1 and then crushed, it does not give an increase of CO2 production. Mathews has shown that while a .2 m. KC1 solution renders the nerve unexcitable, yet it will recover its excitability by being replaced into n/8NaCl. These two facts, therefore, support the idea that any agents that suppress excita- bility of the nerves also decrease the C02 production and that C02 production by crushing the nerve must be largely due to stimulation. This hypothesis is strikingly supported by similar observations on the dry seed. Etherized seeds give much less CO2 and cannot be stimulated to give more C02 by crushing, while under normal con- ditions, crushing a seed always increases its CO2 production. Quan- titative experiments in this direction will be given in another paper. These facts directly support Mathews' hypothesis that substances which suppress irritability must act on the tissue respiration pri- marily. If such an hypothesis is correct, we can easily picture what is happening in the nerve fibre. Vernon 53 considers that a tissue contains certain substances which can absorb oxygen from their sur- 53 VERNON: Journal of physiology, 1909-10, xxxix, p. 182. Carbon Dioxide From Nerve Fibres 135 roundings to form an organic peroxide, and by the help of a peroxidase can transer this to amino acid and carbohydrate molecules bound up in the tissue, just as H2 O254 can oxidize, with the help of an activator, an acid of formula R. CHNH2 COOH to C02, NH3 and an aldehyde RCHO, and then oxidize this aldehyde to RCOOH and ultimately to CO2 and H2 O. Poisons such as HNC, NaHS03 and NaF, which he found to decrease CO2 production, temporarily paralyzed respiration, he thought, by uniting with aldehyde groups, while formaldehyde, acid and alkali temporarily paralyze C02 forming power of the tissue by destroying the peroxidase. The organic peroxide, though it can still affect some oxidation, cannot of itself carry it to the final C02 stage. Recovery of CO2 forming power is due to the regeneration of the peroxidase. Although I doubt that such a process occurs in nervous respiration, the idea of two similar metabolic phenomena involved in the nervous metabolism is very helpful to understand the behavior of the nerve during continued activity. Most recently Tait discovered that a refractory period has two phases, absolute and relative.55 When he treated the sciatic nerve of a frog with yohimbine, the relative phase is greatly prolonged, while the absolute one is little affected, a result quite different from other common anaesthetics. Waller56 has already observed that protoveratrin slows up the positive variation of the nerve, while the negative variation is little in- fluenced. Waller contends that this drug does not alter cata- bolic change, but retards anabolic activity to a considerable degree. Since pharmocological action on animals of protoveratrin and yohimbine are very similar, Tait concludes that these drugs must attack the nerve in similar manner, and that a refractory period, too, must consist of two phases corresponding to the catabolic and anabolic processes which Waller observed in the case of protovera- trinized nerves. Thus, he considers that his " absolute phase" of the refractory period corresponds to negative variation or catabolic process of the nerve, and the " relative " to the positive return or anabolic. Yohimbine, in other words, retards anabolic processes con- siderably, thus prolonging the refractory period, or increasing nerve 64 DAKIN: Journal of biological chemistry, 1908, iv, pp. 63, 77, 8 1, 227. 55 TAIT: Journal of physiology, 1912, xl, p. xxxviii. 66 WALLER: Brain, 1900, xxiii, p. 21. 136 Shiro Tashiro fatigue easily. These considerations suggest very strongly that the absence of fatigability in the nerve as measured by the ordinary methods, is not a question of absence of metabolism, but merely the speed by which these two processes come to an equilibrium. Although we have an infinite number of facts still unexplainable, by our present knowledge of nerve physiology, we have established a few new facts around which we may build up some idea concerning this most essential phenomena of living matter, — i.e., irritability. As to the true nature of the nerve impulse, I can only confess my ignorance. SUMMARY 1. All nerve fibres give off CO2. The resting, isolated nerve of the spider crab produces 6.7 X icr7 gram per 10 milligrams per ten minutes. The frog's sciatic 5.5 X io~7 grams. 2. When nerves are stimulated they give off more CO2. .The nerve of the spider crab claw produces 16. X io~7 gram when stimu- lated, the frog nerve 14.2 X icr7 grams. The rate of increase of C02 by stimulation amounts to about 2.5 times. 3. The CO2 output of resting nerve is due to a vital active process. 4. Anaesthetics greatly reduce the carbon dioxide output of nerves and dry seeds. 5. Mechanical, thermal and chemical stimulation also increases the carbon dioxide output of nerves. 6. Single dry living seeds (oat, wheat, etc.) react in most par- ticulars similar to nerves as regards their irritability, relation to anaesthetics, mechanical stimulation and carbon dioxide outputs. 7. The general conclusion is drawn that irritability is directly dependent upon and connected with tissue respiration and is primarily a chemical process. These results strongly support the conception that conduction is of the nature of a propagated chemical change. To Prof. A. P. Ma thews, under whose direction I have carried on these experiments, I express my appreciation and gratitude. For many suggestions, I am under obligation to Dr. F. C. Koch. Reprinted from the American Journal of Physiology Vol. XXXII — June 2, 1913 — No. II A NEW METHOD AND APPARATUS FOR THE ESTIMATION OF EXCEEDINGLY MINUTE QUANTITIES OF CARBON DIOXIDE 1 BY SHIRO TASHIRO [From the Department of Biochemistry and Pharmacology, the University of Chicago, and the Marine Biological Laboratory, Woods Hole, Mass.] IN connection with the study of the metabolism of the nerve fibre, I undertook, at the suggestion of Prof. A. P. Mathews, to work out a method for the detection of exceedingly minute quantities of carbon dioxide. Following a suggestion made by Dr. H. N. McCoy, a very simple method was devised, which I reported first to the Chicago Section of the American Chemical Society; 2 later in con- junction with Dr. McCoy, its further details were reported to the Analytic Section,3 of the Eighth International Congress of Applied Chemistry. The principle of the new method is as follows : 1. Exceedingly minute quantities of carbon dioxide can be precipi- tated as barium carbonate on the surface of a small drop of barium hydroxide solution. 2. When a drop of barium hydroxide is exposed to any sample of gas free from carbon dioxide, it remains perfectly clear, but when more than a quite definite minimum amount of carbon dioxide is intro- duced, a precipitate of carbonate appears, detectable with a lens. 3. By determining, therefore, the minimum volume of any given sample of a gas necessary to give the first visible formation of the precipitate, its carbon dioxide content can be estimated accurately, since this volume must contain just the known detectable amount of carbon dioxide. 1 One of these apparati was described at the biochemical section, Eighth International Congress of applied chemistry, September, 1912; see also, Journal of biochemistry, 1913, xiv, p. xli. 2 May 18, 1912. 8 Original Communication: Eighth International Congress of applied chemis- try, 1912, i, p. 361. 138 Shiro Tashiro I have constructed two apparati, based on this principle, which are especially adapted for the estimation of the output of carbon diox- ide for very small biological specimens. With these apparati, one cannot only detect easily a very small amount of gas, given off by a small dry seed, or a small piece of a frog's sciatic nerve, but can also estimate it with considerable accuracy. The apparatus shown in Fig. i consists of two glass bulbs. The upper bulb A, is a respiratory chamber, having a capacity of about 15 c.c., which can be diminished to 9 c.c. by means of mercury. The lower bulb B is an analytic chamber with a volume of 25 c.c., which can be made to 5 c.c. by filling up with mercury. These two bulbs are con- nected with a capillary stop-cock D. The respiratory chamber is fitted with a tight glass stopper, R, which is connected to a three-way capillary stop-cock C. This glass stopper is so arranged that the chamber can be sealed by putting mercury above the stopper. The tubes are thick walled capillaries of about i mm. internal diameter, excepting upturned tubes inside the bulbs, which should be rather thin walled, especially at F and H, where it is widened to an internal diameter of about 2 mm. It is important that the glass of which these tubes are made should be of a quality not readily attacked by barium hydroxide. The details of the method of procedure are as follows : The apparatus is first cleaned and dried.4 The specimen is 4 The apparatus is made in such a way that it can be cleaned and dried in ten minutes without being taken apart. For this, the stop-cock D is closed and E and L are opened. The arm at L is connected to the suction pump. Then a little acidulated water is introduced through G. By closing E, and opening D and G, the excess of water is drained off. Then the process is repeated with dis- tilled water, alcohol, and alcohol ether. The last drying is completed by passing a current of air through G while D is closed. FIGURE 1 One-third the actual size. Apparatus For Estimating Carbon Dioxide 139 placed on a glass plate 5 and weighed. The glass plate is hung on n and m, which are electrodes fused into the side of the respiratory chamber A. The chamber is now closed with the stopper R and sealed with mercury. Through L, a connection is made with a pump 6 and about 20 c.c. of mercury is introduced through G. Not too much mercury should be used; its surface should not be within 5 mm. of the cup F. Then wash the whole apparatus with carbon dioxide-free air,7 which is introduced through C, by successive evacuations. After the evacuation and washing out with pure air, which is repeated three or four times, the pressure inside of the bulbs is made equal to the atmospheric pressure by adjusting it at the nitro- meter in the usual fashion. Stop-cock E is then closed, and the space between E and L is evacuated so that the barium hydroxide can rush in, a process which is very advantageous to obtain a clear barium hydroxide solution. Then clear barium hydroxide solution is run in through L. By opening E very slowly and carefully, the solu- tion is now introduced into the chamber so that a small drop stands up upon the upturned end of the capillary at F. Then the connection between the two chambers is closed by D. It is imperative that this drop of the solution should be perfectly clear at the start. If no deposit of barium carbonate forms on the surface of the drop within ten minutes,8 a portion of the sample gas is drawn into B by with- drawing mercury through G and opening the stop-cock D. The volume of mercury withdrawn, which may be readily determined by volume, or more accurately by weight, gives the volume of the sample 5 The kind of glass plate used in connection with the nerve and small animals like Planaria is shown on p. 120, Fig. i. (The first paper.) 6 The pump should be capable of giving a vacuum of at least 25 or 30 mm. of mercury. 7 Air cannot be freed completely from carbon dioxide by passing it through wash bottles. In my work, carbon dioxide-free air is prepared by shaking air with twenty per cent solution of sodium hydroxide in a tightly-stoppered carboy, fitted with suitable tubes. When this is to be used, it is driven into a nitrometer which is filled with less concentrated alkaline solution (a weak solution is used so that the chamber may not be too dry) by displacing it by running in a solution of sodium hydroxide. After each evacuation, this air is introduced from the nitro- meter into the chamber A through stop-cock C. 8 If no precipitate appears within ten minutes, it is a sure control that the apparatus is free from carbon dioxide. 140 Shiro Tashiro gas taken from the respiratory chamber, since the pressure in A and B is kept equal to the atmospheric during the transfer. One now watches the surface of the drop at F with a lens to see whether any formation of barium carbonate occurs within ten minutes. With this apparatus, I have repeatedly introduced accurately known quantities of carbon dioxide of very high dilution into B in the manner just described and as a result have found, with remarkable regularity, that i.o X io~7 gram of carbon dioxide is the minimum amount which will cause a formation of barium carbonate within a period of ten minutes. Smaller amounts of carbon dioxide give no visible results; while larger amounts give a deposit more rapidly, and appear in larger quantities. This minimum detectable amount 'i.o X io~7 gram is about the amount which is contained in J c.c. of natural air, in which we assume 3.0 parts of carbon dioxide in 10,000 by volume.9 In order to determine the concentration of carbon dioxide in the respiratory chamber, one must first find, for the apparatus used, the minimum detectable amount of carbon dioxide. Then one finds, by trial,10 the minimum volume of gas necessary to give the first visible formation of barium carbonate. This volume must, therefore, con- tain the known minimum detectable amount of carbon dioxide. From the ratio between this volume and the original volume of the respira- tory chamber, out of which this amount is withdrawn, the absolute 9 LETTS and BLAKE: Proceedings of the Royal Dublin Society, 1899-03, ix, p. 107. 10 In the case of biological problems, when the specimen gives off carbon dioxide continuously, and sometimes at different rates, varying with the time, it is much simpler not to attempt to determine the minimum volume by a continuous trial with the same sample; but instead to repeat the experiments with a series of samples of known weights for a known time, and determine the minimum volumes which give the precipitates, and the maximum volumes which do not give the precipitates. In this way, it can easily be calculated what is the mini- mum volume which gives the precipitate for the given weight of the specimen for a given time. Table I on page 1 14 will illustrate this more clearly. Another upturned cup H provided in the respiratory chamber A is used in case only the qualitative detection of C02 is wanted. In such a case, the perfectly clear barium hydroxide solution is introduced, after the necessary cleaning and washing, to the respiratory chamber, forming the usual drop at H instead of F. It should be noted that in case a smaller capacity is necessary for the respiratory chamber, the mercury is introduced by a pipette to the bottom of the chamber at K. Apparatus For Estimating Carbon Dioxide 141 quantity of carbon dioxide, given by the specimen, may be computed. At the suggestion of Dr. F. C. Koch, another apparatus was con- structed, which provides a control drop of the barium hydroxide solution, side by side with the other. The apparatus (Biometer) shown in Fig. 2, although it appears complex, is nothing more than apparatus i, inclined 90°, but each of its chambers is provided with a barium hydroxide cup d and f . It is made of glass consisting of two respi- ratory chambers, serving also as analytic chambers, connected by a three-way stop-cock L, the other arm of which is connected to one arm of another three-way stop-cock K. Each of the other two arms of stop-cock K is connected to a nitrometer W and X. The nitro- FIGURE 2. Biometer, one-third actual size. The shaded portions of the apparatus indicate the rubber connection which is first coated by shellac, and then sealed with a special sealing wax. also sealed with mercury. Some parts are meter on the right, is connected to a carboy with air free of CO2; and the other, on the left, to a similar reservoir with air free of C02 plus any gas which is desired as a medium for conducting the experi- ment. Chamber A is drawn to a capillary stop-cock C; chamber B is drawn to the three-way stop-cock G, one arm of which is con- nected with a mercury burette T, which is used for adjusting the pressure. Both of the chambers have a capacity of 20 to 25 c.c. and are provided with a pair of platinum electrodes n and m, and also with the glass stoppers S and R, which can be sealed as usual with mercury. The pump is connected through J, and the barium 142 Shiro Tashiro hydroxide solution is introduced through V to d and f, where drops are formed as before. As stated above, this apparatus can be used for the combined purposes of qualitative detection, quantitative estimation, and com- parative determination of the output of C02 from the various biolog- ical specimens. It has a decided advantage over the other in the fact that we have a control drop, side by side, under exactly the same conditions, and that the comparative estimation of C02 produced by different specimens can be made very easily and accurately. The de- tailed method of procedure is described under three different headings : (a) For the Qualitative Detection of Carbon Dioxide. — After the apparatus is cleaned and dried,11 a weighed tissue is placed on the glass plate and hung on n and m of the chamber A, and no tissue in the other chamber. After both chambers are closed with the stop- pers S and R and sealed with mercury, they are so filled with mercury that the remaining volumes in both chambers are now exactly the same. The chambers are now evacuated and washed with pure air. When evacuation and washing with pure air is complete, the pressure is made atmospheric, by adjusting with the nitrometer the connec- tion between A and B is then closed with stopcock L. If any CC>2 is given off by the tissue, the desposit of carbonate will soon appear on d, while in the control chamber the drop on f remains perfectly clear. In order to avoid any possible error of a technical nature this experiment is repeated by exchanging the chambers, now using chamber B for the respiratory chamber and the other A as a control. (b) For Comparative Estimation of CO2 from Two Different Samples. — By repeated quantitative experiments, it was found that the speed with which the first precipitate appears and the sizes of the deposits on the drops at d and f represent corresponding quanti- ties of carbon dioxide. Thus with remarkably simple means, we can determine simultaneously the comparative outputs of the gas from two different tissues or from the same tissues under different conditions. The method of procedure is best illustrated by the following example. Two pieces of the sciatic nerve are isolated from the same frog and exactly weighed. One piece is laid on one glass plate, and the other 11 This, too, can be cleaned and dried without being taken apart. See foot- note on p. 138. Apparatus For Estimating Carbon Dioxide 143 on the other plate in such a way that one part of the nerve lies across the electrodes of the glass plates as shown in Fig. i, page 120. In this way, when the plates are hung on the electrodes i> and m, any desired nerve can be stimulated with the induction current. These- plates are now hung on the electrodes in each chamber, and the usual procedure is followed for the cleaning and the washing of the appara- tus to make it CO2 free. After the connection between the two chambers is closed by means of stop-cock L, the nerve in chamber A is stimulated by the current. Then if one can watch over the surfaces of the drops carefully from the start, he finds the first deposit of the carbonate on cup d of chamber A in which the stimulated nerve is placed. Later, the total amount of the precipitates grows much larger in the case of this cup. This increased output of the carbon dioxide from the stimulated nerve, thus observed, can be duplicated by repeating the similar experiment, after exchanging the chambers, as usual. This comparative estimation can be more accurately made by exact quantitative measurement, the method for which the follow- ing will illustrate. (c) For Quantitative Measurement of Gas. — The detailed method is exactly analogous to that of apparatus i. Here we use chamber B as the respiratory chamber and A as the analytic cham- ber. Barium hydroxide should be introduced into chamber A only at d, and the stop-cock F is always closed except at the time of wash- ing. The pressure should be adjusted by mercury burette T, or by the potash bulb of the nitrometer. In case the mercury burette is used, the remaining volume in the respiratory chamber should be recorded.12 The introduction of a known amount of gas from the respiratory chamber B to the analytic chamber A is accomplished by withdrawing the mercury from C into a very narrow graduated cylinder, while the stop-cocks L G and H are opened. After a quick adjustment of the mercury burette to equalize the pressure, the stop- cock L is closed and the presence of carbonate is looked for exactly in the same manner as described in connection with the other appara- tus, determining the minimum volume that gives the precipitate for the known mass of tissue for a known time. 12 The bulbs are marked at the point where their capacity became 15 c. c. by introducing mercury. The variation of capacity can easily be read by noting the mercury burette. 144 Shiro Tashiro In summarizing, I may emphasize the following points: 1. Particular care must be taken to test the air- tightness of the apparatus. 2. Purifying the air must be done with greatest care, as this is essential. 3. The apparatus must be perfectly dry. 4. A weak suction pump cannot be compensated by frequency of washing. 5. As long as the ratio between the c.c. taken from the chamber and the original volume of the chamber is needed, it is most important to have the pressure in A and B equal to the atmos- pheric. If this is accomplished we can neglect any caution against pressure and temperature variations — a correction which is always necessary for ordinary methods of analysis of exceedingly minute quantities of any gas. In devising this method and in constructing this apparati, I am under great obligation to Professors McCoy and A. P. Mathews and to Dr. F. C. Koch. In order to test the accuracy with which an estimate of concen- tration of carbon dioxide could be made, many determinations were carried out with samples of air which contained accurately known concentrations of carbon dioxide prepared by Dr. F. C. Koch. The experimenter did not learn the concentrations of the samples until after the analysis had been completed. In making up the test sam- ples, pure carbon dioxide, made by heating sodium bicarbonate was diluted with the carbon dioxide free air several times in succession, as illustrated by the following example: 5.5 c.c. of pure carbon diox- ide was diluted to 52.0 c.c. over mercury and thoroughly mixed; 5.5 c.c. of the first mixture was diluted to 52.0 c.c.; i.i c.c. of the second was diluted to 50.7 c.c.; of this third mixture 5.6 c.c. was received from Dr. Koch. I diluted this a fourth time to 255.6 c.c. to form a mixture to be analyzed. The following observa- tions were made: 0.5 c.c. was introduced into the apparatus and pro- duced no precipitate in ten minutes; 0.5 c.c. more of the same sample, gave no precipitation in another interval of ten minutes; 0.5 c.c. more, a total of 1.5 c.c., was run into the bulb. In six minutes the first evidence of a precipitate appeared on the surface of the drop at d of apparatus 2 and in eight minutes was well developed. Since Apparatus For Estimating Carbon Dioxide 145 the amount of carbon dioxide required to give the precipitate is i.o X io~7 grams, this amount is contained in 1.5 c.c. of the sample or i c.c. contained 6.7 X icr8 grams of carbon dioxide. The amount of carbon dioxide actually contained in the sample was 5-5 X 5-5 X 7-1 X 5-6 52 X 52 X 50.7 X 255-6 C'C' = 6" X I0 In six such determinations, all made with samples the concentra- tion of which were unknown to the experimenter at the time of the analysis, the results given in the following table were obtained: Volume of sample re- quired to give a precipitate Weight of carbon dioxide in one c.c. Found Taken 1.0 c.c. 1.0 X 10-7 g. 0.92 X 10-7 g. .5 c.c. 2. X 107 g. 2.3 X 10-7 g. .55 c.c. 1.82 X 10-7 g. 1.83 X 10-7 g. 1.5 c.c. .67 X 10-7 g. 0.62 X 10-7 g. 2.25 c.c. .45 X 10-7 g. 0.45 X 10-7 g. ERRATA IN JUNE NUMBER OF THE AMERICAN JOURNAL OF PHYSIOLOGY (VOL. XXXII, No. II) Substitute " apparatus" for "apparati" in the following places: Page no, lines 7, n, 23. Page 129, line i. Page 137, line 28. Page 144, line 16. Substitute "7.1 cc." for " i.i cc." on page 144, line 29. In figure i, page 120, correct as indicated in the following drawing UNIVERSITY OF CALIFORNIA LIBRARY BERKELEY Return to desk from which borrowed. This book is DUE on the last date stamped below. JUL 1 3 1949 APR 3 1952 LD 21-100m-9,'47(A5702sl6)476 B1OLOG/ LIBRARY G UNIVERSITY OF CALIFORNIA LIBRARY