No. 4243 November 15, 1942 NUCLEAR PHENOMENA IN MOUSE CANCERS By John J. Biesele Herbert Poyner and Theophilus S. Painter PUBLISHED BY THE UNIVERSITY OF TEXAS AUSTIN Publications of The University of Texas PUBLICATIONS COMMITTEE E. J. MATHEWS R.H. GRIFFITH C. F. ARROWOOD C. D. LEAKE D. CONEY A. SCHAFFER A. C. WRIGHT General Publications R. H. GRIFFITH H. R. HENZE LoUISE BAREKMAN A. SCHAFFER FREDERIC DUNCALF E. G. SMITH FREDERICK EBY M. J. THOMPSON Administrative Publications E. J. MATHEWS B. MCLAURIN C. F. ARROWOOD C. D. SIMMONS R. J. WILLIAMS The University publishes bulletins four times a month, so numbered that the first two digits of the number show the year of issue and the last two the position in the yearly series. (For example, No. 4201 is the first publication of the year 1942.) These bulletins comprise the official publica­tions of the University, publications on humanistic and scientific subjects, and bulletins issued from time to time by various divisions of the University. The following bureaus and divisions distribute publications issued by them; communications concerning publications in these fields should be addressed to The University of Texas, Austin, Texas, care of the bureau or division issuing the publication: Bureau of Business Research, Bureau of Economic Geology, Bureau of Engineering Research, Bureau of Industrial Chemistry, Bureau of Public School Extracurricular Activities, and Division of Extension. Communications concerning all other publications of the University should be addressed to University Publications, The University of Texas, Austin. Additional copies of this publication may be procured from the University Publications, The University of Texas, Austin, Texas TH! UNIVERSITY OF TEXAS PRESS ~ The University of Texas Publication No. 4243: November 15, 1942 NUCLEAR PHENOMENA IN MOUSE CANCERS By John J. Bieaele Herbert Poyner and Theophilus S. Painter PUBLISHED BY THE UNIVERSITY FOUR TIMES A MONTH AND ENTERED AS SECOND·CLASS MATTER AT THE POST OFFICE AT AUSTIN, TEXAS. UNDER THE ACT OF AUGUST 24, 1912 The benefits of education and of useful knowledge, generally diffused through a community, are essential to the preservation of a free govern­ment. Sam Houston Cultivated mind is the guardian genius of Democracy, and while guided and controlled by virtue, the noblest attribute of man. It is the only dictator that freemen acknowledge, and the only security which freemen desire. Mirabeau B. Lamar CONTENTS PAGE Foreword ------------------------------------------------------------------------------------------------------------7 Chapter 1-Introductory Preview ------------------------------------------------------------------9 Chapter II. Procedure ------------------------------------------------------------------------------------12 Chapter III. Observations A. Nuclear Volumes ---------------------------------------·--------------------------------------15 B. Mitotic Figure Volumes ______________________________________________________________________ 24 C. The Chromosomes ------------------------------------------------------------------------------27 D. The Plasmosomes ------------------------------------------------------------------------------33 E. Endomitosis ----------------------------------------------------------------------------------------42 Chapter IV. Observations of Other Workers A. Chromosomal Evidence for Polytene Chromosomes_____________________ ___ 49 B. Nuclear Volume Evidence for Polytene Chromosomes ____________________ 51 C. Nucleolar Evidence for Polytene Chromosomes ____ __________________________ 55 Chapter V. Discussion ------------------------------------------------------------------------------------57 Chapter VI. Summary ------------------------------------------------------------------------------------61 Bibliography --------------------··----·--···----·-----·------------···--------·----···--------··-·----·---·----·-· 63 LIST OF FIGURES PAGE Frn. 1. Axes of Nuclear Volume Distribution Histograms .................................. 16 2. Nuclear Volume Distribution in Adult Mouse Liver .............................. 16 3. Nuclear Volume Distribution in Fetal Mouse Liver ................................ 16 4. Nuclear Volume Distribution in Fetal Mouse Skin ................................ 16 5. Nuclear Volume Distribution in Fetal Mouse Kidney ............................ 16 6. Nuclear Volume Distribution in Striated Muscle of 3 Day Old Mouse .. 16 7. Nuclear Volume Distribution in Striated Muscle of Adult Mouse........ 16 8. Nuclear Volume Distribution in Healing Ear Wound of Adult Mouse.. 16 9. Volume Distribution of 1,932 Nuclei of Mouse Tumor XG.................. 17 10. Volume Distribution of 22,390 +145 Nuclei of Mouse Tumor XG.... 17 11. Asymmetrical Mitosis of Tumor-"Diploid" Cell in Telophase .............. 18 12. Nuclear Volume Distribution in Sarcoma 180 .......................................... 20 13. Nuclear Volume Distribution in Tumor 10591A ..........................•......... 20 14. Nuclear Volume Distribution in Transplant of a Spontaneous Mouse Tumor ...................................................................................................... 20 15. Nuclear Volume Distribution in Primary Methylcholanthrene Sar­coma, Mouse 16 ................................................................................., ...... 21 16. Nuclear Volume Distribution in Primary Methylcholanthrene Sar­coma, Mouse 7 .......................................................................................... 21 17. Nuclear Volume Distribution in Primary Methylcholanthrene Sar­coma, Mouse 14 ........................................................................................ 21 18. Nuclear Volume Distribution in Primary Methylcholanthrene Sar­coma, Mouse 3 .......................................................................................... 22 19. Nuclear Volume Distribution in Primary Methylcholanthrene Sar­coma, Mouse 5 .......................................................................................... 22 20. Nuclear Volume Distribution in Primary Methylcholanthrene Sar­coma, Mouse 2....... ........... ........ ................................................................ 22 21. Mitotic Figure Volume Progressions in Normal and Cancerous Tissues .. 26 22. Middle Prophase, Subdiploid, of Tumor XG............................................ 28 23. Middle Prophase, Diploid, of Fetal Mouse Liver ...................................... 28 24. Metaphase, Diploid, of Tumor XG............................................................ 28 25. Metaphase, Diploid, Fetal Mouse Liver .................................................... 28 26. Metaphase of Fetal Liver............................................................................ 30 27. Metaphase of Healing Ear Wound............................................................ 30 28. Metaphase in Muscle Cell in Methylcholanthrene Tumor.................... 30 29. Metaphase in Methylcholanthrene Tumor.............................................. 30 30. Metaphase in Methylcholanthrene Tumor................................................ 30 31. Metaphase in Transplant of Spontaneous Tumor.................................... 30 32. Metaphase in Tumor XG............................................................................ 30 Fm. PAGE 33. Metaphase in Transplant of Spontaneous Tumor____ _ ___ _____________ ______________ 30 34. Metaphase in Methylcholanthrene Tumor__ _______________ _ __ __ _____________________ 30 35. Class I Liver Nucleus, 4 Plasmosomes__ __ ____ ____ ________ ___________ __ _ __ _______ ____ ___ _ 37 36. Class I Liver Nucleus, 1 Compound Plasmosome______ ________________ __ ______ ___ 37 37. Class II Liver Nucleus, 8 Plasmosomes__________________ _ ______________________________ 37 38. Class II Liver Nucleus, 4 Plasmosomes, Some Compound___ __ __ ___ ____ _ ____ 37 39. Class II Liver Nucleus, 1 Compound Plasmosome___ ____ __ _______ ___ ____ _ ____ __ __ __ 37 40. Diagram of Nucleolar Relations in Simple and Polytene Chromo­somes --------------------------------------------------------------------------------------------------------39 41. Typical Class I Resting Nucleus, Lightly-staining, of Embryonic Mouse Liver --------------------------------------------------------------------------------------------43 42. Class I Nucleus, Transitional from Resting to Prophase, of Embryonic Mouse Liver --------------------------------------------------------------------------------------------43 43. Early Prophase Class I Nucleus, in Optical Section, of Embryonic Mouse Liver ----------------------------------------------------------------------------------------43 44. Middle Prophase of Class I Nucleus in Mitosis, in Optical Section, of Embryonic Mouse Liver--------------------------------------------------------------------------43 45. Endomitotic Metaphase of Class I Nucleus of Embryonic Mouse Liver ----------------------------------------------------------------------------------------------------------43 46. Class I Resting Nucleus, Lightly-staining, of Tumor XG______________________ 44 47. Class II Resting Nucleus, More Chromatic, of Tumor XG_____ ______ __ _______ _ 44 48. Class II or III Nucleus, Transitional to Prophase, of Tumor XG____ __ __ _ 44 49. Very Early Prophase of Class II Nucleus of Tumor XG______ __ ____ _ ______ 44 50. Early Prophase of Class II Nucleus of Tumor XG___ __ __ __ ___ __ ___ __ __ __ __ _______ 44 51. Early to Middle Prophase of Class III Nucleus of Tumor XG__ __ ___ _ ___ 44 52. Endomitotic Metaphase of Class II Nucleus of Tumor XG______ ____ ___ __ ___ __ 44 53. Endomitotic Metaphase of Class II Nucleus of Methylcholanthrene Tumor, Mouse 2--------------------------------------------------------------------------------------44 54. Prometaphase of 885 Chromosomes in a Primary Methylcholanthrene Sarcoma ---------------------------------------·-----------··------------------------------------------·----46 55. Abortive Mitosis in Tumor XG. Chromosome Bundles Scattered through the Cytoplasm.................--·---·---------------·---·--------------------------------47 FOREWORD The following study of the cytology of mouse cancer was undertaken at the suggestion of Dr. Herbert Poyner, M.D., of Houston, Texas, and has been carried out under a private subsidy provided by him. Dr. Poyner' s interest in this problem and his point of view in wanting the work done at The University of Texas are well expressed in the following excerpt taken from his records and written in 1936: "While a medical student in 1923 in a course in Pathology, I was struck by the remarkable difference of the internal organization of cancer cells as compared to normal cells of the same tissue. This difference in organization being witnessed by increase in size of the cancer cells, together with an increase in the size of the chromosomes and in certain instances, what appeared to be a definite increase in the number of chromosomes in the cancer cells. It was my opinion that this change in cell organization, if properly studied, might throw considerable light on some of the fundamentals of cancer, but in order to study this, I felt that it would be necessary to have this work done by a geneticist, and for this reason the problem was discussed with the Depart­ment of Zoology of The University of Texas in 1926, again in 1931, and again in 1935, but at that time the Department apparently was not ready to accept such a study . . ;" Our earlier reluctance to undertake the direction of a study of cancer cytology was not due to a lack of an appreciation of the importance of the problem, or of Dr. Poyner' s generous offer to support the work financially, but to the lack of any tangible lead or clue which gave a reasonable promise that we could do anything more than a number of pathologists had already painstakingly done, that is, describe the extremely varied cytological picture presented by malignant growths. When Dr. Poyner visited us again, in the spring of 1939, and discussed his ideas and his wish to see some cancer work go forward here, the situation had changed. Cytological researches since 1935, some of which were made in this laboratory, were beginning to make it clear that the growth of a cell can be attended either by ordinary cell division (mitosis), the usual way, or by a division of the chromosomes within the nucleus (by "endomitosis") without a breakdown of the nuclear wall. Follow­ing the latter process there is a doubling of nuclear volume and an increase in cell size. If a number of these intranuclear division cycles occur in succession the cell becomes very large, the nucleus enormous and often highly basophilic (e.g., the salivary gland cells, or the nurse-cells in the ovary, of the fruit fly) . Although the details of the endomitotic cycle were worked out largely on normal insect larval tissues there was no reason to suppose that this process was restricted to the insects or the invertebrates. Indeed, it was the earlier work of Jacobj dealing with nuclear volumes in normal mouse liver which gave us the first hint of what was happening in insect larval tissues, and it seemed highly probable, to the writer at any rate, that the mechanism of the "innere Teilungen," which Jacobj postulated to account for the rhythmical doubling of nuclear volumes in the mouse liver, was endomitotic in character. The writer showed Dr. Poyner a preparation of the ileum of an old mosquito larva and he was struck with the similarity between many of the cells in the mosquito gut and those he had seen in some human cancers. Might they both have arisen by the same mechanism? This question seemed to justify an investigation. The mouse was selected as the best material for the first study. There were several reasons for this. First of all, a considerable number of spontaneous and induced cancers, readily propagated by transplantation, were available and easily obtained for study. Furthermore, the chromosome number of the mouse is relatively low ( 40) for placental mammals and a number of studies had been made of mouse chromosomes at this laboratory so that we were familiar with the morphology of the individual elements. Mr. John J. Biesele was selected by Dr. Poyner and several members of our staff to carry on this work. His interest in cytology, his command of languages, his thorough training in chemistry and his willingness to devote the necessary time for special training needed in this field made him unusually well suited to undertake the task. As I have indicated, Dr. Poyner has borne all expenses incident to this study including the employment of Dr. Biesele over a period of three years, first to prepare himself and then to carry on the work. The University of Texas has furnished the space needed, the equipment and the ordinary laboratory supplies. T. S. PAINTER, Dept. of Zoology, The University of Texas. CHAPTER I INTRODUCTORY PREVIEW The confusion evident in the literature on the cytology of malignant growths can be resolved into reasonable order, we believe, by the findings here detailed. We are presenting an explanation for the generally reported larger size of the cancer cell nucleus, for the greater number of its nucleoli, for the reports of reduction divisions and diakinesis-like stages with the formation of chiasmata, and for the precocious split of the chromosomes and their conspicuous doubleness at metaphase. The explanation is simply that the individual chromosome in cancer is a multiple structure. The proof of this follows in Chapter III. In this first chapter we shall summarize what is presented in detail in the rest of the paper. The nuclei of a typical mouse tumor present a highly diversified appearance. There is great variability in nuclear size: nuclear volumes range from those of small, normal diploid nuclei to volumes 100 times as great, but a statistical study indicates that a large majority (72 per cent in one tumor) of the nuclei have a volume about double that of ordinary normal nuclei (Chapter III, A). In the typical tumor there are nuclei with all grades of chromaticity. Many mitotic figures (Chapter Ill, B) are encountered, and they contain chromo­somes (Chapter III, C) which vary in size and in number (mitoses with 25 to 885 chromosomes were found. The diploid chromosome number in the mouse is 40). For the greater part, the mitotic spindles are normal am­phiasters, but there are occasional tri-and multipolar spindles. Although the typical cancer nucleus possesses only about the diploid number of chromosomes, it has a volume (some 700 cubic micra), when fixed in the resting stage, about twice that of the normal diploid nucleus ( approxi­mately 350 cubic micra), examples of which are present in most embryonic cells and in the blood cells and connective tissue cells of tumors. The typical cancer chromosomes in meta phase are twice as large ( 1.1 cubic micra) as ordinary chromosomes (0.5 cubic micron) and are made up of twice as many chromonemata, a fact which is shown by a doubling of the number of nucleolar organizers (which produce the plasmosomes) ; there are 4 nucleolar organizers in normal diploid nuclei and 8 in cancerous diploid nuclei (Chapter III, D). The enlarged and often highly chromatic condition of the nuclei in malignant tissue has long been noted by pathologists, and a few investigators have pointed out that the chromosomes in tumors of man and other animals are larger than the chromosomes in mitoses in normal tissue, but until recently cytologists have understood neither the meaning nor the cause of such phenomena. We now know that, as a normal process, it is possible for certain structures (specifically, chromosomes) within a cell to grow and divide without a breakdown of the nuclear wall or division of the cell as a whole. This process is called endomitosis. Briefly stated, during endomitosis the chromo­somes undergo a division cycle which more or less parallels that of normal mitosis, but there is no spindle. There is a coiling of the threads, a pronounced increase in chromaticity, a slight separation of the chromatids, and subsequently an uncoiling of the threads to form the usual resting nucleus, which assumes a volume twice that which it previously had. Thus we have a mechanism for the rhythmic nuclear growth by doubling of volume which Jacobj (1925) invoked to explain the presence in normal tissues of various classes of nuclei whose average volumes bear a 1 : 2 : 4 : 8 relationship to one another. In some instances the splitting of the chromosomes in endomitosis is complete (i.e., the point of spindle fiber attachment-the centromere-divides) and, as a result, a polyploid nucleus is formed with double the normal number of chromosomes and double the volume of the original nucleus (Geitler, 1939, described this process in the water-strider, in which polyploidization is marked in the resting nucleus by the increase in number of discrete heterochromatic X-chromosomes). In other tissues the centromere apparently does not divide in endomitosis and the chromatids remain attached and form diplochromo­somes, or, after several division cycles, polytene (many-threaded) chromosomes (e.g., the salivary chromosomes of Drosophila and some other flies). Since the typical cancer nucleus has diplochromosomes and a volume double the normal, we infer that such a nucleus has undergone an endomitotic cycle without a division of the centromeres, and the enlarged size of the cancer nucleus is not ordinarily the result of an abortive mitotic division, with fusion of the anaphasic or telophasic nuclei, as some investigators have thought. Endomitosis (Chapter III, E) of either sort-with or without division of the centromeres-can occur several times in the same nucleus, and by this process the nuclei of higher volume-classes can be derived from the typical cancer nuclei. Nuclei of both class II and class III in tumors have been seen in a heavily chromatic condition interpreted as the endomitotic metaphase. Several endomitotic cycles with inhibition of centromere division can give rise to very large chromosomes with many strands (the largest chromosomes found averaged about 6.2 cubic micra, which represents the product of three or four endomitotic cycles). Several successive endomitoses, with division of the centromeres in each cycle but the first, can lead to highly polyploid nuclei with diplochromosomes (the largest mitotic figure found in our mouse cancers was between 32-and 64-ploid, with 885 chromosomes of a mean volume double that of normal embryonic chromosomes) . In tumors the nuclei often go through normal mitotic cycles, and in all these mitoses the chromosomes are larger than normal (except in mitoses of the supposedly normal small connective tissue cells). The varied numbers and sizes of the chromosomes in the cancerous mitoses surely indicate that the two sorts of endomitotic growth are independent of one another to the extent that a nucleus which has undergone one type of endomitosis can later undergo either type. The frequent chromosome numbers which are not euploid multiples of the diploid number may have arisen in several ways. Multipolar spindles and asymmetric mitoses can cause aneuploidy, and the polytene nature of the cancer chromosomes is also likely to lead to aneuploidy because of irregularities in the manner in which the many-stranded chromosomes may disjoin in mitosis. The polytene nature of the chromosomes of mouse cancer accounts for the larger nuclear volume, the greater amount of chromatin in the prophase­anaphase stages, and the increased number of plasmosomes. These cytological sequelae of polytene chromosomes have been seen by others in many sorts of tumors in many organisms (Chapter IV). However, the presence of polytene chromosomes in normal tissues has been demonstrated or inferred in a number of cases, and polytene chromosomes have been experimentally produced with­out formation of a tumor (Chapter V). Hence we may conclude that, while polytene chromosomes are a frequent and perhaps invariable concomitant of cancers, they hardly cause tumors of themselves. The apparent near-universality of polytene chromosomes in cancer indicates that carcinogenesis is somehow closely linked to the reduplication of the chromonema and non-division of the centromere which together produce the many-stranded chromosome (Chapter V). Since the polytene condition of chromosomes is basically similar to the polyploid, cancer cells should be expected to evince differences in the produc­tion of various substances, much as polyploid cells are known to do (Chapter V). The frequency of endomitosis and the relative abundance of nucleic acids in a tumor seem to be proportional to the degree of malignancy of the tumor (Chapter V) . CHAPTER II PROCEDURE Ten malignant tumors of the mouse were used. The tumor most extensively studied is that known as XG, which was produced by Selle, Brindley and Spies ( 1941) by the injection, into other C3H mice, of apparently normal liver cells from mice that bore methylcholanthrene-induced tumors. The XG tumor arose at the point of injection, and its transplants were used in this investigation. Tumor XG is described as a polymorphous cell fibrosarcoma. The other mouse tumors studied included the following: transplants of tumor 10591A on albino mice obtained from the Roscoe B. Jackson Memorial Laboratory; transplants of Crocker mouse sarcoma 180 obtained from Dr. F. C. Wood at Columbia; a transplant of a spontaneous mouse tumor furnished by Dr. W. A. Selle of The University of Texas; and six primary tumors induced by methylcholanthrene in C3H mice, also from Dr. Selle. The primary methylcholanthrene tumors seemed to be sarcomas, and several of them were found to contain dedifferentiating striated muscle cells. Normal tissue used for comparison was taken from some of the cancerous animals and from embryonic, new-born and adult mice of albino laboratory stock. The animal was killed by a blow on the head, and the desired tissue was quickly cut out, minced, and fixed. The slides were prepared by the method described by Painter (1939) for mammalian testis. The method involved fixation in acetic-alcohol for a quarter of an hour, staining in aceto-carmine for half an hour, and pressing out small bits of tissue under coverglasses on slides. The preparations were then dehydrated in 95 per cent alcohol and mounted in Euparal or diaphane. Other methods of preparing the tissues were of subordinate importance. The slides were examined under oil immersion with the aid of green filters. Drawings were made and measurements taken by means of a camera lucida so arranged that the microscopic image was enlarged 2,500 times when pro­jected onto the drawing board on which the microscope was placed. Each millimeter on the table represented 0.4 micron on the slide. Measurements of the two horizontal dimensions of such solid objects as nuclei were made with a thin ruler under the mirror of the camera lucida, and the fine adjust­ment micrometer screw of the binocular microscope gave the vertical dimension, although with questionable accuracy. Since resting nuclei are preserved by the squash method as thick, rounded disks, they may be regarded for purposes of volume-calculation either as flattened ellipsoids or as very short cylinders. It was found by use of plastic models that when spheres are pressed between parallel planes their volumes can be approximated fairly well by averaging the ellipsoidal and cylindrical volumes. There is doubtless considerable error involved in this method of calculation of nudear volume, since the nuclei are often of irregular shape. As observed, the vertical radius (i.e., that perpendicular to the slide and the cover glass) is usually between 2 and 3. 5 micra, while the horizontal radii range in length between the value of the vertical radius and 35 micra. A typical nucleus of 757 cubic micra in average volume has three radii measuring 3, 6, and 8 micra. When counts and measurements of nuclei were made to obtain the relative frequencies of given nuclear types, the mechanical stage was employed in such a manner as to insure that no nucleus was measured twice. Randomness in choice of nuclei was made certain by considering all those nuclei, and only those, which came within a certain blocked-off area in the center of the field of view. Most study was devoted to tumor XG. For each of 20 regions scattered through one of the transplants, some 100 nuclei were picked at random and measured, their plasmosomes or chromosomes counted (where possible) , and their degree of chromaticity and mitotic stage noted. The average of the cylindrical and ellipsoidal volumes was calculated for each nucleus and mitotic figure. Then further study was made of each region in that about 1,000 nuclei, also picked at random, were quickly classified as to mitotic stage; hence the percentage of mitosis and the percentage of each phase of the mitotic cycle could readily be calculated. Since a sample of 100 nuclei does not well disclose the frequency of the rare large nuclei which are highly polyploid, a special count was made of resting nuclei alone. All the resting nuclei were counted, but only those above the arbitrarily set volume of 2,000 cubic micra were measured. This necessarily involved some error in selection. However. the drawing board under the camera lucida had marked on it a square of such dimensions that when the nuclear wall touched the sides of the square the nucleus was 2,000 cubic micra in volume, if its vertical radius was 3 micra long. From about 1,000 to 2,000 nuclei per region were studied for this last determination of high polyploidy. Counts were made of the plasmosomes in the resting nuclei measured at random. In several of the tumors the plasmosomes were not as distinctly preserved as in others (particularly tumor XG) and in normal liver. The apparent fusion of plasmosome initials was noted where observed. For each tissue, volume-measurements of sufficient mitotic figures in the various stages were made to establish the average volumes of the phases in the mitotic cycle. In those stages in which no nuclear membrane is present, as from middle prophase to late telophase, the diameters were measured between the outermost points of the chromosome mass. Camera lucida drawings were made of some metaphases and prometaphases in which the chromosomes were free enough of one another for the boundaries of most of them to be distinguished. The volumes of the chromosomes were calculated as follows: in a given figure enough chromosomes usually lie parallel to the plane of focus so that the average chromosome width can be easily measured; the average chromatid radius was one fourth of this average chromosome width; the total volume of the metaphase chromosome, con­sidered as two parallel cylinders, was then twice the volume of a single chromatid, or 2 pi r2 h, where r is the length of the chromatid radius and h is the chromosome length. In a given metaphase figure, the smallest chromo­some may be one fourth the volume of the largest; therefore the average chromosome volume, calculated for each metaphase figure, served as the means of comparison. Similar studies, which were much less comprehensive, were made of the other nine tumors, of normal adult liver, of striated muscle, of the hetero­geneous tissues at the cut edge of an ear, and of normal embryonic liver, kidney, and skin. CHAPTER III OBSERVATIONS A. NUCLEAR VOLUMES In order to test the accuracy of the previously described method for volume measurement, a short study similar to that of Jacobj (1925) has been made on the liver of the adult mouse. The results may be seen in figure 2. The frequency-maxima of nuclear volumes fall at 200-250, 500-550, and 900­950 cubic micra, which are to each other about as 1 : 2.3 : 4.1. Considering the small number of nuclei measured (132) this deviation from the ideal of 1 : 2 : 4 is well within the limits of error. Similar measurements made on liver, belly skin, and kidney of an embryonic mouse, and on striated muscle of the diaphragm of a three day old mouse, show an essential agreement with one another, as may be seen in figures 3, 4, 5, and 6, respectively. The values are distributed about normally around single maxima of frequency, which are 350-400, 250-300, 300-350, and again 300-350 cubic micra for the liver, skin, kidney, and muscle, in that order. These class I modes are higher than that in figure 2 because the liver in the adult studied had rather small nuclei, perhaps as a result of the XG trans­plant, located elsewhere on the animal, making heavy inroads into the avail­able nutriment. Note that in the fetal liver, fetal kidney, and diaphragm muscles there are a few large nuclei which belong in class II. The fetal liver is hematopoietic and contains megacaryocytes with giant nuclei; some selected nuclei not measured at random had volumes of 1,040, 1,240, 1,710, 3,530, and 4,200 cubic micra. In adult tissues the frequency of nuclei of class II and of higher classes is increased over that in embryonic tissues. This is seen not only in adult liver, but also in adult striated muscle (fig. 7), which has a frequency-maximum at 350-400 cubic micra (class I) and another at 650-700 cubic micra (class II). Resting nuclei in close proximity to mitotic nuclei in an ear 12 hours after it was cut show a class I mean volume near 300 cubic micra and a class II mean at 600-650 cubic micra (fig. 8). A survey of the slides made of this ear tissue revealed no nuclei larger than those of class II. When measurements of volume are carried out on the resting nuclei of a tumor, a somewhat similar grouping into classes appears, although the classes are far from being as distinct as in normal tissues. Figure 9 gives the volume distribution of 1,932 resting nuclei of tumor XG. The nuclei were selected at random, except that many of the nuclei of the more easily recognized blood cells like small lymphocytes and polymorphonuclear leucocytes were dis­regarded. The mode of all nuclear volumes in tumor XG is at 650 to 700 cubic micra, which is also the average of class II. The volumes in class I have an average value of nearly 300 cubic micra, which is about half the average 24 w -' ADULT LIVER <.) ::J AXES 16 132 NUCLEI z "­ a: w m ~ ::J z NUCLEAR VOLUME IN CUBIC MICRA 200 400 600 800 1000 1200 F1_9, I FIG. 2. 24 FETAL SKIN FETAL LIVER 92 NUCLEI 16 8 2 200 400 600 800 200 400 600 800 FJG.3. FIG. 4" 24 24 MUSCLE, STRIATED, 16 FETAL KIDNEY 3 DAY OLD MOUSE 31 NUCLEI 94 NUCLEI 16 8 2 200 400 600 800 200 400 600 800 1000 FIG.5. FIG . 6. 24 24 MUSCLE, STRIATED, HEALING EAR CUT, ADULT MOUSE 16 ADULT MOUSE 57 NUCLEI 32 NUCLEI 400 600 800 1000 200 400 600 800 FIG.7. FIG.8. Fig. 1. Axes of Nuclear Volume Distribution Histograms. Fig. 2. Nuclear Volume Distribution in Adult Mouse Liver. Fig. 3. Nuclear Volume Distribution in Fetal Mouse Liver. Fig. 4. Nuclear Volume Distribution in Fetal Mouse Skin. Fig. 5. Nuclear Volume Distribution in Fetal Mouse Kidney. Fig. 6. Nuclear Volume Distribution in Striated Muscle of 3 Day Old Mouse. Fig. 7. Nuclear Volume Distribution in Striated Muscle of Adult Mouse. Fig. 8. Nuclear Volume Distribution in Healing Ear Wound of Adult Mouse. volume of the class II nuclei. Class III is not distinct, since it is greatly out­numbered by class II and overlain by nuclei in the upper range of volume­variation in class II. The greater frequency at 1,300-1,350 cubic micra might represent the mode of class III. Figure 10, which contains the volume distribution of 145 resting nuclei larger than 2,000 cubic micra out of a total of 22,5 35 resting nuclei counted, shows that there is a well-spread maximum from 2,000 to 2,900 cubic micra, 0 .~ 0 ~ Fig. 9. Volume Distribution of 1932 Nuclei of Mouse Tumor XG. Fig. 10. Volume Distribution of 22,390 + 145 Nuclei of Mouse Tumor XG. with a mean at 2,300 to 2,400. There is another group of nuclei ranging from 3,900 to 6,000 cubic micra, with mean at 4,800 to 4,900, and finally there is a group of three large nuclei whose volumes are 8,680, 9, 710, and 11,900 cubic micra and whose average volume is 10,100 cubic micra. These three groups make up classes IV, V, and VI, respectively. If the average of class I (exclusive of the lymphocytes, whose nuclei are condensed and heteropycnotic) be 300-350 cubic micra, then the averages of the higher classes should be 600-700 (observed, 650-700), 1,200-1,400 (ob­served, 1,300-1,350), 2,400-2,800 (observed, 2,300-2,400), 4,800-5,600 (observed, 4,800-4,900), and 9,600-11,200 (observed, 10,100) cubic micra because of progressive doublings of the quantity of chromosomal material. The observed means agree fairly well with the expected values. The variation in volume within a single nuclear class in normal tissue is probably in part the result of differences in degree of hydration under varying external conditions and in different portions of the resting stage of the mitotic and endomitotic cycles. The variation could also stem from differences in material content other than water as the resting nuclei enlarge preparatory to another division. But tumor tissue exhibits less marked frequency-maxima in volume distribu­tion than does normal tissue and has greater variation in an apparently single class of nuclei. The curve is flattened because of the presence of aneuploid nuclei with chromosome numbers falling between the euploid numbers (see the section on chromosomes, p. 27, and see also Winge, 1930). Although most observed diploid or approximately diploid mitoses in tumors are bipolar and symmetric, there are occasional asymmetric mitoses and some multipolar mitoses, the latter being most clearly observed in sectioned material. The literature on cancer cytology frequently mentions asymmetric and multipolar mitoses. As the result of an asymmetric mitosis there are formed two nuclei, one larger and one smaller than usual; an example may be seen in the telo­phasic cell of figure 11. Evidently these two telophasic "nuclei" are sisters, yet one aggregation of chromatin measures 226 cubic micra, and the other 99; both figures are equally condensed. The usual diploid figure at this stage in cancer has a volume of about 150 cubic micra. Multipolar mitoses also 10,.. FIG. II. Asymmetrical Mitosis of Tumor­"Diploid" Cell in Telophase. yield nuclei of abnormal content and usually result in reduction of polyploidy. It is demonstrated later in this paper that cancer chromosomes are made up of more strands than are normal chromosomes; one expects from this polytene condition certain difficulties in disjunction, leading to occasional aneuploidy. Besides aneuploidy, the possible causes of the absence of distinct maxima in the upper classes of tumors are perhaps the facts that: ( 1) there are relatively few nuclei in these classes, so a large error in sampling is likely; (2) the greater the volume of a nucleus, the greater is the range of its possible physiological variations in volume; and (3) the volume step of 50 or 100 cubic micra used in classifying is a far smaller fraction of the volume of the nucleus in the higher volume range than it is in the lower. The spread in the special case of the nuclei in class I is partly the result of heterogeneity. Class I contains blood cells and endothelial cells in addition to cells which seem, from their nuclear appearance, to belong in the tumor. Enough of the small blood cell nuclei were measured to give the frequency­maximum at 200 to 250 cubic micra. In class I there are also various atypically pale (non-staining) nuclei which may be of normal diploid cells somehow included in the cancer and nearly dead; their volumes fall into the group around 300 to 3 50 cubic micra. These nuclei were all measured, as were all the small nuclei whose appearance was that of a diminutive of the larger resting nuclei of the more abundant variety in the sarcoma, except that the small nuclei had a maximum of only four plasmosomes while the predominant tumor nuclei had a maximum of eight nucleoli. Some of these small nuclei may be considered to be normal diploid nuclei of the stroma; perhaps some are nuclei of fibro­blasts, others of endothelial cells, and since there is considerable necrosis in tumor XG, many of them are no doubt macrophages or some other form of the mononuclear round cell. W. H. Lewis (1927), and Lewis and Lewis (1932) report the presence in sarcoma cultures of dividing normal mon­onuclears which are sometimes more common than the malignant cells. Some of the small nuclei in tumor XG are probably derivatives of cancer cells, either as restitution nuclei made from a few lagging chromosomes, or as products of asymmetric and multipolar mitoses. The nuclei in class II are of the prevailing type in cancer and comprise a more homogeneous group. Their chromaticity varies, probably with the mitotic or endomitotic cycle, but none of them is as pale as the "ghost" nuclei in class I. Those nuclei which are still larger differ if at all only in their grossness from the nuclei of the second class. They may have thicker chromatin strands visible in the resting stage, they have usually a higher plasmosome number (some or all of the plasmosomes being rather large), and they may have patches of heterochromatin which in some cases are obviously compound, since they seem to be made up of seriate alveoli. The other nine tumors show resemblances to tumor XG in the distribution of their huclear volumes. The data are presented in figures 12 through 20. 24 S.A,?COMA 180 97 NUCLEI 16 FIG. 12. 2 TUMOR 1059/A 88 NUCLEI 16 8 I FIG . 13. 2 SPONTANEOUS TUMOR 94 NUCLEI 16 8 1700 1900 2750 3100 4300 4850 5250 ... ·--·---·--·--· . Fig. 12. Nuclear Volume Distribution in Sarcoma 180. Fig. 13. Nuclear Volume Distribution in Tumor 10591A. Fig. 14. Nuclear Volume Distribution in Transplant of a Spontaneous Mouse Tumor. 2 2 Fig. 15. Fig. 16. Fig. 17. METHYLCHOLANTHRENE TUMOR, MOUSE 16 100 NUCLEI __ ,,o~---~so'8oo tlG. 15. METHYLCHOLANTHRENE TUMORMOUSE 7 1 113 ~ 5 NUCLEI METHYLCHOLANTHRENE TUMOR, MOUSE 14 108 NUCLEI .._._______._.__.__.__. 7:1) 900 KXJO 1100 1;350 l5!io 1800 21.XJJ Z2iXI 2500 200.J 5000 f'IG.17. Nuclear Volume Distribution in Primary Methylcholanthrene Sarcoma, Mouse 16. Nuclear Volume Distribution in Primary Methylcholanthrene Sarcoma, Mouse 7. Nuclear Volume Distribution in Primary Methylcholanthrene Sarcoma, Mouse 14. 16 8 2 ..........600 400 24 16 8 2 200 400 600 2 16 8 2 200 400 600 MErHYLCHOLANTHRENE IUMOR1 MOUSE 3 86 • 6 NUCLEI 4800 5600 8600 -~-------------· -• -111111-· -· -·-­ 800 ·1000 1200 1400 l500 1600 2000 2500 3700 SKJO 6500 17000 FIG. 18. METHYLC.HOLANTHRENE TUMOR, MOUSE 5 100 • 6 NUCL El __-• --111111--111111--111111--· --· --. 800 1000 1200 1600 2000 2400 3000 3800 4100 7000 FIG .19. METHYLCHOLANTHRENE TUMOR, MOUSE 2 91 NUCLEI. MUSCLE NUCLEI IN WHITE. 800 1000 1200 FIG. 20. Fig. 18. Nuclear Volume Distribution in Primary Methylcholanthrene Sarcoma, Mouse 3. Fig. 19. Nuclear Volume Distribution in Primary Methylcholanthrene Sarcoma, Mouse 5. Fig. 20. Nuclear Volume Distribution in Primary Methylcholanthrene Sarcoma, Mouse 2. It should be noted that in the methylcholanthrene-induced tumor on mouse 16 (fig. 15) the small nuclei, probably for the greater part of normal diploid cells of the stroma, make up about one third of the total. In the primary methylcholanthrene tumor on mouse 14 (fig. 17) the pre­dominant nuclei are those in class I. This observation coincides with those of Lewis (1927), who saw in some of his sarcoma cultures more small, normal round cells than malignant cells proper. The three primary methylcholanthrene tumors on mice 3 (fig. 18), 5 (fig. 19), and 2 (fig. 20) contain varying proportions of dedifferentiating striated muscle cells whose nuclei are rounding up (losing their polarity) under the influence of the hydrocarbon. They may be expected to confuse the picture, because, as is shown in figure 7, normal striated muscle nuclei of adult mice have volumes which fall at least into classes I and II. The pale nuclei of a dedifferentiating muscle cell in the tumor on mouse 5 (fig. 19) were found to belong in class I, since their calculated volumes were 288, 318, 321, 354, and 365 cubic micra. The resting nuclei of the dedifferentiating striated muscle are considerably less chromatic than the nuclei, many of which are presumably malignant, lying free of muscle fibers in the preparation. A much greater proportion of the muscle nuclei seem to be nonmitotic than is true for the free nuclei. Those muscle nuclei which are in mitosis have small chromo­somes like embryonic cells; data on the chromosomes are given later. W. H. Lewis (1939b), in describing the histology of dibenzanthracene mouse sar­comas, said that a number of the primary tumors contained muscle cells which had been modified in various ways and which were no longer present in later transplant-generations, perhaps because of their transformation into malignant cells. The three tumors we examined are presented here in the order of increasing confusion, which is probably the order of decreasing age of the tumors. The particular slide from which the data of figure 18 were secured did not contain any modified muscle cells, and one sees that classes II and III predominate. In mouse 5 (fig. 19), the predominant tumor nuclei are those of class II, although dediff erentiating muscle cells are present. Class I nuclei predominate in the tumor on mouse 2 (fig. 20) . Muscle nuclei make up at least a third of class I, and the blood cells make up much of the rest of class I. The typical tumor-diploid nuclei are present, however. Another portion of the tumor might have contained fewer of the affected muscle cells, as was the case in the methylcholanthrene tumor on mouse 3. An investigation of the mitoses showed that most of the mitotic nuclei were tumor-diploids of class II. We may conclude: (1) that our mouse tumors show the same sedation of volume classes in their nuclei that some normal adult tissues do, although not so markedly, (2) that the series always begins with class I and extends un­broken to some higher class, ( 3) that exceedingly large nuclei may be en­countered in the tumors, and ( 4) that class I is definitely set apart from the other classes, at least in the heterogeneity of its nuclei. We have not been able completely to separate resting normal cells from resting tumor cells. B. MITOTIC FIGURE VOLUMES The data so far presented do not of themselves confirm the statement of Ehrich ( 1936) that, in the change from normal to malignant tissue, there has been a fundamental doubling or quadrupling of the average volumes of the nuclear classes. This is because of the lack of absolute knowledge of the tissue from which the cancer arose and because of the presence in tumors of small class I nuclei, which have about the same size as the smallest nuclei in normal tissue. However, confirmation can be secured by showing that the diploid mitotic figures of the tumors have two or four times the volumes of the corresponding figures in the same phase of mitosis in normal fetal and healing adult tissue, and then by showing that the cancer chromosomes are themselves twice or four times as large as normal chromosomes. The polymodal character of the distribution of nuclear volumes in cancer tissue can be explained in part by a doubling of the chromosome number. The common occurrence of polyploid mitoses in cancer lends strength to this explanation. The nearly exact doubling of volume of the total figure with doubling of chromosome number is very striking when one compares the volumes of metaphase figures. In tumor XG, 19 indubitably diploid meta­phases of 40 chromosomes each, ranging in volume from 399 to 751 cubic micra and averaging 554 cubic micra, may be contrasted with 4 tetraploid metaphases with volumes of 854, 909, 932, and 1,006 cubic micra (average 925 cubic micra), and they in turn may be contrasted with an octoploid early metaphase of about 145 chromosomes in 2,335 cubic micra. As Winge (1930) has demonstrated for tar cancers in mice, the peaks of chromosome number distribution occur at about the diploid, tetraploid, octoploid, etc., numbers, with some intermediate aneuploid numbers. In tumor XG there is a heavy preponderance of diploid or approximately diploid chromosome numbers. In prophases and metaphases in which the chromosome number is countable certain average volumes of the mitotic figures obtain. By measuring the volumes of figures in which the chromosomes can not be counted because of overlapping or clumping, the ploidy of the figure can be judged. The volumes of some 592 mitotic figures of tumor XG were calculated. The distribution of these figures by phase and volume is given in Table 1. Most of the prometaphases and metaphases fall into the diploid class. Hence it may be assumed that the dominant class of any other mitotic phase is also diploid. In this manner the volume changes undergone by an average diploid nucleus and its products may be followed through the cycle of mitosis. The predominant mitotic nuclei should belong to the pre­dominant class of resting nuclei. Because of the predominance of diploid or near-diploid mitoses and of class II resting nuclei in tumor XG, the class II resting nuclei must be diploid or near-diploid, and they are called "tumor­diploids." The volume changes of classes with nuclear and mitotic volumes above and below those of the predominant class are easily followed out. Figure 21 sets forth the volume progressions for diploid and tetraploid mitoses in tumor XG. The other nine tumors show much the same picture as does tumor XG, but the samples are too small for the amount of aneuploidy and classes possible. Hence they are not included in figure 21. But similar data are included for normal embryonic liver and regenerative mitoses about a cut on an ear. TABLE 1 FREQUENCIES OF MITOTIC FIGURE VOLUMES IN TUMOR XG Volume (µ3) Mitotic Phase p M A T R 100-150 ............ 1 150-200 ............ 2 5 200-250 ............ 6 17 4 250-300 ............ 1 2 20 5 300-350 ............ 1 2 23 7 350-400 ............ 10 2 13 12 400-450 ............ 1 12 15 20 450-500 ............ 2 17 16 21 500-550 ............ 1 13 13 23 550-600 ............ 1 16 7 30 600-650 ............ 14 8 20 650-700 ............ 2 15 18 700-750 ............ 4 5 6 20 750-800 ............ 5 6 3 12 800-850............ 2 3 8 850-900 ............ 5 5 2 3 900-950 ............ 4 3 2 950-1,000 ............ 4 1 1,000-1,100 ............ 4 1 1 1,100-1,200 ............ 8 1 3 1,200-1,300 ............ 12 2 1,300-1,400 ............ 11 1,400-1,500 ............ 7 1,500-1,600 ............ 3 1,600-1,700 ............ 9 1,700-1,800 ............ 1 1,800-1,900 ............ 1 1,900-2,000 ............ 1 2,000-2,300 ............ 2,300-2,400 ............ 2,400-2,500 ............ 1 2,500-2,800 ............ 2,800-2,900 ............ 1 2,900-3,000 ............ 1 3,000-3, 100 ............ 1 TOTALS ........ 92 127 15 148 210 SUM=592 P = Prophases M = Metaphases and Prometaphases A = Anaphases T = Telophases R = Reconstruction nuclei As may be seen in figure 21, the average volume of every tetraploid mitotic phase in the cancer is about double the volume of the corresponding diploid phase in the cancer. This is to be expected on the basis of chromosome num­ber. The mitotic phases of the tumor-diploid nuclei are about double in volume the corresponding diploid phases of the two normal tissues. On the basis of chromosome number, this is entirely unexpected, because both the pre­dominant nuclei in the cancer and the predominant nuclei in the two normal tissues are diploid. If it is the quantity of chromatin and not the simple MITOTIC STAGE RESTING PRO PHASE PRO META PHASE METAPHASE ANA PH ASE TELOPHASE MAX. CONTR . OTHER RECONSTRUCTION RESTING 100 300 500 700 900 1100 1300 1500 1700 2000 AVERAG£ VOLUME 0 F FIGURE IN CUBIC M ICRA FIG. 2. Mitotic Figure Volume Progressions in Normal and Cancerous Tissues. chromosome number which determines the volume of the nucleus, it must be that the tumor-diploid nucleus has chromosomes with double the quantity of material which is present in the chromosomes of the normal diploid nucleus. Also, the tumor-polyploids must have chromosomes which are double the size of those found in fetal liver cells, else the tumor-tetraploid nuclei would not be double the size of the tumor-diploid nuclei. One can assume with some justification that the embryonic diploid nuclei of all tissues have about the same nuclear volume and hence the same size of chromosomes. Jacobj (1925) also found roughly the same volumes for his class I nuclei of several embryonic mouse tissues. That class I nuclei of all normal adult mouse tissues also all have the same size chromosomes may be assumed with less justification, but it is rather likely. It is a difficult task to show what sorts of chromosomes are present in nuclei of classes higher than class I in normal adult tissues, since it is usually nuclei of class I which divide in response to a demand in post-embryonic organs, according to Clara ( 1931), who also found class II nuclei to have a tetraploid number of chromosomes of the usual size. However, there are still further complications. Certain mitotic figures in cancers are found to have volumes that usually belong to tumor-tetraploid nuclei, yet on counting the chromosomes one finds there is only the usual diploid number present rather than twice the diploid number. The chromo­somes in these cases are about double the volume of the usual tumor chromo­somes. A case that indicated chromosomes of four times normal size was found in the methylcholanthrene-induced tumor on mouse 3. Two sister early ana­phases, still rather loose and both obviously diploid, had volumes of 658 and 596 cubic micra. Other diploid anaphases measured in this tumor had volumes of 288, 318, 312, 355, and 388 cubic micra. Some of the chromosomes of the two large anaphase figures in question seemed to be double. This doubleness has been found, however, in two sister anaphase figures, each of 368 cubic micra, in tumor 10591A. In other words, an anaphasic split can be seen both in double and in quadruple chromosomes of mouse tumors. A similar case of quadruple chromosomes was present in the methylcholanth­rene tumor in mouse 7. Two sister anaphase figures, each with 40 large chromosomes, had volumes of 699 and 604 cubic micra. This tumor also contained two more sister diploid anaphase figures of 762 and 723 cubic micra, two sister tetraploid anaphase figures of 1,630 and 1,745 cubic micra, and four diploid metaphase figures of 1,050, 1,130, 1,180, and 1,380 cubic micra. All these figures were about twice as large as the volumes usually found in mouse tumors for phase figures of the ploidy concerned. Chromo­somes of four times normal size doubtless furnish the explanation. This same tumor also contained mitotic figures of the usual volumes, as well as some mitotic figures whose volumes were of the same order as those of normal adult and embryonic diploid nuclei. Such diploid mitotic figures of normal volumes included 4 anaphases of 117, 136, 197, and 197 cubic micra, and also meta­phases with chromosomes of the same size as those in fetal tissues. These small mitoses are products of the small diploid nuclei of class I of the tumor, whieh class we consider to be largely made up of normal, non-malignant cells. In concluding the section on volumes of mitotic figures in normal and cancerous tissues, we may say that, with the exception of those of the normal connective tissue cells, mitotic figures in cancer are of two or four times the volume of corresponding normal mitotic figures in the same phase and with the same number of chromosomes. Hence cancer chromosomes should have two or four times the mass of embryonic and normal adult somatic chromosomes. C. THE CHROMOSOMES Our previous evidence has suggested that the cancer chromosomes are larger than normal chromosomes. Let us examine the chromosomes directly. Since the chromosomes are undergoing a process of coiling and uncoiling during the time they are visible, comparisons in dimensions of normal and cancerous chromosomes must be made at definite stages which are marked by other things besides the size and shape of the chromosomes. The equatorial plate stage of metaphase suggests itself first because it is marked by a definite arrange­ment of the chromosomes. The prophase in general, being anything but static, is excluded, except for that stage in prophase in which the nuclear membrane is in the process of breakdown. At this time one or two of the large plasmo­somes may still be discerned, and the chromosomes are seen as elongated, coiled or chromomeric double threads, which are definitely discrete in well­spread preparations. Anaphase and telophase are not considered because of the close packing of the chromosomes. rlG . 24. f/G . 25. Fig. 22. Middle Prophase, Subdiploid, of Tumor XG. Average Chromosome Volume 1.4 µ.3. Fig. 23. Middle Prophase, Diploid, of Fetal Mouse Liver. Average Chromosome Volume 0.66 µ.3. Fig. 24. Metaphase, Diploid, of Tumor XG. Average Chromosome Volume 0.93 µ.3. Fig. 25. Metaphase, Diploid, of Fetal Mouse Liver. Average Chromosome Volume 0.44 µ3. Middle prophase figures of cancer and fetal liver are reproduced in figures 22 and 23, respectively. The drawings were made by means of a camera lucida at a table magnification of about 2,500 diameters. The chromosomes of the two tissues showed in these figures approximately the same degree of coiling or alternation of deeply-staining and lightly-staining regions. The fetal liver prophase illustrated in figure 23 has chromosomes varying in width from 0.48 to 0.84 micron, the mode being at about 0.80 micron and the mean being taken as 0.72 micron. The lengths run from 2.0 to 6.4 micra, averaging 3.23 micra. To arrive at the volume of the chromosomes, they were considered as paired cylinders and calculations were made as out­lined in the chapter on procedure. The average chromosomal volume in the fetal liver prophase was 0.66 cubic micron. The total volume of the 40 chromosomes is 26.3 cubic micra. In the subdiploid prophase of tumor XG illustrated in figure 22, the chromo­some width ranges from 0.60 to 0.80 micron, with a mean taken as 0.72 micron; and the lengths range from 2.8 to 13.6 micra, the average for the 25 chromosomes being 6.94 micra. The average chromosomal volume is 1.41 cubic micra. The total volume of a diploid set of 40 chromosomes would therefore be 56. 5 cubic micra. Since 15 of the 40 chromosomes of the diploid set of the tumor chromosomes were not present, it is possible that their absence led to an average chromosomal volume somewhat in error. We may conclude that the cancerous and embryonic chromosomes measured in middle prophase had about the same width, but the cancerous chromosomes had about twice the length and therefore about twice the volume of the embryonic chromosomes. For each of the two normal tissues and the ten tumors about nine metaphases or prometaphases were drawn and the individual and average volumes of their chromosomes were calculated. The figures drawn were not picked exactly at random; the criteria of selection were: (1) a relative absence of clumping together of the chromosomes, and (2) the location of most of the chromosomes in a plane or planes parallel to the focal plane. At times figures unsatisfactory in these respects had to be copied. Degree of ploidy played no important part in the selection. Table 2 contains data on the average chromosome volumes of 109 normal and cancerous metaphases and prometaphases, and some of the drawings are reproduced in figures 24 through 34. The data of Table 2 require discussion. The numbers of chromosomes re­corded for many of the figures are probably in error by a few per cent because of overlapping of the chromosomes and because chromatids may have been mistaken for whole chromosomes, and vice versa. Such errors in observation also lead to errors in calculation of the mean chromosomal volumes. However, mistakes of this nature do not materially affect the main results : the two normal tissues contained metaphases most of which had average chromosome volumes of about 0.5 cubic micron; the average volumes of the tumor chromo­somes, however, were distributed in a fairly normal manner about a mean value of approximately 1.1 cubic micra, which is double the mean of the normal chromosomes. Unlike the situation in the prophase, the larger size of the flG. 26 flG. 2.7, 10,,µ­ FIG. 30. F'IG. 29. fl(;. Fig. 26. Fig. 27. Fig. 28. Fig. 29. Fig. 30. Fig. 31. Fig. 32. Fig. 33. Fig. 34. 32. f IG. 33. Metaphase of Fetal Liver. 41 (?) Chromosomes. Average Volume 0.46 µ3. Metaphase of Healing Ear Wound. 40 Chromosomes. Average Volume 0.52 µ3. Metaphase in Muscle Cell in Methylcholanthrene Tumor. 40 Chromosomes. Average Volume 0.42 µ3. Metaphase in Methylcholanthrene Tumor. 38 Chromosomes. Average Volume 1.2 µ3, Metaphase in Methylcholanthrene Tumor. 86 Chromosomes. Average Volume 1.0 µ3, Metaphase in Transplant of Spontaneous Tumor. 138 Chromosomes. Average Volume 1.2 µ3. Metaphase in Tumor XG. 40 Chromosomes. Average Volume 2.4 µ3, Metaphase in Transplant of Spontaneous Tumor. 44 Chromosomes. Average Volume 2.2 µ3. Metaphase in Methylcholanthrene Tumor. 39 Chromosomes. Average Volume 6.2 µ3, Nuclear Phenomena in Mouse Cancers TABLE 2 AVERAGE VOLUME AND NUMBER OF CHROMOSOMES IN METAPHASE AND PROMETAPHASE M = Metaphase PM = PROMETAPHASE A=Anaphase Tissue Normal Chromosomes Number Av. Vol. Figure Volume Stage Fetal Liver -----­------------­-----­-----­----­ 41 0.44µ3 333µ3 M 41 0.46 340 M 41 0.46 377 M, late 41 0.48 528 M, late 42 43 37 0.48 0.54 0.57 365 573 434 M , early M, disarranged M Cut Ear ............................................ 37+ 0.42 328 M 45? 0.44 224 M , PM? 33+ 0.46 236 M 36+ 0.49 323 PM? 40 0.52 413 M , PM? 37+ 0.60 314 M 29+ 0.62 273 PM 34+ 0.74 421 M 31+ 0.77 260 PM 27+ 1.17 260 M Tumor XG .................................................... 29 0.52 296 M, or A? 47 0.83 509 M 42 0.93 495 M 57 1.00 580 M 44 1.10 495 M 36 1.20 461 M 41 1.20 570 M 53 1.20 890 M , disarranged 133 1.30 1,640 PM 39-44 1.40 473 M 52 1.50 993 PM 40 1.60 501 M 40 2.40 690 M Sarcoma 180 .................................... 81 0.60µ3 710µ3 M 73 0.64 648 PM? 71 0.65 726 M 94 0.65 649 M , 2A? 62 0.82 855 M 83 0.94 818 M 158 0.97 2,070 M 70 1.00 655 PM? 63 1.10 740 PM? 77-78 1.10 695 M 47 1.10 538 M 75 1.10 1,160 M 54 1.30 842 M Spontaneous, Mouse 6 .................... 76 0.69 745 M 35 1.00 438 M, A? 56 1.10 658 M 71 1.10 753 PM? 138 1.20 2,330 M , PM? 106 1.30 1,540 M 42 1.50 856 M 59 1.50 1,010 M 44 2.20 1,140 M TABLE 2-Continued AVERAGE VOLUME AND NUMBER OF CHROMOSOMES IN METAPHASE AND PROMETAPHASE Chromosomes FigureTissue Number Av. Vol. Volume Stage 10591A .............................................. 56 0.61 598 M 54 0.77 718 M 49 0.86 800 M 80 0.98 1,310 M 66 1.10 1,460 M 97 1.20 2,590 PM? 70 1.40 1,550 M? 45 1.50 663 M 46 1.70 800 M 49 1.70 1,130 M 49 2.00 1,030 M,PM Methylcholanthrene, Mouse 16.... 37 0.75µ3 354µ3 M 39 1.00 260 M, very compact 45 1.00 592 M 38 1.20 503 M 50 1.20 680 PM 36 1.30 553 M 43 1.40 740 M, disarranged 43 1.50 431 M 43 1.50 861 M, disarranged 39 6.20 1,680 M Methylcholanthrene, Mouse 7 ...... 51 0.47 510 M 86 1.00 1,440 M 885 1.10 20,480 PM 130 1.10 2,610 PM 53 1.30 930 M 58 1.70 1,490 M, disarranged Methylcholanthrene, Mouse 14 .... 45 0.48 316 M 42 0.57 M 44 0.92 475 M 40 0.93 593 M 40 1.20 688 M, spread 28 1.30 695 PM 36 1.30 PM, disarranged 40 1.40 767 PM 188 1.40 large PM, disarranged Methylcholanthrene, Mouse 3 ...... 35 0.46 376 PM? 33 0.65 593 M 43 0.37 265 A, one-half 43 0.88 295 M, A? 39 0.94 469 M 36 1.10 499 M 51 1.10 612 PM? 54 1.10 718 PM 44 1.20 552 PM 55 1.60 815 M Methylcholanthrene, Mouse 5 ...... Most of the chromosomes in this tumor were poorly fixed and swollen. 58 1.10 680 PM 35 1.30 618 M 44 1.50 549 M 33 1.70 678 PM? 32 1.90 428 M 29 2.00 452 PM? Methylcholanthrene, Mouse 2...... 40 0.42 354 M, disarranged 42 0.72 740 M, disarranged 40 0.73 458 M 96 0.81 1,330 M 40 0.88? 290 M, disarranged 40 1.60 551 M, disarranged metaphasic chromosomes in cancer is manifested not only in a greater length but also in a greater width than that of normal chromosomes. The tumors contained metaphases with chromosomes in the same size range as embryonic chromosomes; such metaphases in the tumors were largely of about the normal diploid chromosome number and are considered to have been of dividing normal stroma cells. The tumors contained diploid or hyperdiploid metaphases of double or more than double the usual volume which had chromosomes averaging up to 2.4 cubic micra. There was also one large diploid metaphase with chromosomes which averaged about 6.2 cubic micra. With regard to the normal mitoses, it might be remarked that the single high average chromosome volume of 1.17 cubic micra from the cut ear is questionable, for it comes from a metaphase figure whose total volume, 260 cubic micra, is just half of the volume expected on the basis of a diploid number of chromosomes of the size concerned. The chromosomes in this figure were swollen, as though poorly fixed, and they bore considerable resemblance in shape to the obviously swollen chromosomes of the methylchol­anthrene tumor on mouse 5. It is noticeable in this particular tumor that the average chromosome volume, about 1.6 cubic micra, is displaced upward when compared with those of the other tumors, although the average volume of class II resting nuclei in this tumor is not. Therefore this unusual size, as well as that of the swollen chromosomes in the cut ear tissue, is regarded as a fixation artifact and not as evidence of true growth beyond the size level attained by unswollen chromosomes in similar tissues. We may conclude that the chromosomes in adult regenerative tissue do not have more material in them than do the chromosomes in embryonic tissues. We realize, however, that our data are rather limited. In the tumors the mean chromosome volume is about twice that in normal tissues. The tumor chromosomes have really grown materially beyond the normal chromosomes, and in some of the tumor cells the chromosomes are even four or more times as large as embryonic or normal adult chromosomes. Although our attempts to unravel the mitotic chromosomes experimentally were unsuccessful, it may be proved that this growth of the tumor chromosomes is a matter of reduplication of the strands (probably with no reduplication of the centromeres) by considering the plasmosome number in resting cells of tumors and normal tissues. D. THE PLASMOSOMES A study has been made of the plasmosomes, or true nucleoli, in mouse tumors not only because of the attention which has been given these structures by other investigators of malignant tissues, but more especially because the plasmosomes furnish a means of determining the chromatid valence of chromo­somes (i.e., the number of times the chromatids have been reduplicated by endomitosis). Certain fundamentals known about plasmosomes bear on our evidence. It is a well-established fact that plasmosomes arise at definite locations known as nucleolar organizers along specific chromosomes (Carothers, 1913; Wenrich, 1916; Heitz, 1931; McClintock, 1934; and Gates, 1942). The number of nucleoli formed in telophasic nuclei corresponds to the number of organizers present, and since each chromosome is represented twice in simple diploid cells, the number of nucleoli initially present is 2 or a multiple of 2. The number of nucleolar organizers in normal diploid tissue is constant for a given organism, and a given organizer usually produces a plasmosome of characteristic size. In polyploid cells, the number of nucleolar organizers is, of course, increased with the ploidy. Telophase chromosomes are generally conceded to be split; therefore it is not surprising to find that the nucleolus on a single chromosome arises first as two droplets, one on each chromatid. Rarely, as in Ambystoma (Dearing, 1934), these droplets remain separate, but usually the two primordia fuse very early into one globular element which we have termed a unit plasmosome. Unit plasmosomes show a marked tendency to fuse secondarily with one another into larger masses, depending on the proxi­mity of the organizers (Heitz, 1931) which allows the nucleoli to touch as they enlarge (Geitler, 1938a), on the fluidity of the nucleus (Manton, 1935), and apparently on the species and perhaps on the type of tissue in the animal or plant concerned (cf. Geitler, 1938a; Woods, 1937; Kuhn, 1938; Ernst, 1938­9). If fusion (i.e., lobing and increased size of nucleoli) is taken into account, TABLE 3 UNIT PLASMOSOMES IN MousE LIVER NucLEI Nuclear Class Unit Plasmosomes Evident Number of Nuclei I (Diploids) --------------------------------------------------4 57 3 20 2 8 1 0 85 II (Tetraploids) --------------------------------------------8 93 7 52 6 35 5 15 4 5 3 and less 0 200 III (Octoploids) ----------------------------------------------16 8 15 3 14 6 13 7 12 4 11 2 10 2 9 1 8 and less 0 33 IV (Hexadekaploids) --------------------------------------32 1 31 1 27 1 25 1 4 the number of plasmosomes gives an indication of the number of chromatids present in each chromosome which carries a nucleolar organizer. When the plasmosome numbers of normal mouse liver nuclei are investigated (see Table 3), it is found that the class I nuclei, which are undoubtedly simple diploids, contain a maximum of 4 unit plasmosomes (of 85 class I nuclei, 5 7 were found to have 4 unit plasmosomes). Of 200 class II or tetra­ploid nuclei, 93 had the maximum number of 8 unit plasmosomes, and progres­sively fewer nuclei seemed to have progressively lower numbers of plasmo­somes. Of 33 class III or octoploid nuclei, 8 were observed with the maximum number of 16 unit plasmosomes. Four nuclei of class IV (hexadekaploids) were found to show from 25 to 32 unit plasmosomes. Table 4 gives data on the number of plasmosomes in the first 3 nuclear classes of tumor XG. Here it must be emphasized that we have set approxi­mate volume limits to each class; thus all nuclei with a volume of 450 cubic micra or less are included in class I. Class II begins with 450 cubic micra and extends to 1,200 cubic micra, and class III includes nuclei from 1,200 to 1,800 cubic micra in volume. Undoubtedly class I includes not only normal diploid cells which are enclosed within the cancer but also some small tetraploid­volume cells of class II, and the same principle applies to classes II and III. Turning now to Table 4, one sees that 164 nuclei out of 194 in class I have 4 plasmosomes or less and are probably largely normal diploids. Thirty-two nuclei show from 5 to 8 unit plasmosomes and hence must be at least tetraploid in nucleolar organizers. In class II 370 nuclei out of 661 have from 5 to 8 unit plasmosomes and hence must be at least tetraploid in the number of nucleolar organizers, although we know from other evidence that the nuclei in this class have a more or less diploid number of large chromosomes. The 277 nuclei in class II with from 1 to 4 plasmosomes, although they show much the TABLE 4 UNIT PLAsMOSOMES IN MousE TUMOR XG Evident Unit Number of Nuclei Plasmosomes Class I Class II Class III 1 ···················································------· 2 3 2 ----·---········--------------············---·-------····· 12 26 3 ·········-·············-----------· -··········---·-------· 32 61 2 4 ---······························--­---­·---·············· 118 187 4 5 ······--------··--··········-··········--···------·····--­ 7 65 1 6 7 --·-················--··-··········-·····-··········--·--· ----­-·-­-················----·---··········· ·-----· ······ 9 8 78 89 4 2 8 -·--· -····----·---················----·-·················· 8 138 14 9 10 ·­··--·-···--·········----·-· ·······------------·-········ --·····----·--·············-------·--·-···········--·---·· 3 5 2 3 11 ····----­-······---------···········---·-···-···········-­ 1 5 12 13 ··········--·--···----·---·-------···········-----········ ·················-···-············--------········-------­ 0 1 2 1 14 ······--·-···-·········----········--­·····-------········ 2 2 15 -------····················--­····­----············------­ 1 0 16 ··········································-·--······--·--· 1 2 17 ········-­-·--·---···············-········----····----···· 0 18 ··-·····-·······--------------·--­···········-····----···­ 0 19 ·-··--·---·-----­-· ·················------·-······----··· · 1 Total ···············---------·-·---·-········----·-·· 194 661 45 same progression of frequencies with regard to plasmosome number as do the diploids in class I, are not to be interpreted as diploid in nucleolar organizers, but as tetraploid, because their plasmosomes are quite large, indicating com­poundness (see figures 3 5 to 39, which are of normal liver nuclei and illustrate the point that the volume of a compound plasmosome approximates the sum of the volumes of its individual components). Furthermore, the mechanics of diplochromosomes could often cause the 8 nucleolar organizers of tumor­diploid cells to be distributed in 4 pairs instead of singly at random in the nucleus, as occurs in normal cells. This point is discussed in more detail later. In class III of tumor XG, 44 out of 45 nuclei were found to have 16 or less plasmosomes (1 had 19) and may be regarded as octoploid in nucleolar organizers. We note the predominance of nuclei (14 out of 45) with 8 plasmosomes; this parallels the case of the 4-plasmosome nuclei in class II, and again in class III it appears that the 16 nucleolar organizers may often be distributed in the nucleus in 8 pairs. Class III in the tumor, we remember, is largely tetraploid in chromosome number, and its chromosomes, if they were normal, would have only 8 nucleolar organizers. That they have twice as many substantiates the inference that cancer cells contain diplochromosomes. For an extension of the evidence on tumor nucleoli, let us consider Table 5, which was made up of the combined data on the plasmosome number of 6 other mouse tumors. The total situation is similar to that in tumor XG. In class I, 119 out of 161 nuclei have 4 plasmosomes or less and are mostly normal diploids. There are 42 nuclei which have from 5 to 9 plasmosomes and are chiefly tetraploid in nucleolar organizers. In class II the nuclei build up to a peak frequency of 86 out of 259 nuclei with 8 plasmosomes. The 220 nuclei with 8 plasmosomes or less are probably largely tetraploid in nucleolar organizers but diploid in chromosome number. The 38 nuclei in class II with 9 to 16 plasmosomes are octoploid in nucleolar organizers and belong in class III. The 1 nucleus which seems to have 28 unit plasmosomes either belongs in class IV or is anomalous. In class III, 111 out of 112 nuclei have 16 plasmosomes or less and are octoploid in nucleolar organizers (although tetraploid or diploid in chromosome number). The high frequency of nuclei with 6 to 8 plasmosomes again suggests that the 16 nucleolar organizers are distributed in 8 pairs, as in tumor XG. The 28 nuclei of class IV have up to 29 plasmosomes; very likely they have 32 nucleolar organizers. We note a tendency toward distribution of the organizers in 8 groups of 4, as witnessed by the high frequency of nuclei (7 out of 28) with 8 plasmosomes. In class V the 11 nuclei have up to 47 plasmosomes, and we suppose the nuclei of this class to have, ideally, a 32-ploid number of 64 plasmosomes and a 16-ploid number of diplochromosomes. One class VI nucleus with 48 plasmosomes was seen, as well as 1 class VII nucleus with about 143 plasmosomes. What do these data signify? Firstly, since a given set of the usual chromo­somes of cancer contains twice as many nucleolar organizers as the same set of normal chromosomes, it appears that cancer chromosomes are split in the region of the nucleolar organizers and, by inference, elsewhere (except at the CLASS I NUCLEI flG. 35. r(G. 36. N UCLEU51 275 p3 NUCLEUS, 27 4~3 NUCLEOLl NUCLEOLUS, 1) 1 1.4 1) 10.3 1.4 10 .3µ3 ~ 2.6 4) 10,N CLASS Il NUCLEI 8 2 I 3 4 2 ~ f"IG. ~7. NUCLE us\ 49„ NUCLEOLI, NUCLEUS, s2a,.3 I) 0.5 NUCLEOLI, 2) 0.5 1) 3 .0 3) 5.6 2) 0.2 4) 6.4 3) 1.4 13. of'\3 4) 0.5 5) 0 .3 6) 2-6 7) 4.8 8) 2.1 14 .9,..3 Fig. 35. Class I Liver Nucleus, 4 Plasmosomes. Fig. 36. Class I Liver Nucleus, 1 Compound Plasmosome. Fig. 37. Class II Liver Nucleus, 8 Plasmosomes. Fig. 38. Class II Liver Nucleus, 4 Plasmosomes, Some Compound. Fig. 39. Class II Liver Nucleus, 1 Compound Plasmosome. TABLE 5 UNIT PLASMOSOMES IN 10591A, THE SPONTANEOUS TUMOR, AND METHYLCHOLANTHRENE TUMORS ON MICE 3, 5, 7, AND 14 Numbers of Nuclei in Classes: Unit Plasmosomes I II IIJ IV v VI VII 1 2 5 1 1 3 43 4 1 4 70 20 10 5 20 23 10 6 11 42 21 1 7 8 44 15 3 8 2 86 19 7 9 1 16 12 3 10 8 5 1 11 2 5 1 12 5 7 1 13 1 1 1 14 1 1 3 15 4 1 16 ·--··-·········-·······--··-······ 1 --··-··-············-···--··-····­ 1 3 1 17 18 19 1 20 3 1 24 1 1 23 ··----·-·-····-·······-··········­ 27 -·-······························· 28 29 32 ···········-···-····-········-···· 1 38 ·····················------------­ 1 48 ··············-··················· 47 ----------------------·····-······ 1 143 -------··························· 161 259 112 28 11 1 Total ·········-·············· centromere, which maintains structural unity). In other words, cancer chromo­somes contain at least a doubled number of chromonemata and chromatids, and this is the explanation of the larger size of cancer chromosomes. Secondly, since there is more fusion of nucleoli in cancer tissue than in normal tissue, tending to give plasmosome numbers at one-half or one-fourth the supposed maximum (which is not true in normal mouse tissues), we may assume that the nucleolar organizers of cancer cells are not distributed singly at random in the nucleus, as in normal cells, but in pairs or fours at random. This is just what is expected if the cancer chromosomes are polytene, with the nucleolar organizers on sister-halves or -fourths of the chromosomes in closer juxtaposition than would be true, on the average, of the nucleolar organizers on unattached, normal chromosomes. When the number of chromatids is more than 2 per chromosome, the total number of visible plasmosomes will depend, among other things, on the degree of separation of the strands which go to make up the polytene chromo­some (see the diagram, figure 40) . The fact that in tumor-diploids 8 unit plasmosomes are so very frequent suggests that usually the 4 strands of the diplochromosome spread out in pairs. Indeed, if we assume that these chromosomes open out in pairs we can explain a puzzling feature: increased length rather than width of cancer diplochromosomes in mid-prophase (fig. 22) to give a volume double that of normal fetal chromosomes (fig. 23) . However, since mouse cancer chromosomes in metaphase appear to have terminal spindle attachments, the two arms must be brought together some­how, perhaps by formation of the matrix or pellicle. With further reduplica­tion of the chromosome strands there is no more opportunity for very extensive separation as long as the centromere is intact; hence 8 plasmosomes would be most frequently encountered, as may be seen in Table 5 in class III and IV and in Table 4 in class III. The opening out of the polytene chromosome does not always occur, as attested by the great number of class II nuclei in tumor XG with only 4 plasmosomes, as well as by some very large resting nuclei of tumor XG of about 10,000 cubic micra with only 3 or 4 plasmosomes, which are very large. SIMPLE CHROMOSOME =O l / tHROMATID ~UCLEOLUS / CENTROMERE i::1'ucLEOLAR ORGANIZER DIP LOCHROMOSOME OR POLYTENE CHROMOSOME OR i="IG . 40. Diagram of Nucleolar Relations in Simple and Polytene Chromosomes. 4o The University of Texas Publication These large nuclei could have had, so far as our criteria go, a diploid number (40) of chromosomes, each with 32 times as many strands as the chromosomes of non-cancerous tissues have. Of course, it is also possible that the chromo­somes of these nuclei were not highly polytene, but that there was great polyploidy and the sister plasmosome-organizers on the reduplicated chromo­somes remained in close proximity to each other after every endomitosis, not­withstanding a division of the centromeres. In either case, very extensive fusion of the plasmosomes occurred. The extreme opposite of this case is furnished by the methylcholanthrene­induced tumor on mouse 7. In it there is a resting nucleus of 29,800 cubic micra with about 143 plasmosomes. This nucleus probably has chromosomes averag­ing twice the size of fetal chromosomes, for it would then have about 1,280 chromosomes bearing 256 nucleolar organizers, while if the chromosomes were one grade larger, only about 640 whole chromosomes would be present, and although they would still bear 256 nucleolar organizers, it is very likely that these organizers, being distributed in fours, would form less than 143 definitive plasmosomes in telophase. Besides the distribution of nucleolar organizers in multiples of 2 in cancer nuclei, another factor leading to greater fusion of tumor nucleoli is their rela­tively greater size, which makes it more probable that several will come in contact. This larger size has been noted by many workers (cf. Chapter IV) and is also supported by our observations. In this connection, see figures 3 5 to 39, of normal liver nuclei with various degrees of fusion of plasmosomes. Calculations of nucleolar volume, based on the ellipsoidal formula, 4/3 pi times the product of the three semi-axes (two being measured and the other taken as their average) , show that in a given nuclear class the total plasmosome volume occupies a relatively narrow range regardless of the degree of fusion. The compound plasmosome is larger, the greater the degree of fusion (since its volume approximates the sum of the volumes of its components), and it also has more heterochromatin attached to it. These nuclei and some others, also of normal liver, have ratios of total nucleolar volume to nuclear volume, expressed in percentages, of 1.3, 1.9, 2.2, 2.2, 2.6, 2.6, 2.8, 3.0, 3.4, 3.7, 3.8, and 3.8, while three average-sized resting nuclei of tumor XG were found to have ratios of 4.5, 8.4, and 9.6 per cent. The relatively larger size of the cancer nucleoli has been ascribed to greater metabolic activity (cf. Frugoni, 1941, and the review of non-cancerous nucleoli by Gates, 1942). Darlington ( 1942) infers that a larger nucleolus, i.e., a large amount of nucleic acid, causes rapid synthesis of protein. Without going so far, we may say that a large nucleolus is somehow associated with a high rate of protein synthesis. This may well be through the intermediation of ribonucleic acid, which is found in quantities in the nucleolus and the cytoplasm and contains the same physiologically active sugar as do many nucleotides which function in enzyme systems. It is to be noted that Tennant, Stern, and Liebow (1942) find that sodium salts of both ribo-and thymonu­cleic acids stimulate the growth of mouse heart fibroblasts in culture. The close relationship between the metabolisms of thymonucleic and ribonucleic acids is emphasized by Schultz, Caspersson, and Aquilonius (1940). Possibly the two acids are related through an equilibrium or equilibria. A high con­centration of one type of nucleic acid would then make for a high concen­tration of the other type, also. We note that cancer cells of many sorts have been described as hyperchromatic (i.e., containing much thymonucleic acid in heterochromatic knots) , as, for example, by Fried (1924), Goforth (1927), Fried and Buckley (1930), Tuft (1930), M. R. Lewis (1930), Schopper (1933) , Beverwijk (1934) , Hurst (1937), Dimitri and Alem (1938), Selbie (1938), Kirschbaum and Strong (1939), Gotshalk, Tessmer, and Smith ( 1940) , Jaffe, Lichtenstein, and Portis ( 1940) , and Mallory ( 1940) . Hand in hand with this hyperchromatism (increased thymonucleic acid) should go enlarged nucleoli and more ribonucleic acid. Actually, we do find larger nucleoli. A proof of an increase in the amount of ribonucleic acid is furnished by the method described by Brachet ( 1940), involving fixation in Helly' s fluid, treating the experimental slide with a solution of ribonuclease (ours was from Kunitz) for a number of hours, and then staining control and experimental slides with Unna's pyronin and methyl green. After the ribonuclease has acted completely, none of the basic pyronin dye is taken up by the tissue. Without ribonuclease treatment, the nucleoli and cytoplasm of the cancer cell stain heavily with pyronin, considerably more so than do cells of normal mouse liver treated in the same manner. Insofar as the enzyme is specific and the method valid, one can say that ribonucleic acid is more concentrated in the nucleoli and cytoplasm of cancer cells than in normal liver cells, although the latter are noted for their physiological activity, and are especially rich in ribo­nucleic acid (Brachet, 1940). If the basophily of cytoplasm is largely determined by ribonucleic acid, then the concentration of cytoplasmic ribonucleic acid must increase with malignancy, for Evans, Barnes, and Brown (1942) find the intensity of cytoplasmic staining to be directly proportional to the degree of malignancy in carcinoma of the prostate. It is not known how much of the hyperchromatism of the cancer nuclei accompanies the increased physiological activity and how much of it is asso­ciated with the frequent mitotic and endomitotic cycles. Possibly they are all interrelated. In conclusion, it is evident that nucleoli are more numerous in cancer cells than in normal cells of the same ploidy. This is because there are more organizers in the cancer nuclei, since the chromosomes are polytene. For a given number of nucleolar organizers, more fusion of nucleoli occurs in cancer cells than in normal cells because the organizers are not distributed singly at random in the cancer nuclei, but in groups of 2, 4, or more. The nucleoli also fuse more readily because they occupy a relatively greater part of the nuclear volume in tumors. This larger size is, no doubt, associated with the high rate of metabolism of the cancer cell, and it is reflected in hyperchromatism of the nucleus and in greater concentration of ribonucleic acid in the cytoplasm. E. ENDOMITOSIS We have invoked an endomitotic mechanism to explain the origin of the large nuclei found in normal liver and in tumors. We have now to consider the direct evidence that such a mechanism is in operation. As explained in the first chapter, in an endomitotic cycle the chromosomes are reduplicated, but the nuclear wall never breaks down, a spindle does not form, and neither the nucleus nor the cell divides. Briefly stated, the evidence for endomitosis in late-embryonic liver and in tumors consists in the fact that there are intact nuclei in both tissues in which the chromosomes show various stages of a division cycle. In the usual mitotic cycle, the nucleus starts in the resting stage (figs. 41, 46, 47) with plasmosomes, with chromosomes which are uncoiled and appear as a lightly-staining reticulum, and often with well-defined masses of hetero­chromatin. With the onset of mitosis (figs. 42, 43, 48-50) the chromosomes condense into thread-like structures, and by the middle prophase (figs. 44, 51, 22, 23) one sees the split chromosomes. About this time the nuclear membrane breaks down, the plasmosomes begin to disappear, and the mitotic spindle develops. The chromosomes reach a maximal condensation at a stage later than the prophase. On the other hand, both in embryonic mouse liver and especially in tumors one finds intact nuclei (figs. 45, 52, 53) in which there are no plasmosomes and in which the chromosomes often appear as split rods, condensed almost as completely as in the mitotic metaphase. Such nuclei do not fit into the mitotic prophase, nor can they be telophasic or post-telophasic, because they are not found in pairs. We consider them to be examples of an endomitotic meta phase. When the chromatids move apart slightly in the endomitotic anaphase, the chromosome number is visibly doubled if division of the centromeres has occurred; but if the centromeres do not divide, each chromosome becomes a diplochromosome and the chromosome number remains constant. Therefore the resting nucleus formed after endomitosis contains, in either event, twice as many discrete nucleolar organizers as formerly. Class II nuclei, which comprise some 70 per cent of the nuclei of tumor XG, typically show a diploid number of diplochromosomes. These enlarged elements must have arisen by an endomitotic division of normal chromosomes in which the centromere, or point of spindle fiber attachment, either was not reduplicated or the two halves did not separate, and as a consequence double the number of chromatids ordinarily present in normal chromosomes are held together. This accounts for the increased size of mouse tumor diplochro­mosomes, which have twice the volume ( 1.1 cubic micra) of normal elements ( 0. 5 cubic micron). From this point on, tumor cells exhibit several types of behavior. A great majority seem to undergo a more or less normal mitotic division of the diplochromosomes, forming two daughter cells, each with diplochromosomes. Thus the malignant cells are perpetuated. We must ric; . 42. FIG . 43. /Oµ r 1c; . 4S. Fig. 41. Typical Class I Resting Nucleus, Lightly-staining, of Embryonic Mouse Liver. Fig. 42. Class I Nu'cleus, Transitional from Resting to Prophase, of Embryonic Mouse Liver. Fig. 43. Early Prophase Class I Nucleus, in Optical Section, of Embryonic Mouse Liver. Fig. 44. Middle Prophase of Class I Nucleus in Mitosis, in Optical Section, of Embryonic Mouse Liver. Fig. 45. Endomitotic Metaphase of Class I Nucleus of Embryonic Mouse Liver. Plasmosome Absent, although Present in Figures 41-44. f lG. 50. FIG. 51. 10µ nG. sa. f IC:.. 5 3­ infer that in this division the centromeres divide normally, just as do the centrioles, and disjunction of the double chromosomes is prevailingly regular. Irregularity in centromere separation would, of course, produce aberrant chromosome numbers. The remaining cells of class II undergo either of two types of endomitosis, depending on behavior of the centromeres and centro­somes (see Chapter V). The contromeres of the diplochromosomes may fail to divide, thus giving rise to polytene chromosomes, much as the gigantic chromosomes of Drosophila salivary glands are formed by endomitosis in which the reduplicated chromosomes fail to separate (Painter, 1941) . If endomitosis occurs with a division of the centromeres, highly polyploid cells arise. Nuclear classes higher than class II in the mouse tumors are made up of nuclei ranging from those containing highly polyploid sets of diplo­chromosomes to those containing diploid sets of highly polytene chromosomes. Therefore it seems that either type of endomitosis can succeed a previous endomitosis of the same or the other sort. It is obvious that a normal mitotic cycle can follow endomitosis in the same cell. This suggests that endomitosis, and perhaps the type of endomitosis, is impressed on the cell by some external agent or condition. In this connection it should be noted that in transplants of tumor XG the proportion of large nuclei (i.e., those which have undergone several endomitoses) in small regions is independent of the mitotic rate pre­vailing there at the moment of fixation. The endomitotic increase of chromosome numbers in cancer can proceed rather far. The most highly polyploid mitosis found was a prometaphase (fig. 54) of 885 chromosomes, which averaged 1.05 cubic micra in volume and had therefore passed through one endomitosis without centromere division in their transformation from normal chromosomes. The chromosome number of this mitotic figure is intermediate between 32-and 64-ploidy and could have arisen from the diploid by 4 to 5 successive doublings. The total volume of the figure is some 20,500 cubic micra. Eight hundred eighty-five chromo­somes of the size found would produce a resting nucleus of about 15,000 cubic micra, which would put it into class VI or VII. This prometaphase was in the primary methylcholanthrene tumor on mouse 7. A mitosis with 583 chromosomes (i.e., about 24-ploid) was seen in human cancer by Andres (1932). Fig. 46. Class II Resting Nucleus, Lightly-staining, of Tumor XG. Fig. 47. Class II Resting Nucleus, More Chromatic, of Tumor XG. Fig. 48. Class II or III Nucleus, Transitional to Prophase, of Tumor XG. Fig. 49. Very Early Prophase of Class II Nucleus of Tumor XG. Fig. 50. Early Prophase of Class II Nucleus of Tumor XG. Fig. 51. Early to Middle Prophase of Class III Nucleus of Tumor XG. Note Plasmosome. Fig. 52. Endomitotic Metaphase of Class III Nucleus of Tumor XG. Note Absence of Plasmosome, Presence of Nuclear Wall. Fig. 53. Endomitotic Metaphase of Class II Nucleus of Methylcholanthrene Tumor, Mouse 3. Note Absence of Plasmosome, Presence of Nuclear Wall. FIG. 54. 10I' Fig. 54. Prometaphase of 885 Chromosomes in a Primary Methylcholanthrene Sarcoma. Average Chromosome Volume 1.05 µ,3, The most extreme examples of polytene chromosomes found were some averaging 6.2 cubic micra in a diploid metaphase (fig. 34) in the methylchol­anthrene-induced tumor on mouse 16. To obtain such large chromosomes from normal chromosomes of 0.5 cubic micron would require 3 to 4 successive endomitoses without division of the centromeres. The occurrence of endomitosis in tumors should lead to an increase in the proportion of large nuclei as the tumors age, unless there are certain com­pensating factors. We have some evidence that the mitotic rate is the same for all nuclear classes in tumor XG. However, there may be greater mortality among the high polyploids because of the disparity between the great nuclear volume, which increases as the cube of a linear dimension, and the nuclear surface, which only increases as the square. Furthermore, the highly polyploid mitoses are often multipolar (see also v. Moellendorff, 1940), and the poly­ploidy of large nuclei dividing thus is reduced irregularly. It is possible that the resultant aneuploidy and genie unbalance lead to cell lethality, which may help to explain the extensive death of tumor cells. Lethal aneuploidy may also arise from mitosis of diploid tumor cells whose diplochromosomes fail to divide regularly at the centromere. Ludford (1930a) described an abortive mitosis in cancer which involved the reduplication of chromosomes without the formation of a spindle. In de­generate stages the chromosomes, contracted and quite chromatic, were scattered through the cytoplasm, and the cell eventually died. Levine (1931) likewise described such figures. He said they were formed by the giant, polyploid cells; the chromosomes divided repeatedly, although no spindle was present and no resting nucleus was ever formed again. We have observed such cells in tumor XG. In figure 55, for example, the chromosomes are present as heavily chro­matic multivalent bundles scattered through the cell. It is quite clear that these pathological mitoses are not endomitoses, although they bear some superficial resemblance to them in the absence of a spindle. In endomitosis the nuclear Abortive Mitosis in Tumor XG. Chromo­ some Bundles Scattered through the Cytoplasm. wall remains intact and the nucleus soon takes on the usual "resting" appear­ance as it completes the cycle. A true endomitosis was, no doubt, seen in human lip carcinoma by Evans and Swezy (1929), for they described paired chromosomes inside nuclei whose walls were still intact. Further evidence of like nature for endomitosis in cancers is given in the review of the literature in Chapter IV. The concept of endomitosis as a mechanism for the production of polyploid cells is in conflict with older views on the origin of polyploidy in cancers, so well summarized by Levine (1931). He lists 7 possible mechanisms which are: asymmetrical divisions, the lagging or extrusion of chromosomes, the fragmentation of chromosomes, the reduplication of chromosomes without nuclear reorganization, the occurrence of monasters, cell fusion, and the fusion of sister nuclei. The first two possibilities can be dismissed immediately, for asymmetrical divisions (see figure 11) and the lagging or extrusion of chromosomes from the mitotic field (upheld by v. Moellendorff, 1940) give rise to aneuploidy, not to polyploidy. The fragmentation of chromosomes as a means of achieving polyploidy is highly questionable, for the polyploid cells of cancer do not have chromosomes smaller than do the diploid cells; moreover, in order that all fragments of the chromosomes remain viable, it is necessary that the spindle fiber attachment be diffuse, not localized as it is in the mouse. Chromosomal reduplication without reconstitution of the nucleus, as in the abortive mitosis of Ludford ( 1930a), can obviously not give rise to polyploid resting cells. As for monasters, Levine (1931) states that he could find none in his tumors; they may have been misinterpretations of sectioned material. Our own preparations give no evidence for the two remaining mechanisms, cell fusion and the fusion of sister nuclei after failure of cytokinesis. Moellen­dorff (1940) is a strong advocate of the latter, by which he would explain the occurrence of multipolar mitoses and "pseudoamitotic" figures in cancers. Whatever the final verdict, endomitosis and cellular or nuclear fusion are not mutually incompatible. In conclusion, emphasis should be placed on the fact that, except for the supposedly non-malignant nuclei in class I, every nucleus in the tumors has undergone at least one endomitotic cycle without centromere division, inasmuch as each nucleus has diplochromosomes or polytene chromosomes. These chromosomes of the cancerous nuclei have at least twice as many chromonemata as have the chromosomes of normal or class I nuclei, and they are usually at least twice the size of normal chromosomes. Still larger size and greater number of chromosomes in cancer nuclei result from endomitosis without and with division of the centromeres, respectively. CHAPTER IV OBSERVATIONS OF OTHER WORKERS Are polytene chromosomes characteristic not only of mouse tumors, but also of all other neoplasms? The question will be answered in this chapter insofar as the literature on cancer cytology and histology contains intimations of larger, multiple-stranded chromosomes. A. CHROMOSOMAL EVIDENCE FOR POLYTENE CHROMOSOMES An increased size of the chromosomes was noted in cells of Ehrlich mouse adenocarcinoma by Goldschmidt and Fischer (1929), according to Wermel and Scherschulskaja (1934) . A strand number double or quadruple the normal has been seen occasionally in cancer chromosomes. Such cases have often been interpreted as reduction divisions, in which homologous chromosomes are associated together. Arnold ( 1879) thought divisions which were like those in germ cells took place in cancer. Farmer, Moore, and Walker (1904) announced that they had seen reduction divisions in human cancers. Emmy Stein (1935b) saw double and quadruple chromosomes in the cancerous growths (determined by a homozy­gous recessive mutant gene produced through radium irradiation) in snap­dragons, and she interpreted them as chromosomes undergoing somatic reduc­tion divisions. The idea of heterotypic divisions in cancer was attacked by Hansemann (1904), who said that he had already seen the peculiar shapes of chromosomes which Farmer, Moore, and Walker (1904) figured and that they were merely chromosomes of an abnormal equational somatic mitosis. While Bashford and Murray (1904) at first thought they could confirm the occurrence of reductional divisions in cancer, they later (1905) decided otherwise. An interesting side-light on this point is afforded by the work of Painter and Reindorp (1939), according to which the definitely 8-stranded chromosomes of young Drosophila nurse cells take on the appearance of meiotic bivalents at a certain stage of endomitosis. One can say with fair assurance that the pairing which some of the early workers reported in cancers is really the necessary association of 4 or 8 sister chromatids whose common centromere has not yet divided; i.e., the chromosomes are polytene. Further evidence of a strand number at least double the normal is given by the wide prophasic and metaphasic split in cancer chromosomes. The re­stained cancer mitoses figured by Bashford and Murray (1908) contain very plainly split chromosomes. Wide prophasic splits in cancer chromosomes have been reported by Ludford (1930b), by M. R. Lewis (1932) in both diploid and polyploid cells of tumors studied in vitro, by Lewis and Lewis (1932) in cells of Walker rat sarcoma 338, and by Lewis and Strong (1934) in both diploid and polyploid mitoses in spontaneous mouse tumors. Potter and Richter (1933) described precocious splitting of chromosomes in leukemic cells of mice. In a case of human leukemia Hamilton-Paterson (1941) found anaphase chromosomes to be clearly double. A beautiful demonstration by Ludford ( 1933) showed differences between chromosomes of normal and malignant cells. The normal macrophages, monocytes, fibroblasts, and perhaps endothelial cells of tissue cultures of a rat sarcoma and a rat carcinoma segregated trypan blue, while the malignant cells did not take up the dye. The normal connective tissue cells thus identified had small chromosomes, while the chromosomes of the malignant cells were larger and had a prominent split at metaphase. Ludford (1930a) described tetrad-like chromosomes in abortive mitoses in cancer, and he accepted the view of Evans and Swezy (1929) that precocious splitting gives rise to chromosomes having a false resemblance to paired homologues of meiosis. Winge (1927) noted paired chromosomes and false diakinesis in beet tumors induced by Bacterium tumefaciens. He suggested that polyploidy arose through these diakinesis-like stages. He also thought that the diploid cells in the largely tetraploid beet tumors arose from tetraploid cells by reduction divisions. Winge (1930) extended his investigations to mouse tumors induced by tar. In them he found pseudo-diakinetic stages without spindle formation, which he thought doubled the ploidy. Some figures had single pairs of chromosomes, which Winge interpreted as meaning that doubling of some but not of others of the chromosomes might occur. It is very likely that Winge was seeing endomitotic figures. Likewise, Crew and Koller (1932) saw in mouse tumors paired chromosomes in metaphase of mitosis; the thin chromosomes, originally chromatids, lay beside each other and had not separated farther because the spindle was lacking. The result, they thought, was polyploidy. Another explanation of polyploidy is advanced by Dermen (1941) for bean tumors caused by napthalene-acetic acid. He thinks the chemical induces a splitting of each of the two chromatids of the normal chromosome before mitosis, but it inhibits division of the centromere. Thus double chromosomes are formed. Further action of the chemical would allow the four chromatids of the diplochromosome to split to form eight chromatids attached to a common single centromere. In later division cycles these polytene bundles would probably be unstable and break down at the centromere, forming a tetraploid number of diplochromosomes. With still more effect of the chemical, highly polyploid cells would eventually be formed. Various degrees of polyploidy were seen by Hansemann (1904) in human cancers; by Goldschmidt and Fischer (1929) in Ehrlich mouse adenocar­cinoma; by Heiberg and Kemp (1929) in some human carcinomas; by Hirschfeld and Klee-Rawidowicz (1929) in Jensen's rat sarcoma; by Lewis and Lockwood (1929) in the Walker rat sarcoma; by Winge (1927) in crown gall of the sugar beet; by Levine (1931) in crown gall of tobacco and beet, in Jensen's rat sarcoma, in tar tumors and tumor 180 of mice, in Rous chicken sarcoma, and in human lip cancer; by Andres (1932) in a human skin car­cinoma; by Crew and Koller (1932) in mouse tumors; by Alexenko and Natansohn (1933); by Schopper (1933) in Walker rat sarcoma 319; by Mendolsohn (1935) in cysticercus tumors in rat livers; by Stein (1935a, b) in tumors of the snapdragon; and doubtless by others. Since polytene chromosomes can give rise to aneuploidy by irregular dis­junction, the aberrant chromosome numbers of many cancers suggest that their chromosomes are polytene. If aneuploidy is sometimes caused by the failure of chromosomes to attach to the spindle (v. Moellendorff, 1940), it may reflect injury to the centromere, which may be related to the endomitosis without centromere division that we have postulated as giving rise to cancer diplo­chromosomes. Aberrant chromosome numbers particularly distinguish malig­nant tumors, according to Kemp ( 1930), who said that other pathological tis­sues have little aberration in chromosome number, while fetal membranes display great variation and normal somatic mitoses are nearly always diploid. W. H. Lewis (1939a) noted that normal fibroblasts have the usual diploid number of chromosomes, while sarcoma cells have varying numbers. To summarize this section, we note that chromosomal evidence for polytene chromosomes in cancers is afforded by the observed larger size of the chromo­somes, by the observation of pseudo-reductional divisions, and by the wide split in the chromosomes at metaphase. Earlier in this paper (Chapter III, E) we have explained how endomitosis in cancer gives rise first to diplochromo­somes and if continued may give rise either to more complex polytene bundles or to polyploidy and aneuploidy, depending on the occurrence and regularity of division of the centromeres. Polyploidy and aneuploidy in tumors then may also indicate the presence of polytene chromosomes. The facts cited above from the literature all fall into place and are not at variance with the concept of polytene chromosomes in cancer. B. NucLEAR VOLUME EVIDENCE FOR PoLYTENE CHROMOSOMES Many pathologists and experimental biologists have noted an increase in nuclear size in malignant cells as compared with their mother cells. Among the first to point out the fact that tumor nuclei are comparatively large was Heiberg (e.g., 1908), who stressed the disparity in the sizes of embryonic and tumorous nuclei. All the nuclei, not merely single nuclei, had increased in size in skin cancers (Heiberg, 1921), and the larger nuclear volume was therefore diagnostic for malignancy. Nomikos (1910) reported that a comparison of the sizes of nuclei in carcinomas and their normal mother tissues revealed that most carcinomas had nuclei larger than the nuclei of the tissue of origin, although in some cancers the nuclei were not increased in size and in a very few carcinomas they were actually decreased in size. Borst, who was the teacher of Nomikos, summarized (1924) the available information and concluded that the primary phenomenon in the malignant condition was a nuclear variability expressed chiefly in larger size and probably resulting from a previously disturbed nuclear division. Carcinomas and sarcomas could be distinguished from harmless epithelial growths and the like by the greater nuclear variability in the malignant tumors. Schmitz (1934) found that all sorts of epithelial cells in a lingual cancer had nuclei whose sizes were two or four times the normal. The increase in nuclear size he thought to be due to a doubling of the chromosome number. Wermel and Scherschulskaja (1934) found an increase in nuclear size over normal in a number of cases of human skin cancer, papilloma, carcinomas of the pancreatic duct and gland, fibrosarcomas, and rhabdomyomas. Since an increase in nuclear volume and also, in some cases, a preponderance of diploid mitoses had been reported in cancer literature, and further, since Heiberg had observed that the equatorial plates were larger, they concluded that cancer chromosomes had probably enlarged in size before they gave rise in mitosis to polyploid nuclei. Stein (1935a) described a radium-induced mutant in snapdragons which was expressed in the whole plant in a swelling of all parts of the cells, including the nucleus. This swelling was not accompanied by an increase in the number of chromosomes. The occasional polyploid cells in the plant had still larger dimensions. Here and there cancerous growths broke out which showed all the aberrant cytological phenomena common to animal cancers. This swelling of cells and nuclei of the whole plant is similar to the enlargement of hyper­plastic cells about animal cancers (cf. Page, 1938, and Cowdry and Paletta, 1941). Arndt (1935) found that in cancers of the liver the nuclei were mostly double the volume of the predominant liver nuclei. Arndt thought that the larger nuclei in normal liver had become larger by progressive doublings of the volume of each chromosome without any chromosome splitting. When such cells were excessively stimulated, as by a carcinogenic agent, and under­went mitosis, the large chromosomes broke down to yield polyploid cells which were cancerous. Arndt seems to associate malignancy and polyploidy. Ehrich (1936) noted in 22 human cancers a general shift of the average volume in the predominant nuclear class to a volume two or four times as great as that prevailing in the tissue of origin. He left the question of the cause of the nuclear volume shift undecided between polyploidy and a larger size of the chromosomes. One or the other of these two chromosomal conditions was important in the formation of the cancerous cell. Frugoni (1941) found that active adenomas were characterized by larger average size of nuclei and by greater variability of nuclear size than were normal pituitaries. Many of the active adenomas contained giant nuclei, but Frugoni did not seem to think their large size was closely related to the larger average size of the nuclei. The notion that the nuclear size is necessarily increased in malignancy has been attacked by a number of workers. Outstanding among them has been Schairer (1935, 1937), who claimed that nuclear size was not increased in some of the malignancies he studied, although it was increased in others. Epantschin ( 1928), on studying tar tumors in mice, found that the smallest nuclei in the tumor agreed in size with normal nuclei (but we note that possibly the smallest nuclei in his tumors were those of included normal cells) . Deuticke (1935) wrote that the variations in nuclear volume in skin cancers were paralleled by those in normal skin; the very large nuclei were the result of inflammation rather than of malignancy. Heinkele (1936) found that nuclear sizes in mammary adenocarcinomas of the mouse were the same as the nuclear sizes in the normal lactating breast in one case, and smaller in two cases; however, it should be remembered that the lactating gland has a gen­eralized shift toward larger nucleus and cell size, according to Schairer ( 1935), Jacobj (1935), and Ehrich (1936). Strodtbeck (1937) found that the nuclear volumes of a squamous-cell carcinoma of the cervix agreed with those of the stratum spinosum; however, he admitted that the nuclei of the stratum spinosum were twice as large as the nuclei of the basal cell layer of the squamous epithelium of the cervix, and he said nothing of the chromosomes, so his attempted refutation of the views of Ehrich is not convincing. Deckner (1938) decided that cancer was the result of genie defect, inasmuch as he could discover no common denominator in the aberrations of the nucleus in several cancers. Schlesinger ( 1939) thought it was far safer to diagnose cancer on the basis of the finding of free groups of epithelium-like cells in body fluids, rather than on the basis of finding single cells with enlarged nuclei. Cowdry and Paletta (1941) reported that the cells of mouse and human carcinomas had nuclei no larger than had the neighboring hyperplastic cells. However, Paletta, Cowdry, and Lischer (1941) found that methylcholanthrene hyper­plasia differed from benign hyperplasia of the epidermis in having greater variability in nuclear size and often greater hyperchromatism and larger nucleoli. The relatively uninterpreted reports on enlarged nuclei in tumors are nu­merous. Examples will be cited here. The illustrations in the article by Woglom (1912) show that the Flexner-Jobling adenocarcinoma of the rat has definitely larger nuclei than the tissue into which it was transplanted. Smith ( 1916) stated that the tumor cells of crown gall had large nuclei. The four cases of lymphosarcoma described by MacKenzie (1918) had cells which were chiefly nucleus and which exceeded the size of normal lymphocytes. A liver tumor of the mouse was said by Itami (1918) to have nuclei which were rather large, and the same author (1919) declared the nuclei of a mammary carcinoma of the cat to be large and oval. Kilgore (1920) stated that human malignant skin epithelium differed from non-malignant in several respects, among them being a larger size of the nucleus. The nuclei in a gold fish fibrosarcoma were said to be "prominent," the cells "fairly large," by Scharn­berg and Lucke ( 1922). Largeness and hyperchromatism are attributed to the nuclei in human liver carcinoma by Helvestine (1922). Healthy tumor cells show a slight enlargement of the nucleus, according to Sokoloff (1922). A primary carcinoma of the human lung contained enlarged and hyperchromic nuclei (Seecof, 1924). A case of human lymphatic leukemia had pathological lymphocytes with nuclei larger than normal, as may be seen from the photo­graphs in an article by Wollstein and Bartlett (1925). A human spongio­blastoma of the brain was found to exhibit great variation in the size of its nuclei, some of them being gigantic (Leavitt, 1929) . W. H. Lewis (1927) declared that the malignant spindle cells in rat and mouse sarcoma cultures had nuclei larger than the nuclei of the normal connective tissue cells in the same cultures. MacCarty (1927) described some cells in stomach ulcers which had rounded up and contained large nuclei and nucleoli; he suspected these cells of being carcinomatous. Crawford and Weiss (1928) found that the typical cell in a case of human subacute leukemia seemed to be immature and had a large nucleus. Lewis and Lewis (1932) declared that different rat sarcomas varied in their nuclear size, but all had nuclei somewhat larger than the nuclei of normal fibroblasts. According to Stapel (1935), an adenocar­cinoma of the human salivary gland had nuclei twice as large as the nuclei of the normal gland duct. Graef, Bunim, and Rottino (1936) described adenomatous cells of the pituitary with enlarged nuclei. Barnes and Furth ( 1937) noted that a mouse leukemia contained giant cells which resembled megacaryocytes. Seeger ( 1937) gave illustrations of ascitic cells of Ehrlich mouse carcinoma which clearly have nuclei larger than those of the normal exudate cells of the mouse. Page and MacCarty (1937) found larger nuclei in human ovarian tumors. Page (1938) induced increases in size of nucleus and nucleolus in mouse skin on carcinogenesis with methylcholanthrene, but neither carcinoma-formation nor changes in nucleus and nucleolus were brought about by benzene. Even the malignant cells of the Rous sarcoma in chickens were furnished with large nuclei, although they had developed from several normal cell types with small nuclei in the region of the injection of the cell-free filtrate (Mauer, 1938). According to Tompkins and Cunningham ( 1938), in their review of some leukemias, a fair number of myelogenous and lymphatic leukemias have larger nuclei in the malignant cells. Watson ( 1938), review­ing lymphosarcoma and leucosarcoma, reported the prevalence of large cells and large nuclei among the malignant cells; this did not seem to be true of all. There were large nuclei in the tumor cells of a primary reticulum cell sarcoma of a cow investigated by Bengston (1938). It was related by Reese (1938) that in precancerous melanosis of the conjunctiva there was an increase in the size of some of the cells. McDonald and Broders (1939) maintained that the presence of large cells with large nuclei and nucleoli in human body fluids was diagnostic for cancer. The nuclei of sarcoma cells were declared by W. H. Lewis (1939a) to be larger than the nuclei of adult fibroblasts in vitro. Pullinger ( 1940) stated that the nuclei of mouse skin did not swell after application of non-carcinogenic hydrocarbons, while swelling did occur after the use of several carcinogens. Des Ligneris (1940) said that in the develop­ment of cancer about injected carcinogenic hydrocarbons, the nearby fibro­blasts and their nuclei enlarge, the chromatin is often increased in amount, and irregular mitoses set in. Battle and Stasney (1941) declared that the nuclei of bone marrow melanoma cells were large. In short, the nuclear hypertrophy of many sorts of tumor cells is well estab­lished. By those who have studied the matter more closely, the increase in size over the nuclei of the cells of origin is seen to be a matter of progressive doublings. c. NucLEOLAR EVIDENCE FOR POLYTENE CHROMOSOMES Attempts have been made to use nucleoli in the diagnosis of cancer by workers who thought the nucleoli were more numerous and both actually and relatively larger in the malignant cell than in its normal analogue. Pianese (1896) first noted differences in the nucleoli of normal and mal­ignant cells. MacCarty ( 1927) observed that the carcinomatoid cells of a stomach ulcer contained large nucleoli. Goforth (1927) found enlarged nucleoli in a carcinoma of the parotid duct. Quensel (1928) counted up to 9 or 14 nucleoli in single nuclei of human cancer. Karp (1932) attributed diagnostic significance to the larger nucleolus in cancer cells. According to Soto (1933), sarcomas are more malignant the greater the number of nucleoli per cell. The most important diagnostic feature of tumor cells was thought by Zadek ( 193 3) to be greater numbers of big and irregularly-shaped nucleoli in small nuclei. Pool and Dunlop (1934) declared that cancer cells could be distinguished by their larger nucleolus from the large monocvtes in the blood. According to the work of Guttman and Haloerin (1935) , the nucleolar volume in normal tissues is smaller than that in benign and malignant tumors and hyperplastic tissue. MacCartv and his students have long been concerned with the nucleolus in cancer. The paper of MacCarty (1936) gave measurements on a number of human tissues showing a relatively larger size of the nucleolus in tumors than in corresponding regenerative tissues. Page and MacCarty (1937) found larger nucleoli in ovarian tumors. In fifty human tumors, chiefly carcinomas, Haam and Alexander (1936) reported an increase in the number and the relative size of the nucleoli. Some bizarre shapes of the nucleoli we would attribute to fusion of nucleoli. Weitz­mann (1938) likewise found peculiar shapes of the large nucleoli of human carcinomas. The nucleoli seemed sometimes to survive cytolysis, and Weitz­mann suggested that they might contain a high concentration of carcinogen. Tompkins and Cunningham (1938) mentioned multinucleolaritv in the cells of various leukemias. Kirschbaum and Strong (1939) also found the nucleoli to be more prominent in mouse leukemia. W. H. Lewis (1939a) described sarcoma cells as having more nucleoli, which were often peculiarly shaped, than normal fibroblasts had. McDonald and Broders (1939) attributed larger nucleoli to malignant cells. A human melanoma bore conspicuous nucleoli, according to Gotshalk, Tessmer, and Smith (1940). Mallory (1940) described cells of human stomach carcinoma as having larger nucleoli. Frugoni (1941) noted enlarged nucleoli in active hypophyseal adenomas, and he attributed their larger size to a metabolic activity greater than normal. Admittedly, as Ludford (1922) pointed out, a larger size of the nucleolus may accompany a higher metabolic activity of the cell, and some of the dis­proportionate increases in nucleolar volume in malignant cells may result from their greater activity (see Sokoloff, 1922). However, as we have shown else­where in this paper, the presence of more chromonemal strands than normal lying free of one another in the resting nucleus leads to the formation of a greater number of nucleoli, and the fusion of these plasmosomes makes for fewer but larger nucleoli. Often the compound nucleoli are of lobulate shapes which clearly indicate their plural nature and incomplete and perhaps relatively recent fusion. CHAPTER V DISCUSSION In the foregoing chapters it is seen that our own study, as well as a review of the literature, indicates that in most, if not in all, tumors the chromosomes are abnormally large. This we believe is due to the fact that they contain more chromonemata (strands) than do normal chromosomes. There are other possibilities, however, which must be considered. Since it has been shown that the size of a given chromosome may be altered by different states of hydration (Shinke, 1939), it might be that the larger nuclear size in cancer, as well as the large chromosome size, results from hydration. However, hydration ought to affect contiguous cells in like manner, but, as a matter of fact, class I and class V nuclei may be immediately adjacent to one another in tumors, and chromosomes of several sizes may be seen in the same microscopic field. Hydration does not explain the periodicity of nuclear volumes, and it could not change the number of nucleolar organizers. According to Politzer (as quoted by Geitler, 1938b), the increase in chromo­some size in carcinomas would be the result of a change in spiralization. This explanation is in conflict with the observation of Ludford (1936) that colchi­cine contracts fetal kidney chromosomes more than it contracts the chromosomes of mouse carcinoma 63, as though the normal chromosomes had been more loosely coiled than the cancerous. There is evidence for the normal occurrence of polytene chromosomes. Their formation is part of normal differentiation in some insect tissues (cf. Painter and Griffen, 1937). The work of Hertwig (1933, 1939) indicates a polymerism in spermatogonial chromosomes in certain mammals and in early cleavage chromosomes of the mouse. Jacobj (1935) found many nuclear volume classes in human tissue and expressed the belief that they did not reflect polvploidv so much as discontinuous size changes of the chromosomes; he claimed that in the human there were several nuclear classes ( erythroblasts and microlymphocytes) below class I of the embryo in volume, and he there­fore suggested that the chromosomes of class I nuclei were polymeric. Karplus ( 1930) declared there was no constancy of form or size of chromosomes in non-pathological human pleura, peritoneum, and amnion. Levi (1925) wrote that it was generally assumed that chromosomes varied in size in different tissues in order to account for different nuclear and cellular volumes. It appears from the above that there are polytene chromosomes in tissues other than tumors. Polytene chromosomes have been experimentally produced in a number of ways, without forming a cancer. Haecker (1904) described a heterotypic mitosis (which possibly involved misinterpreted diplochromosomes) in the first cleavage of Cyclops, after etherization. Barratt (1907) applied Scharlach R to the rabbit ear and caused hypertrophy of the Malpighian layer; the chromo­somes were larger than those in the testis, and in some cases they were in pairs or in undivided pairs. After the Scharlach R was withdrawn, no further growth took place. White (1935) x-rayed grasshoppers and obtained V­shaped diplochromosomes instead of the usual rod-shaped chromosomes in spermatogonial divisions. Interestingly, the crowded metaphase plate which held the diplochromosomes was no larger than normal. Barber (1940) reviewed the work of investigators who had found what Barber interpreted as diplochromosomes, produced variously under the in­fluence of heat, auxins, gene mutation, age, or occurring naturally. Barber is of the opinion that diplochromosomes are formed when chromatids divide and the nucleus does not, as in an abortive early prophase in which the centro­meres fail to divide because the nuclear membrane is still intact (according to Darlington, 1937). A break of the centromeres in a subsequent anaphase produces cells with tetraploid numbers of normal, unpaired chromosomes. Barber thinks that this also accounts for the polyploidy in tumors. We do not agree with Barber's opinion that tetraploid cells in mouse tumors are formed by breakdown of diplochromosomes. We hold rather that they are formed by a growth beyond the diplochromosome stage and a splitting of the reduplicated diplochromosomes in an endomitotic cycle. If Barber is correct in his idea that tetraploids are formed from diploid cells when diplochromo­somes break down in anaphase, then the new tetraploids should have chromo­somes one-half the size of the diplochromosomes and the nuclear volume should be no greater for the new tetraploid than for the diplo-diploid cells from which they were derived. But this is not true in tumor XG and the other mouse tumors studied. The chromosomes of the tetraploid set are the same size as those of the tumor-diploids, and the tetraploid nuclei have twice the volume of the diploid nuclei. Moreover, Lewis and Strong (1934) point out that cancer chromosomes have wide prophasic and metaphasic splits both in diploid and in polyploid mitoses. Hence the cancer diplochromosome does not seem to be an ephemeral thing, and its formation is apparently irrever­sible. The widespread occurrence of polytene chromosomes in cancers indicates their association with the fundamental cancerous change and emphasizes the notion that the many diverse "causes" of cancer must be channeled through the same stream. We might profitably inquire into the origin of polytene chromosomes in carcinogenesis. There is doubtless a stimulation of the synthesis of nucleic acid and new chromonemal proteins. Once the strands are reduplicated, what holds them together? It could not be the matrix which does this, for in plant tumors (Stein, 1935a; Dermen, 1941) the multiple chromatids may be seen as fairly separate structures. The strands could very well be held together by an undivided centromere. In polytene chromosomes the centromere may be assumed to lag behind the chromonemata in growth and division. This assumption may be strengthened for the case of cancer by the considerations which follow: ( 1) The work of Pollister {1939) has demonstrated a close genetic relationship between the centromeres and the centrioles, which form the focal points of the asters in cell division; according to Schrader ( 1936), the centriole and the centromere stain similarly with the Kull stain, while the chromosome proper stains differently; hence it is to be expected that the centromere and the centriole should behave similarly. (2) The majority of the approximately diploid mitoses in cancer are bipolar, with the normal number of two centrosomes; Fogg and Warren ( 1941) state that about 96 per cent of the cells of rat carcinoma 256 and mouse sarcoma 180 have only two centrioles per cell; therefore we may say that the centrioles have not been reduplicated beyond the normal, although the chromonemata have been. ( 3) If consideration ( 1) holds the centromeres should likewise be expected not to have been doubled beyond the normal in the diplochromosomes of tumors. Normal polyploid cells have reduplicated centrioles which often elicit multi­polarity in mitoses (Clara, 1931, on tetraploids in rabbit liver; and Heiden­hain, 1912, on megacaryocytes). In cancers, too, the polyploid mitoses often seem to be multipolar. This means that in the growth and complete splitting of chromosomes to yield polyploid nuclei, the centrioles divide in step with the centromeres. Nuclei which have become pol;rploid by fusion would be expected to form multipolar spindles in mitosis. Lewis and Lewis (1932) suggested that upsets in the mechanics of the centrosomes might be responsible for both heteroploidy and malignancy. But, we are still faced with the question of what causes the centromere-chromonema upset. That this is one of the early changes in carcinogenesis is evident from the observation of Hearne (1936) that mouse fibroblasts treated in vitro with methylcholanthrene underwent divisions in which chiasmata seemed to be formed, and from her later work (Hearne-Creech, 1939) on the same subject. She found that those treated cultures which grew most rapidly also exhibited the largest number of precocious prophase splits of the chromosomes, which made for a resemblance to meiosis. The inference from her papers is clear: methylcholanthrene produces a doubleness of the chromosomes beyond their customary valence. If the carcinogen stimulates growth and splitting of the chromonema, why does it seemingly inhibit growth or division of the centriole and centromere? We may speculate on the possibility that the carcinogens differentially affect the proteins of the centromere-centriole combination and those of the chromo­nema proper. It might be that some of the enzymes concerned in the synthesis of the centromere substance are lowered in activity, since it is known that some of the enzyme systems in cancer are variously altered in their activities (Dodds and Dickens, 1940). Later recovery might allow mitoses with the necessarily polytene chromosomes. This would in a sense parallel the idea of carcinogenesis held by Badger, Elson, Haddow, Hewett, and Robinson (1942); they think that carcinogenesis involves an optimal, not a maximal, inhibition of normal cell growth followed by an adaptation which allows the changed and often dedifferentiated cells to continue growth; the cells of the new cancerous line are permanently altered, and their further maintenance and multiplication do not depend on the continued presence of the carcinogen. The cause of the formation of polytene chromosomes is obviously unsettled, and we leave it thus. What effects can we expect the multiplication of genes involved in polytene chromosomes to have on the cells that contain them? Jordan (1939) went so far as to consider the possibility that cancers might be caused by reduplicated parts of chromosomes present in excess, either vaga­bonding as viruses or reduplicated linearly or in parallel, as in the salivary chromosomes of some flies. We see that cancer chromosomes do bear some resemblance to the far more highly polytene salivary chromosomes. Whether or not polytene chromosomes have anything to do in causing cancer, they should nevertheless lead to some changes. Polyploidy is closely akin to multiple-strandedness of diploid chromosomes. Some of the pheno­typic effects of polyploidy might be expected to appear in a cell with polytene chromosomes. The changes incident to polyploidization are thought to be due to incomplete dominance of some genes. This may be expressed in another way: some genes have a cumulative effect when present several times, and others do not have a cumulative effect. This results in a changed balance when all the genes are present several times more than normal in the same nucleus. Besides the changes effected by polyploidy in cell size, organ size, and growth rate (Miintzing, 1936), there are differences produced in the proportions of definite substances, among which in various instances have been sugars, nico­tine, ascorbic acid, and carotenoids (cf. review by Randolph, 1941) . Some of the known chemical differences between normal and cancerous tissues might arise because of the polytene nature of the cancer chromosomes. A polyploidy masked in polytene chromosomes, as in tumor-diploid cells, might not have quite the same effect as polyploidy of simple chromosomes, . because there might be something in the nature of a position effect from the polytene chromosomes. This would be especially likely for the genes near the centromere, for the genes in that region would be closer to their homologous sister genes than would be true, on the average, in the random distribution of chromosomes in simple polyploidy. Hence the local concentration of cer­tain primary gene-products might be greater near the centromeres of polytene chromosomes than near the centromeres of simple chromosomes in polyploid nuclei. Therefore, concentration thresholds of certain reactions might perhaps be reached sooner in the case of the polytene chromosomes. A very interesting deduction may be drawn from the paper of Greenough ( 1925), in which he described certain morphological characteristics that accompany various stages of malignancy in breast cancer. Among those as­sociated with high malignancy are greatly variable nuclear size, hyper­chromatism, and abnormal mitoses. Furthermore, Evans, Barnes, and Brown ( 1942) state that high malignancy in carcinoma of the human prostate is marked histologically by enlarged nuclei, more visible nucleoli, a greater num­ber of mitotic figures, and scantier but more heavily staining cytoplasm (i.e., the cytoplasm probably contains more ribonucleic acid). Since these char­acters are individually expressions either of nucleic acid unbalance or of injury to the centromeres and repeated endomitosis, we may conclude that malignancy is paralleled in degree by frequency of endomitosis and concen­tration of nucleic acids. CHAPTER VI SUMMARY 1. Ten mouse tumors, largely sarcomas, were studied, as well as some normal adult and embryonic tissues of the mouse. 2. Aceto-carmine squash preparations were chiefly used. 3. The methods of counting and measuring are described. 4. Cancer nuclei fall into volume classes like the nuclei of normal tissues do, although the classes are far from being as distinct, partly because of con­siderable aneuploidy in the tumors. 5. Class I in the tumors is composed largely of small diploid nuclei, pre­sumably non-malignant, with a maximum of four plasmosomes and with chromosomes which are quite like the chromosomes of normal diploid em­bryonic cells. Other nuclei in the volume range of class I may be small, hypodiploid malignant nuclei. 6. Class II, which is usually the prevailing class in the tumors, is made up of nuclei which are diploid or about diploid in chromosome number, but which contain chromosomes twice as large as the chromosomes of normal cells. The chromosomes are larger because they consist of twice as many chromonemata as do normal chromosomes. Hence there are twice as many nucleolar organ­izers present and twice as many plasmosomes may be formed. 7. Class III nuclei are ideally twice as large as the nuclei of class II. The tetraploid nuclei of class III have chromosomes of the same size as the chromo­somes of class II. But in the diploid class III nuclei, each chromosome has four times as many chromonemata as normal chromosomes or twice as many as the diplochromosomes of tumorous class II nuclei. The maximum plasmo­some number is sixteen, but this is often not attained because the close proxi­mity of the nucleolar organizers on sister chromonemata allows plasmosomes to fuse. 8. Among the nuclei of class IV there are octoploids with chromosomes twice normal size, tetraploids with chromosomes four times normal size, and diploids with chromosomes eight times normal size and strand-number. Thirty­two nucleolar organizers are present in the chromosomal complement. 9. The succeeding classes are built up in the same manner, and the nuclei of a given class are presumably derived from the nuclei of the next lower class by endomitosis with or without division of the centromeres. 10. The original formation of diplochromosomes in carcinogenesis-in the endomitosis without division of the centromeres which marks the change from a non-malignant ( ?) class I nucleus to a malignant class II nucleus-seems to be irreversible, for no polyploids with small chromosomes have been found in the tumors. All cancer chromosomes except those of class I nuclei are larger than normal and have more strands. 11. Polytene chromosomes seem to be widespread in cancers, for many tumors of man and other organisms have been found to be characterized by one or more of these sequelae of polytene chromosomes: greater nuclear volume, an increased number of nucleoli, and larger size of chromosomes, which often appear double. 12. Several other possible interpretations of the larger size of chromosomes in cancer are refuted. 13. Polytene chromosomes are known to occur normally, and they have also been produced experimentally in tissues which did not thereafter become malignant. Therefore the polytene nature of cancer chromosomes could hardly cause cancer, but it must be a phenomenon resulting from or concomitant with carcinogenesis. 14. The chromosomes of cancer probably become polytene because the centromere lags behind the chromonema in growth and division in endomi­tosis, possibly as an effect of the carcinogen. 15. Since the polytene condition of chromosomes is related to the polyploid, and because polyploidy is known to influence the production of various sub­stances, there are probably changes in the chemistry of the cancer cell elicited by its polytene chromosomes. 16. The frequency of endomitosis in tumors is proportional to the degree of malignancy. The increase of cytoplasmic basophily with malignancy in tumors seems to be the result of a greater concentration of ribonucleic acid, whose presence is associated with protein synthesis. BIBLIOGRAPHY Alexenko, B., and Natansohn, A. 1933. Karyologische Untersuchung der boesartigen Geschwuelste. I. Z. Krebsforsch., 38: 264-273. Andres, A. H. 1932. Zellstudien an Menschenkrebs. Der chromosomale Bestand im Primaertumor und in der Metastase. Z. Zellforsch., 16: 88-122. Arndt, G. 1935. Kernstudien zur Unterscheidung von Regeneration und Geschwulstbil­dung. Z . Krebsforsch., 41: 393--444. Arnold, J. 1879. Beobachtungen ueber Kerntheilungen in den Zellen der Geschwuelste. Virchow's Arch. path. Anat., 78: 279-301. Badger, G. M., Elson, L. A., Haddow, A., Hewett, C. L., and Robinson, A. M. 1942. The inhibition of growth by chemical compounds. Proc. Roy. Soc., B, 130: 255-299. Barber, H. N. 1940. The suppression of meiosis and the origin of diplochromosomes. Proc. Roy. Soc., B, 128: 170-185. Barnes, W. A., and Furth, J. 1937. A transmissible leukemia in mice with atypical cells. Amer. lour. Cancer, 30: 75-94. Barratt, J. 0 . W . 1907. On mitosis in proliferating epithelium. Proc. Roy. Soc., B, 79: 372-377. Bashford, E. F., and Murray, J. A. 1904. The significance of the zoological distribution, the nature of the mitoses, and the transmissibility of cancer. Proc. Roy. Soc., 73: 66-77. Bashford, E. F ., and Murray, J. A. 1905. On the occurrence of heterotypical mitoses in cancer. Proc. Roy. Soc., B, 77: 226-232. Bashford, E. F., and Murray, J. A. 1908. On the occurrence of heterotypical mitoses in cancer. Third Sci. Rep. Imp. Cane. Res. Fund: 61-68. Battle, J. D., and Stasney, J. 1941. Malignant melanoma cells in the bone marrow, Arch. Path., 31: 631-633. Bengston, J. S. 1938. Primary reticulum cell sarcoma of a cow with widespread metas­tases. Amer. lour. Path., 14: 365-376. Beverwijk, A. van. 1934. Zuechtung von Walker-Rattenkarzinom Nr. 256 in vitro. Arch. exp. Zellforsch., 16: 151-158. Borst, M . 1924. Allgemeine Pathologie der malignen Geschwuelste. Leipzig. Brachet, J. 1940. La detection histochimique des acides pentosenucleiques. C. r. Soc. de Biol., 133: 88-90. Carothers, E. E. 1913. The Mendelian ratio in relation to certain orthopteran chromo­somes. lour. Morph., 24: 487-512. Clara, M. 1931. Ueber den Bau der Leber beim Kaninchen und die Regenerations­erscheinungen an diesem. Gewebe bei experimenteller Phosphorvergiftung. Z. mikr.­anat. Forsch., 26: 45-172. Cowdry, E. V., and Paletta, F. X. 1941. Changes in cellular, nuclear, and nucleolar sizes during methylcholanthrene epidermal carcinogenesis. lour. Nation. Cancer Inst., 1: 745-759. Crawford, B. L., and Weiss, E. 1928. A consideration of leukemia, with especial reference to a subacute form. Amer. lour. Med. Sci., 175: 622-630. Crew, F. A. E., and Koller, P. C. 1932. Chiasma frequency and crossing-over in the mouse. lour. Genet., 26: 359-383. Darlington, C. D. 1937. Recent Advances in Oytology. 2d Ed. London: Churchill. Darlington, C. D. 1942. Chromosome chemistry and gene action. Nature, 149: 66-69. Dearing, W. H . 1934. The material continuity and individuality of the somatic chromo­somes of Ambystoma tigrinum, with special reference to the nucleolus as a chromo­somal component. lour. Morph., 56: 157-179. Deckner, K. 1938. Cytologische Studien an Krebszellen. Z. Krebsforsch., 48: 129-148. Dermen, H . 1941. lntranuclear polyploidy in bean induced by naphthalene-acetic acid. lour. Hered., 32: 133-138. Deuticke, K. 1935. Die Kernverhaeltnisse bei Hautcarcinomen. Z. Krebsforsch., 43: 39-53. Dimitri, V., and Alem, C. 1938. Tumor suprarenal con metastasis en pulmon y cerebro. Rev. Neural. Buenos Aires, 3: 33-63. Dodds, E. C., and Dickens, F. 1940. The biochemistry of malignant tissue. Ann. Rev. Biochem., 9: 423-458. Ehrich, W. E. 1936. Nuclear sizes in growth disturbances. With special reference to the tumor cell nucleus. Amer. lour. Med. Sci., 192: 772-790. The University of Texas Publication Epantschin, W. 1928. Kernmessungen beim Teerkrebs der weissen Maus. Z. Krebsforsch., 26: 439-449. Ernst, H. 1938-9. Meiosis und Crossing over. Z. Bot., 33: 241-294. Evans, H. M., and Swezy, 0. 1929. The chromosomes in man, sex and somatic. Mem. Univ. Calif., 9: 1. Evans, N., Barnes, R. W., and Brown, A. F. 1942. Carcinoma of the prostate. Correlation between the histologic observations and the clinical course. Arch. Path., 34: 473-483. Farmer, J. B., Moore, J. E. S., and Walker, C. E. 1904. Ueber die Aehnlichkeit zwischen den Zellen maligner Neubildungen beim Menschen und denen normaler Fortpflanz­ungsgewebe. Biol. Centralbl., 24: 1-7. Fogg, L. C., and Warren, S. 1941. Some cytologic effects of therapeutic radiation. Cancer Research, 1: 649-652. Fried, B M. 1924. Primary carcinoma of the liver. Amer. lour. Med. Sci., 168: 241-267. Fried, B. M., and Buckley, R. C. 1930. Primary carcinoma of the lungs. Arch. Path., 9: 483-527. Frugoni, P. 1941. Cytologic studies on hypophyseal adenomas. Trans. Amer. Mier. Soc., 60: 261-272. Gates, R. R. 1942. Nucleoli and related nuclear structures. Bot. Rev., 8: 337-409. Geitler, L. 1938a. Ueber das Wachstum von Chromozentrenkernen und zweierlei Hetero­ chromatin bei Bluetenpflanzen. Z. Zellforsch., 28: 133-153. Geitler, L. 1938b. Chromosomenbau. Protoplasma-Monographien, 14. Berlin: Borntraeger. Geitler, L. 1939. Die Entstehung der polyploiden Somakerne der Heteropteren durch Chromosomenteilung ohne Kernteilung. Chromosoma, 1: 1-22. Goforth, J. L. 1927. Carcinoma developing in the parotid (Stensen's) duct. With clinico­pathological report of a case. Amer. lour. Med. Sci., 173: 624-629. Goldschmidt, R., and Fischer, A. 1929. Chromosomenstudien an Carcinomzellen in vitro. Z. Krebsforsch., 30: 281-285. Gotshalk, H. C., Tessmer, C. F., and Smith, J. W. 1940. Malignant melanoma of the palate. Arch. Path., 30: 762-765. Graef, I., Bunim, J. J., and Rottino, A. 1936. Hirsutism, hypertension and obesity associ­ated with carcinoma of the adrenal cortex. Arch. Intern. Med., 57: 1085-1103. Greenough, R. B. 1925. Varying degrees of malignancy in cancer of the breast. ]our. Cane. Research, 9: 453-463. Guttmann, P. H., and Halperin, S. 1935. Nuclear-nucleolar volume ratio in cancer. Amer. lour. Cancer, 25: 802-806. Haam, E. von, and Alexander. H. G. 1936. Cytological studies of malignant tumors. Amer. lour. Clin. Path., 6: 394-414. Haecker, V. 1904. Ueber die in malignen Neubildungen auftretenden heterotypischen Teilungsbilder. Biol. Zentralbl., 24: 787-797. Hamilton-Paterson, J. L. 1941. Mitosis in monocytic leukemia. lour. Path. and Bact., 52: 273-274. Hansemann, D. von. 1904. Ueber Kernteilungsfiguren in boesartigen Geschwuelsten. Biol. Zentralbl., 24: 189-192. Hearne, E. M. 1936. Induced chiasma formation in somatic cells by a carcinogenic hydro­carbon. Nature, 138: 291. Hearne-Creech, E. M. 1939. Carcinogenic and related non-carcinogenic hydrocarbons in tissue culture. I. Amer. lour. Cancer, 35: 191-202. Heiberg, K. A. 1908. Ueber die Erklaerung einer Verschiedenheit der Krebszellen von anderen Zellen. Nord. med. Ark., 41: (Abt. 2): 1-20. Heiberg, K. A. 1921. Studien ueber Haut-Epithel-Atypie bei Krebs-und Granula1'ions­ gewebe und die diagnostische Verwendung der Kerngroesse. Virchows Arch. path. Anat., 234: 469-480. Heiberg, K. A., and Kemp, T. 1929. Ueber die Zahl der Chromosomen in Carcinomzellen beim Menschen. Virchows Arch. path. Anat., 273: 693-700. Heidenhain, ~-1?12. Ueber '?willings-, Drillings-und Vierlingsbildungen der Duenndarm­zotten, em Be1trag zur Te1lkoerpertheorie. Anat. Anz., 40: 102-147. Heinkele, T. 1936. Kernmessungen an Mammacarcinomen der Maus. Z. Krebsforsch 43· 323-336. ., • Heitz, E. 1931. Die Ursache der gesetzmaessigen Zahl Lage Form und Groesse pflanz­licher Nukleolen. Planta, 12: 775-844. ' ' Helvestine, F. 1922. Primary carcinoma of the liver. lour. Cane. Res., 7: 209-227. Nuclear Phenomena tn Mouse Cancers Hertwig, G. 1933. Die dritte Reifeteilung in der Spermiogenese des Menschen und der Katze und ihre experimentelle Ausloesung durch Prolan im jugendlichen Rattenhoden. Z. mikr.-anat. Forsch., 33: 373-400. Hertwig, G. 1939. Der Furchungsprozess des Maeuseeies, ein Beispiel fuer die wiederholte Volumenhalbierung polymeren Kerne und Chromosomen durch multiple Succedanteil­ungen. Z. mikr.-anat. Forsch., 45: 37-45. Hirschfeld, H., and Klee-Rawidowicz, E. 1929. Zytologische Untersuchungen am Sarkom­gewebe in der in-vitro-Kultur. Z. Krebsforsch., 30: 406--427. Hurst, E. W. 1937. Myxoma and the Shope fibroma. 1: The histology of myxoma. Brit. lour. Exp. Path., 18: 1-15. Itami, S. 1918. Adenocarcinoma or adenoma of the liver in a mouse. lour. Cane. Res., 3: 275-277. ltami, S. 1919. A mammary carcinoma in the cat. lour. Cane. Res., 4: 19-22. Jacobj, W . 1925. Ueber das rhythmische Wachstum der Zellen durch Verdoppelung ihres Volumens. Arch. Entw.-mech., 106: 124--192. Jacobj, W. 1935. Die Zellkerngroesse beim Menschen. Ein Beitrag zur quantitativen Cytologie. Z. mikr.-anat. Forsch., 38: 161-240. Jaffe, H. L., Lichtenstein, L., and Portis, R. B. 1940. Giant cell tumors of bone. Arch. Path., 30: 993-1031. Jordan, P . 1939. Heuristische Bemerkungen zur Krebstheorie. Fund. Radio[., 4: 177-180. Karp, H. 1932. Die Cytodiagnostik maligner Tumoren aus Punktaten und Sekreten. Z. Krebsforsch., 36: 579--005. Karplus, H. 1930. Ein Beitrag zur Kenntnis der somatischen Mitose beim Menschen. Z. Zellforsch., 10: 38--52. Kemp, T . 1930. Ueber die somatischen Mitosen bei Menschen und warmbluetigen Tieren unter normalen und pathologischen Verhaeltnissen. Z. Zellforsch., 11: 429-444. Kilgore, A. R. 1920. Skin involvement in breast cancer with reference to its bearing on the interpretation of appearances of transition between normal epithelium and cancer. lour. Cane. Res., 5: 291-300. Kirschbaum, A., and Strong, L. C. 1939. Leukemia in the F strain of mice: observations on cytology, general morphology, and transmission. Amer. lour. Cancer, 37: 400--413. Kuhn, E. 1938. Ueber die Trabantenchromosomen bei normalen und immerspaltenden Sippen von Matthiola incana. Z. indukt. Abstamm.-u. Vererbungsl., 74: 388-408. Leavitt, F. H. 1929. Brain tumors in childhood. A clinicopathological study. Amer. lour. Med. Sci., 178: 229-236. Levi, G. 1925. Wachstum und Koerpergroesse. Ergeb. Anat. u. Entw., 26: 87-342. Levine, M . 1931. Studies in the cytology of cancer. Amer. lour. Cancer., 15: 144--211, 788--834, 1410-1494. Lewis, M. R. 1930. Nuclear structures of malignant cells. Anat. Rec., 45: 268-269. Lewis, M. R. 1932. Behavior of the chromosomes in cells of malignant growths. Anat. Rec., Suppl., 52: 65. Lewis, M. R., and Lewis, W. H. 1932. The malignant cells of Walker rat sarcoma no. 338. Amer. lour. Cancer, 16: 1153. Lewis, M. R., and Lockwood, J. 1929. The tetraploid number of chromosomes in the malignant cell of the Walker rat sarcoma no. 1. Bull. Johns Hopkins Hosp., 44: 187-197. Lewis, M. R., and Strong, L. C. 1934. A study of spontaneous tumors of the mouse by the tissue culture method. Amer. lour. Cancer, 20: 72--95. Lewis, W. H. 1927. Sarcoma cells. Arch. exp. Zellfo;5.;h., 5: 143-156. Lewis, W. H. 1939a. Some cultural and cytological characteristics of normal and malig­nant cells in vitio. Arch. exp. Zellforsch., 23: 8-26. Lewis, W. H. 1939b. Dibenzanthracene mouse sarcomas. Histology. Amer. lour. Cancer, 37: 521--630. Des Ligneris, M. J. A. 1940. Precancer and carcinogenesis. Amer. lour. Cancer, 40: 1-46. Ludford, R. J. 1922. The morphology and physiology of the nucleolus. lour. Roy. Mier. Soc., 1922, pp. 113-150. Ludford, R. J. 1930a. Chromosome formation without spindle development in cancer cells, and its significance. Ninth Sci. Rep. Imp. Cane. Res. Fund, 109-120. Ludford, R. J. 1930b. The chromosomes of transplantable tumor cells. Ninth Sci. Rep. Imp. Cane. Res. Fund, 149-153. Ludford, R. J. 1933. The behavior of the malignant and non-malignant cells of trans­plantable tumors in tissue cultures. Arch. exp. Zellforsch., 14: 42--55. The University of Texas Publication Ludford, R. J. 1936. The action of toxic substances upon the division of normal and malignant cells in vitro and in vivo. Arch. exp. Zellforsch., 18: 411-441. MacCarty, W. C. 1927. Chronic ulcer and carcinoma of the stomach. Amer. lour. Med. Sci., 173: 466--472. MacCarty, W. C. 1936. The value of the macronucleolus in the cancer problem. Amer. lour. Cancer, 26: 529. McClintock B. 1934. The relation of a particular chromosomal element to the develop­ment of the nucleoli in Zea mars. Z. Zellforsch., 21: 294--328. McDonald, J. R., and Broders, A. C. 1939. Malignant cells in serious effusions. Arch. Path., 27: 53-60. MacKenzie, G. M . 1918. Cases of typical and atypical lymphosarcoma. lour. Cane. Res., 3: 93-106. Mallory, T. B. 1940. Carcinoma in situ of the stomach and its bearing on the histogenesis of malignant ulcers. Arch. Path., 30: 348-362. Manton, I. 1935. Some new evidence on the physical nature of plant nuclei from intra­ specific polyploids. Proc. Ror. Soc., B, 118: 522-547. Mauer, G. 1938. Die Histogenese des Rous-Sarkoms Nr. 1. Z . Krebsforsch., 48: 58-63. Mendelsohn, W. 1935. The significance of abnormal mitosis in the development of malig­ nancy. Amer. lour. Cancer, 24: 626-636. Moellendorff, W. von. 1940. Zur Kenntnis der Mitose X. Ueber das Vorkommen patho­logischer Mitosen in Gewebskulturen des Brown-Pearce-Tumors. Zugleich ein Beitrag zur Frage der Tumormitosen. Z. Zellforsch., 31: 60-90. Miintzing, A. 1936. The evolutionary significance of autopolyploidy. Hereditas, 21: 263-376. Nomikos, B. 1910. Vergleichende Untersuchungen ueber die Kerngroessen bei den ver­ schiedenartigsten epithelialen Neubildungen, insbesondere bei Carcinomen. Inaug.-Diss. Wiirzburg. Page, R. C. 1938. Cytologic changes in the skin of mice during application of carcino­genic agents. Arch. Path., 26: 800-813. Page, R. C., and MacCarty, W. C. 1937. Cytology of ovarian tumors. Arch. Path., 24: 1-7. Painter, T . S. 1939. An aceto-carmine method for bird and mammalian chromosomes. Science, 90: 307-308. Painter, T. S. 1941. An experimental study of salivary chromosomes. Symp. Quant. Biol., 9: 47-53. Painter, T. S., and Griffen, A. B. 1937. The structure and the development of the salivary gland chromosomes of Simulium. Genetics, 22: 612-633. Painter, T. S., and Reindorp, E. 1939. Endomitosis in the nurse cells of the ovary of Drosophila melanogaster. Chromosoma, 1: 276-283. Paletta, F. X., Cowdry, E. V., and Lischer, C. E. 1941. Comparison of methylcholan­threne hyperplastic epidermis with benign hyperplastic epidermis in healing wounds. Cancer Research, 1: 942-952. Pianese, G. 1896. Beitrag zur Histologie und Aetiologie des Karzinoms. Jena: Fischer. Pollister, A. W. 1939. Centrioles and chromosomes in the atypical spermatogenesis of vivipara. Proc. Nat. Acad. Sci., 25: 189-195. Pool, E. H., and Dunlop, G. R. 1934. Cancer cells in the bloodstream. Amer. lour. Cancer, 21: 99-102. Potter, J. S., and Richter, M. N. 1933. Mouse leukemia. VIII. Continuity of cell line­age in transmission lines of lymphatic leukemia. Arch. Path., 15: 198-212. Pullinger, B. D. 1940. The first effects on mouse skin of some polycyclic hydrocarbons. lour. Path. and Bact., 50: 463-471. Quensel, U. 1928. Zrtologische Untersuchungen von Erguessen der Brust und Bauchhoehle, mit besonderer Beruecksichtigung der karzinomatoesen Exsudate. Uppsala: Almquist & Wiksells. Randolph, L. F. 1941. An evaluation of induced polyploidy as a method of breeding crop plants. Amer. Nat., 75: 347-363. Reese, A. B. 1938. Precancerous melanosis and diffuse malignant melanomas of the con­junctiva. Arch. Ophth., 19: 354--365. Schairer, E. 1935. Kernmessungen und Chromosomenzaehlungen an menschlichen Gesch­wuelsten. Z. Krebsforsch., 43: 1-38. Schairer, E. 1937. Die Beziehungen von Kerngroesse und Geschwulstwachstum, untersucht am experimentellen Teerkrebs. Z. Krebsforsch., 45: 279-297. Scharnberg, J. F., and Lucke, B. 1922. Fibrosarcoma of the skin in a gold fish (Carassus auratus). lour. Cane. Res., 7: 151-161. Schlesinger, M. J. 1939. Carcinoma cells in thoracic and in abdominal fluids. Arch. Path., 28: 283-297. Schmitz, W. 1934. Das Gesetz des Wachstums in konstanten Proportionen (Heidenhain) in Beziehung zum Krebsproblem. Z. Krebsforsch., 41: 372-381. Schopper, W. 1933. Das Walker-Rattensarkom No. 319 in der Gewebekultur. Arch. exp. Zellforsch., 14: 14-41. Schrader, F. 1936. The kinetochore or spindle fibre locus in Amphiuma tridactylum. Biol. Bull., 70: 484-498. Schultz, J., Caspersson, T., and Aquilonius, L. 1940. The genetic control of nucleolar com­position. Proc. Nat. Acad. Sci., 26: 515-523. Seecof, D. P. 1924. Multiple primary malignant neoplasms. Report of a case of carci­noma of lung and carcinoma of rectum each producing metastases. lour. Cane. Res., 8: 213-220. Seeger, P. G. 1937. Untersuchungen am Tumoraszites der Maus. 1. Mitteilung: Vital­faerbbarkeit der Asziteszellen. Arch. exp. Zellforsch., 20: 280-335. Selbie, F. R. 1938. Tumors in rats and mice following the injection of thorotrast. Brit. lour. Exp. Path., 19: 100-107. Selle, W. A., Brindley, P., and Spies, J. W. 1941. The production of tumors by trans­plantation of normally appearing liver cells from animals previously injected with methylcholanthrene. Cancer Research, 1: 618-619. Shinke, N. 1939. Experimental studies of cell-nuclei. Mem. Coll. Sci., Kyoto Imp. Univ., B, 15: 1-126. Smith, E. F. 1916. Studies on the crown gall of plants. Its relation to human cancer. lour. Cane. Res., 1: 231-309. Sokoloff, B. 1922. The nucleo-cytoplasmic ratio and cancer. lour. Cane. Res., 7: 395-415. Soto, E. V. 1933. Los constantes nucleolares de Gonzalez Guzman y el metabolismo de los tejidos normales y cancerosos. Estudio de los nucleolos en los sarcomas. Rev. Mex. Biol., 13: 49-91. Stapel, E. 1935. Ueber die Wachstumsproportionen einiger Speicheldruesengeschwuelste. Z. Krebsforsch., 42: 488-496. Stein, E. 1935a. Weitere Analyse der Gruppe A von den durch Radiumbestrahlung veraenderten Erbanlagen bei Antirrhinum. Kern-und Zellveraenderung in der krebsigen Gewebeentartung. Z. indukt. Abstamm.-u. Vererbungsl., 69: 303-326. Stein, E. 1935b. Zur Genetik, Histologie und Zyto!Ggie einer neuen, durch Radium­bestrahlung erzeugten krebsigen Entartung von Antirrhinum majus. (Aus Gengruppe B.) Z. indukt. Abstamm.-u. Verebungsl., 70: 525-531. Strodtbeck, W . 1937. Ueber Kernmessungen an Portioepithel, Portiocarcinom, und Uterus­schleimhaut. Z. Krebsforsch., 45: 268-278. Tennant, R., Stern, K. G., and Liebow, A. A. 1942. The stimulating action of nucleic acids upon mouse heart fibroblasts. Cancer Research, 2: 218-222. Tompkins, E. H., and Cunningham, R. S. 1938. The application of the supravital method to the study of blood in pathological conditions. Handbook of Hematology, 1: 585-640. Ed. Hal Downey. New York: Paul B. Hoeber. Tuft, L. 1930. Primary squamous cell carcinoma of the vagina complicating late preg­nancy in a patient sixteen years old. Arch. Path., 9: 654-658. Watson, C. J. 1938. Lymphosarcoma and leucosarcoma. Interrelations of leucemia and diseases of the "pseudoleucemia" group (Hodgkin's disease, aleucemic lymphadenosis, lymphosarcoma) . Handbook of Hematology, 4: 3049-3106. Ed. Hal Downey. New York: Paul B. Hoeber. Weitzman, G. 1938. Epithel und Karzinom des erwachsenen Menschen in vitro. Arch. exp. Zellforsch., 22: 347-355. Wenrich, D. H. 1916. The spermatogenesis of Phrynotettix magnus with special reference to synapsis and the individuality of the chromosomes. Bull. Mus. Comp. Zool., Harvard Coll., 60. Wermel, E. M., and Scherschulskaja, L. W. 1934. Studien ueber Zellengroesse und Zellen­wachstum. VII. Ueber die Groesse der boesartigen Zellen und ihre Variabilitaet. Z. Zellforsch., 20: 54-76. White, M. J. D. 1935. The effects of x-rays on mitosis in the spermatogonial divisions of Locusta migratoria L. Proc. Roy. Soc., B, 119: 61-84. Winge, 0. 1927. Zytologische Untersuchungen ueber die Natur maligner Tumoren. I. "Crown Gall" der Zuckerruebe. Z. Zellforsch., 6: 397-423. Winge, 0. 1930. Zytologische Untersuchungen ueber die Natur maligner Tumoren. II. Teerkarzinome bei Maeusen. Z. Zellforsch., 10: 683-735. Woglom, W. H. 1912. The nature of the immune reaction to transplanted cancer in the rat. Proc. Roy. Soc., B, 85: 197-200. Wollstein, M., and Bartlett, F. H. 1925. Lymphatic leukemia in infancy with the report of a case. Amer. lour. Med. Sci., 169: 819-830. Woods, M. W. 1937. The nucleolus in Tulipa. Amer. lour. Bot., 24: 528--536. Zadek, J. 1933. Die zytodiagnostischen Kennzeichen der Krebszellen. Acta med. Scand., 80: 78-92.