CHROMATIN MASS AND CELL VOLUME IN RELATED SPECIES BY M. NAVASHIN UNIVERSITY OF CALIFORNIA PUBLICATIONS IX AdKIi TLTL'RAL SCIENCES Volume 6, No. 8, pp. 207-230, 3 figures in text Issued April 14, 1931 University of California Press Berkf.ley, California Cambridge University Press London, England CHROMATIN MASS AND CELL VOLUME IN RELATED SPECIES BY M. NAVASHIN INTRODUCTION At the present time the existence of a parallelism between nuclear mass and morphology and external structure of organisms can be demonstrated only for polyploids. There is no such clear relationship, however, among the different species. A tetraploid species is not necessarily larger than a diploid species ; moreover, instances are known where analogous changes in the chromatin mass have exactly opposite results in two different species. Such a lack of straight parallelism between nuclear mass and morphology and external organization may be reasonably explained by secondary influences of certain genes modifying the development in such a way that the original relations become obscured. This may be observed even in some polyploids, for instance, in Oenethera gigas naneJla, which, instead of having the giant stature typical for tetra- ploids, is a dwarf, owing to a specific influence of a particular gene. In order to arrive at a clearer understanding of this problem, one must study, therefore, earlier stages of development, prior to any differentiation or mutual interaction of tissues. Such conditions are realized best in the primary meristem of a plant, where one might expect to observe the original relation between the nucleus and development, unobscured by differentiation. But because of lack of differentiation only a few characters can be studied in meristematic cells. Among these primordial characters the size of the cell may be considered the most important ; for it can be accurately measured and, moreover, it undoubtedly reflects the fundamental property of the cell, the ability of synthesizing a definite amount of protoplasmic substance. 208 University of California Publications in Agricultural Sciences [Vol.6 The cell size was selected, therefore, for the present investigation in the hope that it may give a clue to some understanding of the significance of specific differences in nuclear organization. The present investigation was conducted in the Division of Genetics of the University of California under a fellowship of the International Education Board. The writer acknowledges with gratitude the many helpful suggestions of Professor E. B. Babcock and Professor R. E. Clausen during the course of the work. MATERIAL AND METHOD Primary meristem of root tips was chosen as material for the present investigation. Only primary roots of very young seedlings were used, since it was found that they show less variation in cell size than the adult roots. The seeds of different Crepis species were planted in sand and germinated under conditions of moisture and temperature held as constant as was possible under the circumstances. Immediately after the cotyledons of the seedlings had spread, the root tips were fixed in S. Navashin's fixative (chromic acid, acetic acid, and formalin) ; special care was taken to perform all the following procedures in a uniform manner in order to prevent errors due to unequal shrinkage of different lots of material, etc. The sections were uniformly 10 microns thick ; they were stained in iron haematoxylin and mounted in Canada balsam. Thirteen Crepis species belonging to different sections of the genus were selected for the investigation; in their total chromosome length they ranged from 42 to 112.1 relative units. After several preliminary trials, cells of dermatogen were selected for measurements. They appeared to be more suitable than others since they showed less variation in different species than the cells of the periblem; this was probably due to their superficial position in the root and corresponding possibility to grow more freely as com- pared with cells surrounded, and therefore compressed, from all sides by other cells. It should be further expected that the physiological conditions, such as water and oxygen content, would be more uniform in the cells of dermatogen because of their exposed position. Certain morphological peculiarities, moreover, as well as complete lack of intercellular spaces in the dermatogen. facilitated the obtaining of reliable results of measurements; this will be shown in detail later. 1!>31J Navashin: Chromatin Mass and Cell Volume in Belated Species 211!) C. ptlhlnu ('. dioscoridis C. rubra C. foetida c. tectorum ('. alpinti c. aspera C. bursifolia c. parvifiora c. setosu c. neglecta (\ capillaris C. senecioides Fig. 1. Portions of the dermatogen of the thirteen Crrpis species drawn to the same scale. The typical dimorphism of cells expressed in the occurrence of cells of the two kinds (primary cells, P, and secondary cells, S) may be seen, as well as the differences in cell areas of different species. 210 University of California Publications in Agricultural Sciences [Vol. 6 Finally, it was found, that the sizes of the cells stand in a certain relation to the number of cells present in the whole root, probably owing to differences in mutual pressure and of conditions of nutrition ; this source of error was less pronounced in the cells of the dermatogen. Several tests showed that, owing to the nature of the cell shapes, only direct measurements of the areas on cross-sections could be applied. Measurements of linear, tangential, and radial dimensions, on the contrary, appeared to be impracticable. This can easily be seen from an examination of figure 1, representing the typical cells of the dermatogen of all thirteen species investigated. The following method of measuring cell areas was devised ; adjacent cells along the circumference of the cross-section of the root were drawn to the same scale by means of the camera lucida. The magni- fication applied was equal to 1400 diameters. A sample consisting of ten adjacent cells was used for obtaining the average area of a cell. Each root gave from four to six samples, according to the number of cells present. The drawn samples of ten cells each were then out out with scissors and weighed; a set of weights measured directly in square microns was prepared from the same paper as that which was used for the drawings. This simple, although laborious method of making measurements proved to be far better than any other possible way, including the use of a planimeter. In order to check the accuracy of the measurements, the same sample of cells was drawn and weighed repeatedly ; the error of measurements was found to be negligible for it never exceeded 1 per cent of the measured value. Naturally, before using it, the pack- age of paper was tested as to its uniformity ; it was found that the weight varied from sheet to sheet to a negligible degree. The uni- formity of the paper was tested, nevertheless, repeatedly during the progress of the work. It was of the greatest importance to determine, as a preliminary step, the amount of variation of cell areas along the length of the root in cross-sections belonging to different levels. Several samples from different levels were measured; it was found that the cell area rapidly increased toward the base of the root. This region of growth, however, occupied only the short conical part of the root; but, as soon as the root assumed nearly cylindrical shape, the growth ceased immediately and the cell areas remained con- stant, so that it appeared immaterial from what part of the root the cell samples were taken, provided they did not belong to the conical 1931] Navashin : Chromatin Mass and Cell Volume in Belated Species 211 apex of the root. (The cylindrical part of the root usually begins at the level where only one cell layer of the root cap is present. All measurements were confined to this part of the root.) Table 1, repre- senting some data of measurements of cells belonging to different levels of the cylindrical part of the root clearly demonstrates this. TABLE 1 Average Cell Areas Belonging to Different Levels of the Cylindrical Part of the Root. The Levels are 30 Microns Apart. Data Obtained on Crepis capillaris Levels l 2 3 4 Samples: 1 215 230 215 210 2 220 225 215 220 3 225 220 * 220 235 4 210 230 225 225 5 190 165 170 190 Average 212 214 209 216 It soon was found, however, that simple average areas obtained from samples could not be used for comparative purposes, owing to the fact that the cells of dermatogen were found to be of two different kinds, the larger primary and the smaller secondary cells (see fig. 1). The two kinds of cells greatly differed in size, the average ratio being equal to 1.5 ; the presence of secondary cells thus considerably reduced the average figures obtained simply by dividing the total area of a sample by ten (= the number of cells in a sample) . From the figures given in table 2 it is plainly evident that the areas of the two kinds of cells do not even overlap but form a typical bimodal distribution. This cell dimorphism was not peculiar to the dermatogen; on the contrary, it occurred in any part of the root where regular cyclic rows of cells were present. Obviously, it was an inevitable result of growth and cell multiplication. The cells of the two kinds did not alternate regularly ; conse- quently their relative proportions could vary in different individual samples. Furthermore, different species and occasionally different individuals of the same species had different average proportions of the cells of the two kinds. All these circumstances made it necessary to devise some method of correcting the errors of measurements result- ing from intra- and interspecific variation in the proportions of primary and secondary cells. As has been shown above, the typical 212 University of California Publications in Agricultural Sciences | Vol. 6 TABLE 2 Comparison Between Primary and Secondary Cells. The Figures Represent the Average Areas op Cells of the Two Kinds for each Sample. Data on Crepis pulchra Areas Primary Secondary Samples: 1 286 167 2 250 163 3 243 150 4 220 150 5 275 188 6 267 163 7 292 200 8 307 167 9 300 150 10 250 175 11 267 188 12 242 188 13 283 150 14 300 200 15 295 175 16 270 170 17 229 150 18 320 180 19 225 163 20 225 150 21 260 190 22 280 190 23 270 200 24 290 220 25 283 200 26 250 170 27 292 175 28 275 200 29 270 190 30 267 175 31 250 163 Mean 271 3 ±3.0 181. 3±2 3 1031] Navashin: Chromatin Mass and Cell Volume in Belated Species 213 ratio of the average areas of primary and secondary cells closely approximated 1.5. The necessary correction, however, could not be obtained by simple arithmetical computation since it had been found that the average sizes of the cells changed at a higher rate than one might expect assuming the two kinds of cells to be independent one from another. Table 3 presents the corresponding data. Since the computation of the correlation coefficient did not seem to be profitable for determination of the nature of relation between both variables, owing to the low number of classes, a graphical method was employed and a straight line was fitted by the method of least squares. 1.4 • S .6 .7 .8 .9 1 .0 •*• *y^ t \ ^JC i h t | / c \ t f f i \/ r \ \ ) Fig. 2. Graph representing the method of correcting the errors of measure- ments of cell areas, due to variation in the proportion of primary cells in cell samples. X = proportion of primary cells; ¥ = corresponding relative average area of one cell. Circles represent the observed values, disks, the corrected values. The horizontal line C represents the theoretical correction level, cor- responding to the conditions prior to any cell divisions and, correspondingly, to differentiation of the two kinds of cells. A nearly linear correlation between the proportion of primary cells and the average area of the cells was found ; so that the desired cor- rection could be readily determined. It seemed to be feasible to devise the correction in such a way as to bring the area of the cell up to the value which it would assume in case no secondary cells were to be formed, i.e., if no cell divisions took place. Figure 2 gives a diagram- matic representation of the method employed. 214 University of California Publications in Agricultural Sciences [Vol.6 TABLE 3 Relation Between the Proportion of Primary Cells and the Average Area of a Cell. Data Obtained on 11 Different Species, Represented by 350 Samples of 10 Cells each. The Figures Represent Relative Values Obtained by Taking the Area Corresponding to the Proportion 0.5 = 1.0 Proportion of primary cells 0.5 0.6 0.7 0.8 0.9 1.0 Relative areas: Observed Expected 1.00 1.00 1.07 1 04 1.12 1.08 1.19 1.12 1.20 1.16 1.20 TABLE 4 Correction Coefficients Computed for Various Proportions of Primary Cells Proportions of primary cells 0.5 0.6 0.7 0.8 9 1.0 Coefficients 1.299 1.226 1.160 1.102 1.048 1.000 TABLE 5 Comparison Between Corrected and Uncorrected Average Cell Areas in Different Crepis Species Proportions of primary cells Coefficients Average areas 5 6 7 of variation C pulchra : Corrected 292 225 293 240 .3 Uncorrected 6.7 C. lectorum : Corrected 282 217 289 236 2.4 Uncorrected 8.8 C. neglecta : 194 149 191 156 1.5 4.7 C. capillar is : 240 196 240 207 5.6 C. parviflora : ■ 174 134 180 147 3 4 9.7 Average coefficients of variation: 1.5 7.1 li»31] Navashin: Chromatin Mass and Cell Volume in Belated Species 215 On the basis of the above data, constant correction coefficients were computed, each one corresponding to a certain proportion of primary cells in one given sample ; through simple multiplication by the cor- responding coefficient and dividing by ten the average area of a given sample was transformed into a corrected value for one cell. The values thus obtained represented the maximal cell areas which would arise when all modifying influences, including cell divisions, are eliminated. Table 4 gives these coefficients together with corresponding proportions of primary cells. In order to check the results of the correction applied, uncorrected measurements were compared with the corrected ones in regard to degree of variability. Table 5 gives the corresponding data for several Crepis species. From this table it is evident that the use of this cor- rection greatly reduces the variation in measured areas, thus eliminat- ing the influence of variation in the proportions of primary cells. The whole method of measuring cross-radial areas of cells then assumed the following form. First, samples of ten cells each were drawn, cut, and weighed in terms of square microns; to every figure thus obtained was added a note regarding the corresponding propor- tion of primary cells. Second, the corresponding coefficient was set in the computing machine and all the figures obtained by direct measure- ment were corrected through multiplication by this coefficient and division by ten ; first for samples containing five primary cells, then for those containing six, and so on. The corrected areas obtained were treated as individual observa- tions in calculating the mean values and other constants. The number of samples measured was never less than 25 ; as will be seen later, this number was sufficient to obtain fairly consistent and reliable results. Nevertheless, a much greater number of samples was measured in the majority of cases; altogether over 1100 samples were measured repre- senting more than 11,000 individual cells. After the true corrected values for cell areas were obtained, it remained to find the length of the cells along the axis of the root in order to determine the actual volume. This apparently simple task, however, proved to be an extremely difficult one, owing to the very high and variable rate of cell division and, correspondingly, to an extreme variation in cell size. In view of this difficulty it was decided simply to calculate the volumes through multiplication of the areas by their square root. This method seemed to be justified, especially when relative values were needed. One might suppose, however, that 216 University of California Publications in Agricultural Sciences [Vol.6 tlif primary cells could be proportionally shorter than the secondary ones, the respective volumes being thus about the same. Although, a priori, such an assumption seemed to be highly improbable, yet it was checked by direct comparison of both kinds of cells on longi- tudinal sections. It was found that there is no detectable difference in relative length between primary and secondary cells, the two kinds showing the same range of variation. Inequalities in cell length in different species were always clearly pronounced, provided the species in question differed in cell areas. Although there was considerable difference, the results of measurements could not be evaluated because of extreme variation in proportion of cells of different length. As measures of the chromatin mass, the values given by Mann- Lesley (1925) and expressed in relative chromosome lengths were used, with the exception of C. senecioides which was not included in her work. THE RESULTS OF MEASUREMENTS OF CELL VOLUMES IX THIRTEEN CREPIS SPECIES Applying the method described above the following mean values were obtained for cell areas and calculated volumes in the thirteen species investigated (see table 6). From this table it may be seen that the cell areas and correspondingly the calculated cell volumes differ widely in different species. Among the thirteen species one may dis- tinguish groups characterized by similarity of cell size. Thus C. dios- coridis, C. pulchra, and C. rubra all have very similar large cells; C. foetida, C. bursifolia, ('. capillaris, and C. alpina have medium-sized cells; C. parviflora, ('. neglecta, and C. setasa form a group with small cells; C. senecioides stands apart with its very small cells. RELATION BETWEEN CHROMATIN MASS AND CELL VOLUME The length of somatic chromosomes as they appear in rn.etaph.ase, wax used as a measure of the chromatin mass. It is obvious that the length of the chromosomes may be applied as a measure of chromatin volume only when the width of the chromosomes is constant. As a matter of fact, however, this is not the case; the different species have in many instances chromosomes of very different widths. The measure- ment of chromosome width appears now to be extremely difficult owing to its exceedingly low value and especially to variation resulting from 1031] Navashin: Chromatin Mass and Cell Volume in Related Species 217 the degree of splitting of each chromosome, the latter being always composed of two daughter strands, It is sufficient at present to under- stand that chromosome length alone cannot be applied without clue reservations. TABLE 6 Cell Areas, Calculated Cell Volumes, and Eelative Total Length of Chromosomes in Thirteen Species Species C. senecioides C. parviflora.. C. neglecta C. setosa C. as-pera C. foetida C. bursifolia, C. capillar is... C. alpina C. tectorum... C. dioscoridis C. pulchru C. rubra Cell area in square microns 156.8 180.3 192.0 201.0 234.6 248.3 249.0 250.9 256.7 275.5 290.2 296.3 296.8 2.8 2.2 2.9 3 2.6 2.7 3.6 5.3 7.1 2.2 4.0 2.4 3.9 Calculated volume in cubic microns 1963 2421 2660 2850 3693 3913 3929 3974 4113 4573 4944 5100 5113 35 30 40 43 38 43 57 84 ±114 ± 37 ± 68 ± 41 ± 67 Corrected average of total length of chromosomes* 42.0 69.9 61. 7t 63.2 82.6 93.7 78.5 61. 4t 87.3 88.7 109.4 112.1 102 9 * From Mann (1925) table 3, p. 302 (except senecioides). t These species have thicker chromosomes, cf. p. 223. TABLE 7 Cell Areas and Calculated Volumes of the Members of a Polyploid Series in C. capillar is Chromatin mass Cell areas Cell volumes Haploid (one plant, several roots) 147.7 250.9 329.0 425.4 1789 ± 22 Diploid (several plants, one root of each) Triploid (several plants, one root of each) Tetraploid (three plants, one root of each) 3960 ± 84 5968 ± 82 8976 ± 100 Prior to comparing different species in the matter of the relation between their chromatin masses and the respective cell volumes, a study of intraspecific variability of chromatin masses was made with the expectation that these conditions would be clearer, i.e., unobscured by eventual specific developmental peculiarities or errors due to inac- curate chromosome measurements. For this purpose a polyploid series in C. capillaris was used, consisting of haploid, diploid, triploid, and tetraploid individuals. The figures obtained from measurements of cells of the representatives of this polyploid series are given in table 7. 218 University of California Publications in Agricultural Sciences [Vol. 6 Essentially the same relations between chromatin mass and cell volume were found in the polyploid series of the two other species: C. tectorum (2n, 3n, and 4n were investigated) and C. parviflora (2n and 3n). The calculated cell volume here was also found to be prac- tically proportional to the chromatin mass. 70 no / 7— / / / '/ rubra | * , \ pulchra ecto i rum //W 1 t y / / /) ' oapillaris JjVipi.. | bureifolla | / ^ | foetida i LBpej *a V Aty aetoaa / A - nej / u / / parvif 1 ora ^ / A senecioides / Fig. 3. Graphic representation of the relation between chromatin mass and cell volume in thirteen Crepis species. The length of the bars, indicating different species, corresponds to the respective probable error. The cell volume is expressed in cubic microns; the chromatin mass, in relative units of chromosome length. As was stated above, these data on the polyploid series were secured chiefly in order to check the method employed; the correctness of which, as may be seen, they have established. As for the accurate determination of the degree of correlation between chromatin mass and cell volume in the polyploid series, the actual computing of the correlation coefficient did not seem to be practicable on account of the few classes of variables. Different species could now be compared safely in the matter of the relation between their chromatin masses to the respective cell volumes. 1931] Navashin: Chromatin Mass and Cell Volume in Belated Species 219 The usual method of graphic representation was here employed. Figure 3 gives the corresponding correlation diagram for all thirteen species investigated. This diagram clearly shows that there is a very' strong positive correlation between the chromatin mass and the cell volume, save for a few exceptions. These exceptional species deviate in the graph from the exptcted positions to a significant degree ; but, as will be shown in the discussion, this circumstance has its natural explanation. Computation of the coefficient of correlation gave the value, r=.90 ± .04. Computation of the correlation ratio did not seem to be practicable owing to the low number of classes ; it appears from the graph shown in figure 3, that the correlation may be regarded as linear. DISCUSSION AND CONCLUSIONS That there is direct relation between the sizes of cells and the nuclei they contain, has been well known since the time of Strasburger (1893). Later, on the basis of numerous observations, R. Hertwig (1903, 1908) developed the idea of the "nucleoplasmic ratio." This theory contends that the quotient, nucleus cytoplasm, is a constant of fundamental importance for the life activity of the cell. Further investigations have shown, however, that the nucleoplasmic ratio varies very much in different tissues (Kleinberger, 1917) ; it varies under external influences as well, especially those of temperature (R. Hertwig, 1903; 0. Hartmann, 1919). The most striking instances of change of the nucleoplasmic ratio is furnished by sexual cells, the male cells containing practically no cytoplasm, the female cells developing a vast amount of cytoplasmic material. From these and many similar data it is evident that the nucleoplasmic ratio could hold true only for cells prior to any differentiation, specialization, or modifying influences of the surrounding medium. All the above mentioned data were obtained from studies of resting nuclei, their sizes being regarded as a measure of the amount of nuclear material. It may be observed, however, that the resting nuclei may greatly differ in the quantity and quality of materials they con- tain, in connection with innumerable changes they undergo in different stages of the cell life, cell division, specialization, and differentiation of cells, etc. Thus the data, obtained from observation on "resting" nuclei may be regarded as scarcely reliable ; at least they should not be accepted without due reservations. 220 University of California Publications in Agricultural Sciences [Vol.6 The comparison of nuclei on the basis of their actual chromatin content, which can be measured in a definite stage of development, gives much more conclusive results. The actual counts and especially the measurement of chromosomes at the metaphase give directly the relative amount of chromatin, as it emerges at this critical moment of karyokinesis. The first to demonstrate the effect of change of chromatin amount was Grerassimoff (1902), who experimentally produced a diploid game- tophyte of Spirogyra. Gerassimoff found that the diploid cells greatly exceed in size the normal haploid cells. Thus it was demonstrated that an increase in the actual amount of nuclear substance stands in causal relation to the cell size and, consequently, to the general amount of protoplasm. Gerassimoff 's results were immediately confirmed by Boveri (1905) on animal material. From his famous experiments on Echinoderwata eggs this investigator concluded that even the addition of a single supernumerary chromosome causes an enlargement of the cell. From his measurements he found that not the volume but the surface of the cell is proportional to the chromatin mass it contains. Following Gerassimoff and Boveri many investigators became inter- ested in this problem. Tischler on Musa (1910), the Marehals on mosses, (1906, 1907, 1909, 1911), Winkler on SoJanum (1916), and especially Gates on Oenothera (1909) have found that an increase in chromatin material causes also an increase in the masses of nuclei and of the cytoplasm. The figures obtained from measurements by some of these investigators showed, however, that there was no straight pro- portionality between chromatin mass and cell size. The cell size was, furthermore, strikingly variable in different tissues of the same organism ; this is, of course, to be expected as a natural result of differentiation. It seems strange, therefore, that none of these workers measured the cells of the primary meristem of the root. Xemec (1910), who was the first to find tetraploid cells in normal tissues. worked with root tips; he did not make any measurements. To cite more recent data, Taylor's work on Acer (1920). Blakes- lee's on Datura (1922), and especially F. Wettstein's investigations on experimental polyploidy in mosses (1924) should be mentioned. Most recently Clausen and Goodspeed (192")) have shown that even under hybrid conditions the cell size tends to correspond to the chromatin mass. 1931] Navashin: Chromatin Mass and Cell Volume in Belated Species 221 All these and many other similar data proved beyond question that there is a definite relation between chromatin ma.ss and cell size. In the great majority of these cases the differences in chromatin mass were due to polyploidy, i.e., to the addition of entire chromosome sets. As a matter of fact, however, natural species show differences in chromatin mass owing' to variation in chromosome size, the chromo- some number being often the same. At present, data concerning the relation between cell size and chromatin mass as determined by the sizes of the chromosomes, are entirely lacking, save for a few observa- tions which contradict one another. Such specific differences a.s inequalities in chromosome size seem to be more important than varia- tions resulting from polyploidy, since the latter could produce only secondary new forms from pre-existing species which had arisen by different processes. In other words, a new species could not arise through polyploidy, unless sufficient specific changes in the organiza- tion of the chromosomes themselves should take place. The visible results of these phylogenetic changes probably represent alterations in both the relative and absolute sizes of chromosomes (M. Navashin. 192."), 1926; L. Delaunay. 1926). Tischler was probably the first worker who became interested in this problem. Comparing a giant strain of Phragmites ("var. Pseudo- donax") with a normal one (1918), he found it to possess much larger meiotic chromosomes, but the size of pollen mother cells was unaffected by this difference in chromatin content. Delaunay reached essentially the same conclusion, while measuring meristematic cells (dermatogen of the root tips) in two Muscari species (1926). This investigator deduced from his measurements that the cell volumes in the two species investigated were identical, in spite of significant differences in chromosome sizes. He found the number of meristematic cells, not their sizes, to be connected with the chromatin mass. Most recently Jaretzky (1928), studying various Polygonaceae, came to an opposite conclusion ; he has found that there is direct con- nection between chromosome size and cell size. But he offers no measurements in proof of his statement. It does not seem difficult to explain the contradictory results obtained by the above mentioned investigators. First, it may be noted that Tischler might have dealt merely with that type of size difference which was not due to actual variation in chromatin mass, but was 222 University of California Publications in Agricultural Sciences [Vol. 6 rather connected with physiological conditions, such as those known to occur in different tissues of the same organism. The figures pre- sented by Tischler do not give a clear idea as to the nature of chromo- some differences in the two Phrag mites strains studied by him. In Delaunay's case there is not the slightest doubt that the two species investigated (Museari longipes and M. teniiiflorum) actually differ in chromatin amount, the respective total chromosome length being in the ratio 1.35 : 1.00. From the data of this writer, one may easily see, however, that his cell measurements could have been sub- ject to errors, owing to the same general causes as those described in detail in the present paper. Thus Delaunay merely applied average values of the linear dimensions of cells; he did not. select cells, but probably used random samples of cells (90 cells of each species) for determination of the radial and the tangential linear dimensions. The axial length of cells was determined simply by dividing the total length of the meristematic zone by the number of cells present in a longitudinal section. Referring now to what was found by the present writer, it may easily be seen that, if Delaunay's two species differed in the proportion of primary cells, his conclusions regarding the radial and tangential dimensions could have no significance; not to mention the impossibility of obtaining a reliable linear measure of such a com- plex shape. To illustrate this, suppose that M. longipes has 0.5 of primary cells and M. tenuiflorum, 0.7. If the relations between the two kinds of cells are similar to those found in Crepis (which is highly probable, since in several of the Liliaceae the present writer found exactly the same conditions as in Crepis), it would, of course, mean that the average areas of cells obtained from random sampling might happen to be equal, the real mean values being in the ratio 1.12 : 1.00. which would give the ratio for volumes approximating 1.2:1.0; and this latter ratio is not so far from the ratio of chromatin masses in the two species. Some details mentioned by Delaunay directly suggest the probability of the existence of a difference in the proportions of primary cells in the two species, although he does not even mention the occurrence of cells of the two kinds. Thus, Delaunay has found that M. longipes has narrower but longer cells as contrasted with M. toiiiiflorniii. It seems to be very probable, therefore, that the former species has actually fewer primary cells, since the width of cells is above all affected by the proportion of additional (secondary) cells. As for the axial dimensions of cells, there is no simple way to eliminate the extremely great error arising from variation in the rate 1931] Navashin : Chromatin Mass and Cell Volume in Belated Species 223 of cell division. In Delaunay's case it would be expected that M. longipes, having more cells would be apt to have shorter cells, owing to the fact that the larger number of cells is merely the result of a higher rate of cell divisions. Consequently Delaunay's divergent con- clusions can by no means be regarded as established. With regard to his conclusion concerning the connection between chromosome length and the number of meristematic cells, the present writer can not confirm them on his material. Although in some cases a parallelism between chromosome length and the number of meristematic cells could be observed, in other cases both characteristics seemed to vary independently. For example, C. pulchra has a very short meriste- matic zone and very long chromosomes, while C. tectorum possesses much shorter chromosomes but about twice as many meristematic cells. The data of the present investigation, however, indicate that there is a strict proportionality between chromosome length and cell volume in related species ; this proportionality is not less in this instance than in the polyploid series. Some deviations from the expected relations, however, seem to contradict the generality of this conclusion. Thus C. capillaris appears to have larger cells than might be expected from the choromosome measurements ; however, C. neglecta with chromo- somes almost as long as those of C. capillaris, possesses much smaller cells. But this apparent disagreement has its simple explanation. As has been stated, in comparing the chromatin masses in different species only the total length of the chromosomes was considered, with- out reference to chromosome width. It is therefore apparent that in case the chromosomes differ in width, the length no longer can be used as a relative measure of volume. This is precisely the situation in C. capillaris and C. neglecta, as a glance at the chromosomes of the two species is sufficient to show, the former having considerably thicker chromosomes than the latter (cf. Hollingshead and Babcock, 1930, fig. 11, p. 19). The same difference exists also in several other species. These differences in relative chromosome length and width raise a very important question as to the relation between the general chromo- some shape and the actual amount of the chromatin content. As may easily be seen, the shape of the colloidal body of a chromosome is deter- mined to a large extent by the properties of the surrounding medium, especially by its acidity, osmotic pressure, etc. Very considerable changes in the shape may, therefore, occur when the conditions of the cytoplasm become changed. This was actually demonstrated by 224 University of California Publications in Agricultural Sciences [Vol. 6 Kuwada and Sakamura (1926), who succeeded in changing the size of the chromosomes by varying the hydrogen ion concentration. It is natural to suppose, however, that different species may differ in the acidity of their protoplasms, so that in some of them the chromosomes may be more contracted (owing to higher acidity), others, on the contrary, may be more elongated (owing to lower acidity). These alterations may go even farther than simple changes of shape due to variation in surface tention ; for the amount of water contained in the chromosome body may vary considerably, thus changing the actual mass of the chromosome itself. That the cause of variation in relative width and length of the chromosomes is not inherent in the chromo- somes, may be deduced from the fact that chromosomes belonging to the same plate never show differences in widths. This very plausible conclusion leads Jaretzky (he. cit.) to explain certain variations in chromosome shape through differences in the acidity of the cytoplasm. It is now perfectly clear that a comparison of chromosome masses of two different species cannot safely be made unless the chromosomes are observed under precisely identical conditions of surrounding medium. Such conditions can exist only in one case : viz., when the chromosomes of the two species are located in the same cell. There is an easy way to accomplish this — by bringing the specific chromosomes together by means of hybridization. The present writer has made a large number of comparative observations on F! interspecific hybrids (1929), and it was found that the parental chromosomes never differ in width in hybrids, even though they previously displayed marked differences in the pure species. It could further be observed that this equality in width was accompanied by corresponding changes in chromosome length ; thus, for instance, in a hybrid, C. capillaris x C. neglecta, the neglecta chromosomes appeared as thick as those of capil- laris but this widening of the chromosomes was accompanied by con- siderable shortening. Table 8 gives the results of chromosome measure- ments of some pure Crepis species and of hybrids between them. TABLE 8 The Total Length of Haploid Chromosome Complexes Measured in Cells of Puke Species as Compared with that Obtained from Measurements in Cells of Corresponding Interspecific Hybrids. Root Tips Total haploid chromosome length of C. capillaris C. tectorum C. neglecta From pure species 107.94 ± .40 155.55 ± .88 97.02 ± .51 From capillaris x neglecta hybrid 125.85 =fc .69 83.33 ± .37 From capillaris x tectorum hybrid 109.43 ± .47 147.57 ± .50 1931] Navashin: Chromatin Mass and Cell Volume in Belated Species 225 It appears, therefore, that the length-width ratio of a chromosome is merely a matter of the ehemieo-physieal condition of the cytoplasm; and the situation would not change at all, if one assumes that the properties of the chromosome itself are responsible (to a certain extent at least) for these conditions. Further discussion of this problem will appear elsewhere. It is clear that comparative measurements of chromosome volume are relatively easy when hybrids are used. It may give reliable results also as far as comparison of pairs of species is concerned. But the conclusions may not be applied without reservations to a larger num- ber of species, since it is obvious, that the conditions in a hybrid cell are determined by the cooperation of the two specific chromosome sets. And, if in certain cases actual change of the absolute chromosome mass should take place, the same species might appear to have different chromosome volumes, depending upon the other species with which it had been crossed. Nevertheless, judging from the available data, it does not seem to be probable that the chromosome changes often go beyond mere shortening and proportional thickening; the total volume being unaffected. Through comparative study of hybrid chromosomes, it was found that C. capillaris has actually much more bulky chromosomes than C. parvifiora, C. nefflecta, and even C. setosa. Therefore the respective positions of these species in the diagram of figure .'1 should be cor- respondingly changed, thus bringing them much closer to the actual regression line. As has been shown above, the connection between chromatin mass and cell volume is not less pronounced in interspecific variation, than in intraspecific changes due to polyploidy. This justifies the assump- tion, with a fair degree of probability, that the same conditions which make the polyploid cells larger, are acting in the case of specific altera- tion of chromatin mass. Although there is no doubt that members of polyploid series actually differ in the amount of genes present, it might be premature at present to conclude that just this difference in the amount of homologous genes is alone responsible for the variation in the production of protoplasmic material ; yet such a conclusion is certainly indicated. In regard to specific differences in chromatin amount, such a conclusion must be made with still more reservations, in spite of its plausibility. From his studies on Polygonaccae Jaretzky (he. cit.) has made much more categorical assertions, namely, that the species having 226 University of California Publications in Agricultural Sciences [Vol. 6 larger chromosomes must be regarded as polyploids, the large chro- mosomes being in reality compound structures, each composed of two or even more small chromosomes. This idea is not new, nor is it war- ranted by a sufficient number of facts; furthermore, this author reverts to the old idea of loss mutations in order to explain the differences between various species. It is, however, perfectly obvious that regarding loss mutations as a cause of evolution, one does not avoid the absurd consequences of such an idea by appealing to poly- ploidy, irrespective of its kind. It may be of interest to discuss briefly another problem arising from the results of the present investigation. From the accurate deter- mination of relative proportions of primary and secondary cells in the dermatogen of the root, it was found that these proportions are remarkably constant for a given root ; it appeared further to be very constant for a given species. Table 9 gives some corresponding data. TABLE 9 Proportions of Primary Cells in the Dermatogen in Different Crepis Species Species Proportion of primary cells C. parviflora .563 ± .012 C. foetida.. .568 ± .010 C. senecioides .567 ± .014 C. bursifolia .570 ± Oil C. tectorum .575 ± .010 C. pulchra .578 ± .008 C. aspera . . .572 ± .009 ( '. capillar is mean value .571 ± .001 .665 db .014 Besides the proportions given in the preceding tabic, some others were determined, for example, the occurrence of proportions 0.62, 0.71, and 0.75 (possibly also 0.60) was established, also that in two species the proportion of primary cells sometimes dropped below 0.5; and the most common proportion closely approximated 0.571. Differ- ent roots of the same species eventually exhibited different propor- tions; but never did a proportion occur which was intermediate between those enumerated above. The variation in the proportions of primary cells thus proved to be discontinuous. It is easy to observe, that, since the proportion of primary cells is merely a direct result of the proportion of preceding cell divisions, the 1931] Navashin : Chromatin Mass and Cell Volume in Eclated Species 227 latter must be strictly limited by a definite number, typically constant for each case. The relations between the proportion of dividing cells and the proportion of primary cells in the resulting tissue are shown in table 10. TABLE 10 Theoretical Relation Between - the Number of Cell Divisions and the Resulting Proportions op Primary Cells Relative numbers of dividing cells Relative numbers of resulting cells of the two kinds Proportion of primary cells No divisions 1 000 1 out of 10 10 prim. + 1 secon. 5 prim. + 1 secon. 4 prim. + 1 secon. 3 prim. + 1 secon. 5 prim. + 2 secon. 2 prim. + 1 secon. 5 prim. + 3 secon. 3 prim. + 2 secon. 4 prim. + 3 secon. 909 1 out of 5 0.833 1 out of 4 soo 1 out of 3 . 750 2 out of 5 714 1 out of 2 667 3 out of 5 2 out of 3 3 out of 4 All dividing 625 0.600 ■0.571 500 The whole process of differentiation of the two kinds of cells (see fig. 1) may be explained in the following way: after the last cell layer of the meristem is formed through divisions in the radial direc- tion, a series of cell divisions in a perpendicular (tangential) direction takes place ; each dividing cell forms two cells, one of which is identical with the original (primary) cell, the other being different in shape and size owing to its position opposite the cell wall of the underlying cell of the row underneath. Thus two kinds of cells arise, the primary. equal in number to the cells of the layer below and characterized by pointed inner ends, which occupy the spaces between the cells of that layer; and the secondary, which are located between the primary cells and opposite the cells of the layer below. The above table gives only a few of the most simple numerical relations which would arise as the result of various proportions of dividing cells; an indefinitely large number of different combinations might be imagined. As a matter of fact, however, only very few of the possible combinations occur in reality ; namely, only the following relative numbers of dividing cells occur: 1 out of 3, 2 out of 5, 1 out of 2. 3 out of 5, 2 out of 3, and 3 out of 4, the latter proportion being the most common (it results in the proportion of primary cells being 0.571). 228 University of California Publications in Agricultural Sciences [Vol. 6 It seems evident, therefore, that the rate of cell division follows an extremely simple rule : only the simplest numerical ratios between dividing and non-dividing cells are realized. Further discussion of this matter not being pertinent to this paper, it is merely stated that the described relations are undoubtedly consequences of some most fundamental physical property of the cell. The study of these rela- tions might possibly lead to some understanding of the most direct causes of cell division and cell differentiation. Further investigation is in progress. SUMMARY Many closely related species are known to differ greatly in their chromatin content. There is no strict parallelism between the specific chromatin amount and external morphology. This phenomenon remains as yet unexplained. In view of the fact that in adult organisms the original relation between chromatin mass and the mode of development could be obscured by some influences of certain genes, it was decided to com- pare the meristematic cells of different species. The size of the cells of the dermatogen of the root was chosen as material for the investi- gation, on the supposition that the amount of protoplasmic material produced reflects the most fundamental property of the cell, viz.. its ability to synthesize a definite amount of organic matter. A special method of measuring cell volumes was devised ; and all sources of error were eliminated, as far as possible. It was found for the thirteen species of Crepis investigated, that the volume of the cell was proportional to the amount of its chromatin. Thus it was established that the specific alterations in chromatin amount have the same primary consequences as the intraspecific varia- tion in the number of the homologous chromosome sets arising through polyploidy. This close analogy suggests the possibility that the specific differences in chromatin amount are connected with the amount of homologous genes, as in the polyploid series. Some data of the present investigation clearly show that the pro- portions of cell divisions in primary meristem are strictly confined to a series of simple fractions; viz., J ;i , -,-,, ' ^, %, -.., and %. This indi- cates that the cell divisions and, consequently, the resulting simplest cell patterns are controlled by some elementary physical process. 1931] Navashin: Chromatin Mass and Cell Volume in Belated Species 229 LITERATURE CITED Blakeslee, A. F. 1922. Variations in Datura due to changes in chromosome number. American Nat., 56:16-31. Boveri, Th. 1905. Zellenstudien V. Ueber die Abhangigkeit der Kerngrosse und Zellen- zahl der Seeigel-Larven von der Chromosomenzahl der Ausgangs- zellen. Jenaische Zeitschr. f. Naturwiss., 39:445-524. Clausen, B. E., and Goodspeed, T. H. 1925. Interspecific hybridization in Nicotiana. II. A tetraploid glutinosa tabacuiit hybrid, an experimental verification of Winge 's hypo- thesis. Genetics, 10:278-284. Delaunay, L. 1926. Phylogenetische Chromosomenverkiirzung. Zeitschr. Zellforsch. Mikr. Anat., 4:338-364. Gates, B. B. 1909. The stature and chromosomes of Oenothera ffi/ias de Vries. Arch. f. Zellforsch., 3:525-552. Gekassimoff, J. .J. 1902. Die Abhangigkeit der Grosse der Zelle von der Menge ihrer Kern masse. Zeitschr. allgem. Physiol., 1:220-258. Hartmann', O. 1919. Ueber den Einfluss der Temperature auf Plasma, Kern und Nucleolus und eytologisehe Gleichgewichtszustande (Zellphysiologische Ex- perimente an Pflanzen). Arch. f. Zellforsch., 15:177-248. Hertwig, B. 1903. Ueber Correlation von Zell und Kerngrosse und ihre Bedeutung fur die geschleehtliche Differenzierung und die Teilung der Zelle. Biol. Zentralbl., 23:49-62; 108-119. 1909. Ueber neue Probleme der Zellenlehre. Arch. f. Zellforsch., 1:1-32. Hollingshead, L., and Babcock, E. B. 1930. Chromosomes and phylogeny in Grepis. Univ. Calif. Publ. Agr. Sei., 6:1-53. .Taretzky, B. 1928. Histologische und karyologische Studien an Polygonaceae. Jahrb. wiss. Bot., 69:357-490. I.SH1KA.WA, M. 1916. A list of the number of chromosomes. Bot. Mag. Tokyo, 30:404-448. Kleixberger, E. 1917. Ueber die Grosse und Besehaffenheit der Zellkerne mit besonderer Beriieksichtigung der System atik. Bot. Centralbl. (Beihefte), 35:219-278. 230 University of California Publications in Agricultural Sciences [Vol. 6 Kuwada, Y., and Sakamuba, M. 1926. A contribution to the colloid-chemical and morphological study of chromosomes. Protoplasma, 1:239-254. Mann, Margaret C. (Mann-Lesley). 1925. Chromosome number and individuality in the Genus Crepis. A com- parative study of the chromosome number and dimensions in nine- teen species. Univ. Calif. Publ. Agr. Sci., 2:297-313. Marohal, E. L., and Marchal, Em. 1906. Recherches experimentales sur la sexualite des spores chez les mousses dioiques. Mem. cour. CI. d. Sci. Acad. Royale Belgique, ser. 2. 1(5): 1-50. 1907. Aposporie et sexualite chez les mousses. Bull. Acad. Royale Belgique (CI. d. Sci.), 1907:728-730. 1909. Aposporie et sexualite chez les mousses II. J hid.. 1909:1249-1288. 1911. Aposporie et sexualite chez les mousses III. Ibid., 1911:750—776. X avashix, M. 1925. Morphologische Kernstudien der Crepis-Arien in Bezug auf die Ait bildung. Zeitschr. Zellforsch. mikr. Anat., 2:98-110. 1926. Variabilitat des Zellkerns bei CVepis-Arten in Bezug auf die Art- bildung. Ibid.. 4:171-215. 1929. Unpublished data. Nemec, B. 1910. Das Problem der Befruchtungsvorgiinge und andere Zytologische Fragen. Berlin. stk assburger, E. 1893. Ueber die Wirkung'sphare der Kerne und die Zellgrosse. Histologische Beitrage (Jena), Heft 5:95-124. Taylor. W. R. 1920. A morphological and cytological study of reproduction in the Genua Acer. Contrib. Bot. Lab. Univ. Penn., 5(2) :111-138. Tischler, 6. 1918. Untersuchungen iiber den Riesenwuchs von Phragmites et mi munis var. Pseudodonax. Ber. Deutseh. Bot. Ges., 36:549-558. Wettstein', F. von 1924. Morphologic und Physiologic des Formwechsels der Moose auf gene- tischer Grundlage. Zeitschr. Indukt. Abst. Vererb., 33:1-236. Winkler, H. 1916. Ueber die experimentelle Erzeugung von Pflanzen mit abweiihenden Cliromosomen/.ahleu. Zeitschr. f. Bot., 8:417-531.