certain forms as centers of distribution. Twenty F2 selections with respect to the character of the leaf base were made and their progenies were studied in F, and subsequent generations. The twenty selections fell into six general leaf types, the names and descriptions of which follow:

1. Stenophylla, long-petioled like angustifolia, and with a minimum wing development along the petiole.

2. Latifolia, short but distinctly petioled like the F1, and with a variable wing development.

3. Lanceolata, broad-based sessile leaves with long lanceolate blades. 4. Loriifolia, sessile leaves with very narrow strap-like blades.

5. Auriculata, leaves with the blades sharply constricted at the base, nearly if not quite to the mid-rib, and typically with clasping auricles. In some auriculata forms the auricles may be absent, in which case the leaf may appear to have a short distinct petiole.

6. Sessilifolia, sessile leaves with a broad base and a broad clasping insertion.

Of the above six types of leaves, three, lanceolata, loriifolia, and sessilifolia, differ merely in width of blade. Genetically they appear to be identical in their leaf-base factors. For the purposes of this discussion, therefore, it is only necessary to distinguish four primarily leaf-base types, stenophylla, latifolia, sessilifolia, and auriculata.

As we have stated, F2 gave no decisive evidence either as to the number of factors involved in the leaf-base contrast or as to the interactions among them. In order to obtain suitable analytical material we were obliged to study populations of F3 and subsequent generations. In such populations a notable simplification was observed in the number of segregation products, and the distribution into classes was more distinct. In the course of the investigations, also, constant derivatives of the various leaf-base types mentioned above were established, and further information as to the genetic relation of the different types was obtained by intercrossing these lines and by crossing them with the original parents. Theoretically such lines should differ less from the parents and from one another than the original parents did from each other, and practically this was found to be the case. The behavior in skeleton of selections of various leaf-base types is given below.

Stenophylla selections were observed to segregate in a variety of ways. Two populations bred true for stenophylla; three segregated into 3 stenophylla: 1 sessilifolia; and one gave 3 stenophylla: 1 auriculata. A latifolia derivative crossed with angustifolia gave F1 stenophylla, and F2 3 stenophylla: 1 latifolia.

Latifolia selections proved rather perplexing. In some cases latifolia is evidently a complex hybrid expression, as is evidenced by the fact that the F1 cross between the original parents was typically latifolia in appearance. One selection repeated the complex F2 segregation in F3; two bred

1

true for the latifolia leaf-base type; one gave a fairly definite segregation into 3 latifolia: 1 sessilifolia with subsequent establishment of both types in constant derivative lines; and one population gave 3 latifolia: 1 sessilifolia.

Sessilifolia selections behaved very simply. Of nine such selections, five bred true for sessilifolia, and the remaining four gave 3 sessilifolia: 1 auriculata.

Auriculata selections appear to breed true whenever tested. One population is an exception to this statement, but it was unfortunately not further investigated. Two other selections bred true for auriculata. An F light pink auriculata derivative crossed with macrophylla (red sessilifolia) gave pink sessilifolia in F1, and 133 pink sessilifolia: 40 pink auriculata: 53 red sessilifolia: 16 red auriculata in F2, a very close approximation to an independent dihybrid expectation.

On the basis of these facts we have been bold enough to offer provisionally the following formulation for the major differences in leaf-base characters: Ss, stenophylla versus sessilifolia, SS is long-petioled like angustifolia, ss broadly sessile like macrophylla. The heterozygote may possibly approach an intermediate condition similar to latifolia.

Ll, stenophylla versus latifolia. SSLL is long-petioled like angustifolia; and SSI, short-petioled like latifolia, and with a distinct but not broad wing on the petiole. Latifolia is evidently a modified petiolate form.

Aa, sessilifolia versus auriculata, ssAA has the broad clasping leaf-base characteristic of macrophylla; and ssaa the deeply constricted leaf-base with flaring auricles of auriculata. Auriculata is clearly a modified sessile form.

As to the completeness of dominance in any of these factor pairs, we can say very little because of the difficulty of evaluating the effect of the residual genotype. Likewise the significance of the wing in latifolia or of the auricles in auriculata is not definitely established on account of the extreme genetic difference which existed between the original parent varieties.

The Calycina-Virginica Series. In the calycina-virginica (Maryland) series most of the attention was given to flower color and flower form. The flower color of calycina was red like that of macrophylla, and that of virginica was light pink like that of angustifolia. The results were in full agreement with those obtained from that series. The F1 was pink and F2 gave 73 pink: 23 red, with the same variation in the pink class as that noted for the preceding series. Five red F2 segregants gave 125 F3 plants all red-flowering. Of twelve F, families grown from pink segregants, three gave only pink-flowering descendants, and nine exhibited segregation into 3 pink: 1 red, the actual totals being 170 pink: 52 red.

Calycina is characterized by the production of split hose-in-hose flowers. The calyx is strongly developed and petaloid and it is also deciduous in

contrast to the persistent and at times accrescent nature of the calyces of normal forms. This character proved to be a simple recessive. Fi was normal except for slight suggestions of a petaloid condition of the calyx tips on some of the flowers of the F1 plants. F2 gave 61 normal: 35 split hose-in-hose. In F3, seven families from F2 split hose-in-hose segregants gave 175 plants all bearing split hose-in-hose flowers. Of ten families from normal F2 segregants two gave only normal descendants, while eight exhibited segregation into normal and hose-in-hose in the numbers 133 normal: 42 split hose-in-hose. Four F2 families were examined for simultaneous segregation of pink versus red and normal versus split hose-in-hose. The results were 109 pink normal: 36 pink hose-in-hose: 43 red normal: 11 red hose-in-hose, which is in substantial agreement with a 9:33:1 ratio for independent segregation in these two pairs of allelomorphs. The results establish the existence of a single pair of allelomorphs, Cc, for the character contrast normal versus split hose-in-hose (calycine) flowers.

The Alba-Macrophylla Series.-In the alba (White)-macrophylla series our attention was largely confined to flower color, alba being white and macrophylla, as we have seen, red. The F1 in this case was definitely pink, and F2 gave (combining 1911 and 1916 results) 180 pink: 58 red: 83 white. This result is in very satisfactory agreement with a 9 pink: 3 red: 4 white ratio, which may be accounted for by adding to the Rr pair of allelomorphs previously demonstrated a second pair, Ww, for colored versus white flowers. Alba is then RRww, and macrophylla, rrWW; and the F1, RrWw, is pink as in the two previous series. Of five F3 families grown from red F2 segregants, three (rrWW) bred true for red; one (rrWw) gave 35 red: 14 white; and one gave the anomalous segregation product of 2 red: 23 white. Seven families were grown from pink F2 plants. Five (RrWw) segregated for pink, red, and white in the numbers 73 pink: 25 red: 28 white; one (RrWW) gave 22 red: 3 pink; and one (RRWw) gave 33 pink: 9 white. Only one pink genotype, RRWW, was not demonstrated, but this is not inexplicable in view of the fact that only one in nine of the F2 pink-flowering plants should belong to this genotype. Seven F, families were grown from white F2 segregants. Among 191 plants, otherwise white-flowering, there was a single pink-flowering individual. It was doubtless a stray of some sort. The difference between the red flowers of macrophylla and the white ones of alba, therefore, finds a satisfactory formulation as a bigenic contrast.

Discussion of Results.-The results of the flower color studies set forth above appear at first sight to conflict with those which have been published by Allard. He found that carmine, a color cnly slightly different from the red of macrophylla, was a simple dominant to pink; and when he crossed it with white, he obtained F1 lighter than carmine, and an F2 progeny which might have been interpreted as 9 carmine: 3 pink: 4 white. Certain

back crosses which he made and out crosses of extracted whites to pink bear out this analysis. Of his varieties his pink (Maryland Mammoth) appears to be identical with our virginica (Maryland), consequently we are forced to assume that his carmine is genetically distinct from our red. It follows that if we add to the two pairs of factors for flower color which we have demonstrated, a third, Pp, for carmine dominant versus pink recessive, his results are brought into accord with ours. Certain experimental results which we have just secured and which we are reserving for future treatment demonstrate the correctness of this suggestion.

A general result of these investigations has been a demonstration of the complexity of difference from a genetic standpoint between any two of these so-called fundamental varieties of Tabacum. In one sense this result confirms the opinion of Comes and of Anastasia as to the manner of origin of the vast assemblage of Tabacum varieties. It goes further, however, and demonstrates the futility of seeking to determine affinities on the basis of morphological studies unaccompanied by experimental investigations. Thus, for example, we have seen such puzzling segregation products as the auriculata and loriifolia leaf types, the former with its peculiarly constricted leaf base, the latter with a much narrower leaf blade than either parent, coming out of the angustifolia-macrophylla series. Moreover, the demonstration of the existence of genetically distinct redflowering varieties is another evidence of the limitations of purely morphological studies and of the errors in the determination of affinities to which such studies are subject. The genetic studies outlined above indicate merely that Tabacum is a group species like corn, barley, oats, etc., possessing a complex series of allelomorphic contrasts. The so-called fundamental varieties of Tabacum intercross freely and produce fully fertile progenies. They cannot genetically, therefore, be regarded as representing anything but a few very distinct genotypes. A demonstration that a few such varieties may contain within them the possibility by means of recombination of producing a host of secondary varieties does not really demonstrate that they are fundamental. In fact when we consider the fugitive nature of our Tabacum varieties, except in so far as they are kept isolated by natural or artificial means, the conclusion appears inevitable that we must regard all our varieties as fundamentally equivalent from a genetic standpoint. The really significant problem in considering the species is the determination of how these allelomorphic contrasts have come into existence. These investigations throw no light upon that problem.

We have been interested in the methodology of Mendelian analysis in Tabacum. Doubtless one of the reasons why our knowledge of the details of inheritance in Tabacum is so meager is because of the prevailingly quantitative or semi-quantitative nature of the character differences. Even in the case of flower color, the impression given by a series of varie

ties arranged in order of intensity from white to intense dark red is one of graduated differences in amount of a single pigment rather than in qualitatively different pigments. A definite alternative analysis can, however, be made even for these semi-quantitative characters by growing successive generations of segregating progenies until the progenies have been freed of segregation products other than those which it is desired to analyze. Furthermore, the establishment of constant derivative races and the subsequent study of intercrosses among them has been found to result in simplification of the difficulties of analysis. Both of these methods evidently depend upon stabilizing the residual genotype, which is a prime desideratum in the accurate analysis of semi-quantitative characters. The fact that such simplification of segregation can be accomplished and that semiquantitative characters may then be subjected to analysis according to the qualitative mode of procedure argues not only for the adequacy of Mendelian principles in these cases, but for the identity in principle of qualitative and quantitative characters.

The experimental data cited above were obtained from cultures made possible by a portion of the Adams' Fund allotted to the Department of Botany by the Department of Agriculture of the University of California. The detailed account of this series of studies will appear in a forthcoming number of Vol. 5 of the University of California Publications in Botany under the title, Studies of inheritance in Nicotiana Tabacum, I. A report on the results of crossing certain varieties.

1 Comes, O.: "Monographie du genre Nicotiana comprenant le classement botanique des tabacs industriels," Atti. R. Inst. Incoraggiamento, Napoli (Ser. 5), 1, 1899. "Della Razze dei Tabacchi Filogenesi, Qualita ed Uso," Ibid., 57, 1905.

2 Anastasia, G. E.: Le varieta Tipiche della Nicotiana Tabacum L. Scafati, 1906.

3 Setchell, W. A.: "Studies in Nicotiana, I," Univ. Calif. Publ. Botany, 5, 1912 (1–88). Allard, H. A., "Some studies in blossom color inheritance in tobacco, with special reference to N. sylvestris and N. Tabacum," Amer. Naturalist, 53, 1919 (79–84).

THE CORRELATION OF COMPOUND FORMATION, IONIZATION AND SOLUBILITY IN SOLUTIONS. OUTLINE OF A MODIFIED IONIZATION THEORY

BY JAMES Kendall

CHEMISTRY Department, COLUMBIA UNIVERSITY*

Communicated by M. T. Bogert. Read before the Academy, November 17, 1920 Introduction. The anomaly of strong electrolytes has been a notoriously weak point in the ionization hypothesis ever since its inception. In fact, satisfactory agreement with the Ostwald dilution law (which follows from the application of the law of mass action to the ionization equilibrium RX — R+ + X-) has been definitely established only for two restricted types of electrolytes (weak acids and bases) in a single solvent (water) through a limited concentration range (dilutions greater than N/16).