tics in answering it is a most subordinate one. When an author deals in standard deviations and coefficients of variation, the first question the reader should ask is: Has the author determined what part of the diversities thus measured have any bearing on heredity? They may have absolutely none.

Thus the mere fact that observable variations exist between individuals can not properly be appealed to as furnishing material for selection and evolution, as has been so generally done. Most such variations have in the organisms studied absolutely no bearing on the evolutionary process, and there seems little doubt but that this is true for organisms in general.

Illustrations of the practical bearing of these points may be found without departing from the particular organism with which we are dealing. It has been shown that the two products of the division of a single individual Paramecium often differ in size at a given time, so that "variation" occurs in non-sexual as well as in sexual reproduction. But these "variations" are mere temporary fluctuations, without effect in heredity, so that their relation to evolution is nil. Again, Pearl10 showed that conjugants are less variable than non-conjugants. This is true even within the limits of a single race, as I can confirm from extensive studies. But all the variations in such a case, both in the conjugants and non-conjugants, are purely temporary matters, without effect on posterity; so far as evolution or heredity or selection goes they can be left quite out of account.11

Comparative studies have often been made of the variability of higher animals under different methods of reproduction, under different conditions, etc.; the varia

'Simpson, J. Y., "The Relation of Binary Fission to Variation,'' Biometrika, 1, 1902, 400-404. Pearson, K., "Note on Dr. Simpson's Memoir on Paramecium caudatum," Biometrika, 1, 1902, 404-407.

10

Pearl, R., "A Biometrical Study of Conjugation in Paramecium,” Biometrika, 5, 1907, 213-297.

"This of course is no criticism of Pearl's paper, which is one of the foundational ones for this line of work. When different races are present, the less variability of the conjugants is of the greatest significance.

bility of parthenogenetic generations has been compared with that of sexual generations, and the like. These have no meaning for evolutionary questions unless we know whether any of the diversities are heritable. In general, it would appear that most observed variations are not heritable.

Perhaps more important even than the distinction between temporary modifications and really heritable differences is another point regarding "variations." Even leaving aside the temporary modifications, much discussion of variation assumes that the word implies an actual change from one condition to another. This is obviously a very different matter from mere observance of two different conditions in different individuals. If our object is to discover how far we have actually observed evolution taking place, the distinction between variation as an active change and variation as an existing condition (of permanent differentiation between two races) is absolutely fundamental. Evidently, observation of the mere fact that permanent differentiations exist between races is a totally different matter from observation of the changes by which evolution occurs; it is compatible with almost any theory of the origin of diversities; for example, with that of special creation. As a matter of fact, do we find the existing races changing or permanent? What light does our study of variations throw on this?

In Paramecium, in the extensive study of many races for hundreds of generations by exact statistical and experimental methods, not one single instance was observed of variation in the sense of an actual change in a race. In the detailed paper, coefficients of variation are given almost by the hundred, and permanent diversities of race are registered minutely and in numbers. But these mean nothing so far as real change in any race is concerned. So far as the evidence goes, every race was essentially the same throughout the work, and may have been the same for unnumbered ages before.

For clear thinking it is of the greatest importance to distinguish variation as a process from variation as an existing static condition of diversity. If this distinction is not made, we may delude ourselves into thinking we have seen evolution occurring, when all we have seen is the complexity that induces us to invent the theory of evolution. The difference is just the difference between seeing a problem, and seeing its solution; between asking a question and answering it.

But is there indeed no evidence that actual racial changes occur in unicellular forms? On this point we have the extremely important work of Barber.11 Barber was the first to undertake in bacteria and yeasts the study of "pure lines"-of races derived entirely from a single individual. In general his results were the same as those set forth above for Paramecium. Many races of yeasts and bacteria exist, and these races are constant (with the exceptions to be noted). Environmental effects were not inherited, and long continued selection was of no effect in changing such a race. Barber studied also unusual individuals; he found, just as I have set forth above for Parameicum, that their peculiarities were, as a rule, not inherited. But he did find a few cases of peculiar individuals within a pure race, that transmitted their peculiarities to their descendants. Here we have then actual changes in a race; variations in the dynamic sense. In this way there were produced races of yeasts having cells of a different form; races of bacteria composed of longer rods than the parents. But such cases were extremely rare. Variations that perpetuate themselves were found only in one individual among thousands. Barber's work goes as strongly as my own against the significance of the common variations among individuals-such variations as are measured by the coefficient of variation-for heredity or evolution.

To recapitulate, we find that the unicellular organisms are made up of numerous races, differing minutely but

11 Loc. cit.

constantly. The individuals of any race vary much among themselves, but these differences are matters of growth and environment, and are not inherited. What is produced in reproduction depends on the fundamental constitution of the race, not on the peculiarities of the individual parent. The fundamental constitution of the race is resistant to all sorts of influences; it changes only in excessively rare instances, and for unknown causes; in a study of thousands of individuals of Paramecium, through hundreds of generations, hardly a single case of such change was observed.12 Most differences between individuals are purely temporary and without significance in inheritance; the others are permanent diversities between constant races. Systematic and continued selection is without effect in a pure race, and in a mixture of races its effect consists in isolating the existing races, not in producing anything new.

To give in brief an account of the general results of extensive work, it is necessary to make definite statements, and to omit conditions, exceptions and qualifications. This the reader is asked to remember; the details may be found in the original papers. The results are

based on study and measurements of more than 10,000 individuals of Paramecium, kept under experimental conditions, for many generations. But science is essentially incomplete and its results at any time are not final. The author expects to make strenuous attempts to overthrow the generality of some of the results set forth.

12 A single doubtful case is described in the first of the author's two papers: certain individuals of a race acquired a hereditary tendency to remain united after fission, while others did not show this tendency, or showed it less strongly.

THE COLOR SENSE OF THE HONEY-BEE: IS CONSPICUOUSNESS AN ADVANTAGE TO FLOWERS?

JOHN H. LOVELL

IN 1895, Professor Felix Plateau, of the University of Ghent, began the publication of a long series of papers, in which he asserted that Hermann Müller, in formulating his theory of the evolution and use of floral colors, had been misled by a too vivid imagination; and that anthophilous insects are attracted chiefly by odor. In a list of his papers prepared by himself and now before me Plateau states that his latest contribution, entitled Les insects et la couleur des fleurs,1 contains a "summary of the whole." The conclusions of many years of patient research are given at the close of this paper as follows:

"In the relations between the insect fertilizers and entomophilous flowers, the more or less bright coloration of the floral organs has not the preponderating rôle which Sprengel, H. Müller and their numerous adherents have attributed to them. All the flowers in nature might be as green as the leaves without their fertilization being compromised. The sense of smell so well developed among most insects far from being a secondary factor is probably the principal sense which discovers to them the flowers containing pollen and nectar."

By fertilization Plateau doubtless means pollination, for fertilization is an entirely distinct phenomenon, often not occurring until many months after the pollen has been placed upon the stigma. Plateau's conclusions have not met with general acceptance; and in some instances, as he himself naïvely remarks, have been criticized in a "merciless manner."

Florœcology is, however, greatly indebted to him not only for many very interesting observations and experi

1 Plateau, F. Les insects et la couleur des fleurs. L'Année Psychologique, 13, 67-79, 1907.