# Popular Science Monthly/Volume 78/June 1911/The Measurement of Natural Selection

THE

POPULAR SCIENCE

MONTHLY

JUNE, 1911

 THE MEASUREMENT OF NATURAL SELECTION
By Dr. J. ARTHUR HARRIS

STATION FOR EXPERIMENTAL EVOLUTION, COLD SPRING HARBOR, N. Y.

I. The Status of Darwinism

BY organic evolution the broad-minded biologist of to-day understands merely the natural as opposed to the supernatural processes by which the hundreds of thousands of kinds of organisms which now inhabit or have inhabited the surface of the earth have come to possess the morphological and physiological peculiarities which distinguish them from each other. He believes that these differentiable types have been derived by a natural and relatively gradual process from earlier and, in the main, simpler forms.

This belief he shares with all his associates; the evidence in favor of it is considered by scientific men to be so strong that it has become scientific faith and scientific dogma. To-day, only the nature of the processes by which this evolution has proceeded interests biologists. Darwin's theory, and in large measure Darwin's evidence, have accomplished this. At one time Darwinism and organic evolution were synonyms, but now the suffixes "ian" and "ism" and the prefixes "neo" and "ultra" and "post" are combined with half a dozen different names and discussed with a glibness which is bewildering to some of those who are more interested in measuring the intensity of the factors which may have been active in organic evolution than in formulating theories concerning it.

Darwin's theory, viewed from such a distance that trivial details blend into large outlines, involves three propositions:

First, that variations from the typical condition of an existing species do occur.

Second, that these deviations may be inherited.

Third, that in the competition for existence which must result from the inadequacy of space and food for all, certain individuals are by virtue of their characteristics better fitted to survive under their existing conditions of life than others.

For Darwin these were merely hypotheses to be conscientiously tested against all known facts. With the facts which Darwin painstakingly collected and sifted they seemed to agree so well that naturalists accepted his theory as the best explanation of the diversity of the organic world. Fifty years has furnished, it seems to me, a highly satisfactory substantiation of the first two propositions. We have forgotten that it was once necessary to convince biologists that variations do occur, and are now trying to measure the frequency and amount of variations, to determine what their proximate causes are, and to classify them. Variation and heredity are so intimately linked together that one can not be extensively investigated without considering the other. Since the pioneer work of Galton a few men have been actively engaged in the measurement of the intensity of heredity, and of recent years many more have been occupied with the experimental study of the physiological phases commonly known as genetics.

In consequence of this activity our knowledge of variation and inheritance is much more extensive than was that of Darwin—it would be to our shame if this were not true—but if biologists could all escape for a moment from the limitations of vision imposed by the tangle of post-Darwinian detail and by assumption and subsidiary qualification, and could look at the problems and the data of organic evolution as a whole and in the large I think they would be almost unanimous in regarding these two first propositions as so well established that they present no difficulty to the acceptance of the Darwinian theory.

The insecurity of the Darwinian tripod is to be seen in the weakness of the proposition that natural selection moulds the species by eliminating variations not adapted to the environment. While the first two hypotheses have been replaced by the masonry of quantitative science, the third remains largely a hypothesis, weakly reinforced by analogy and by the indirect evidence of adaptation.

To make more widely known the fact that natural selection is capable of quantitative treatment, of direct measurement, just as are variation and inheritance, is the purpose of this essay. It is not a brief for Darwinism, but a plea for direct quantitative researches into one of the more neglected problems of organic evolution.

II. The Problems of Selection

By the word selection in its most general evolutionary sense we mean merely that those individuals which leave offspring are not on the average representative of their generation, but that they differ in some regards from those which do not survive to be parents. In statistical terminology they are not a random sample of the population.

For the evolutionist a question of fundamental importance is this: Do the offspring of selected individuals differ from unselected individuals of the same population? This is, however, really a problem of variation and heredity and not of selection at all. Given variations which are heritable, stringent selection will change the type of the population. If after this change of type no more heritable variations occur, selection can effect no further change.

The history of cultivated varieties shows us that much can be accomplished by selection, but neither the history of animals and plants under domestication nor any amount of experimental evidence would be sufficient to demonstrate the correctness of the third Darwinian proposition.

This has often been recognized. "The real difficulty of Darwin's theory is the transition from artificial to natural selection," said Paul Janet. Darwin himself frankly tells us, "I soon perceived that selection was the keystone of man's success in making useful races of animals and plants. But how selection could be applied to organisms living in a state of nature remained for some time a mystery to me."

Experimental breeding and statistical studies in variation and heredity can teach us much, but evolution, for the most part, has occurred outside the breeding pen. Some species have originated in the greenhouse and some in a hanging drop culture, but most species have come into existence and biological dynasties risen to dominance and sunk into decadence in the fields and swamps and mountains where organisms live in competition and cooperation, as host and parasite, as destroyer and destroyed. From the standpoint of evolution the vital question concerning selection is: Does selection (natural, sexual or genetic) occur in nature?

Is a selective death rate such an important factor that equipped with proper instruments the biologist can go out into free nature and measure its intensity? If he can, then the Darwinian theory of evolution must detain us longer; if he can not, we must lay Darwinism on one side, and maintain towards it, as towards all other theories for which critical evidence is wanting, an attitude of agnosticism.

III. The Measurement of Selective Elimination'

The hypothesis of the existence of the evolutionary factor known as natural selection is dependent upon the assumption that individuals vary in their capacity to withstand the pressure of their environment, and that the differences in resistance to untoward external conditions are associated with and due to differences in the physical, physiological or psychical characteristics of the organism.

It does not assume that every death is selective. Many are due to factors which eliminate irrespective of any particular character; many survivals are due to a fortuitous combination of favorable environmental conditions. Concerning the fate of any individual we can say nothing. The problem is, therefore, a statistical one; evolution, as it occurs in nature, is not a problem of "individuals" but of "populations."

A. The Protective Value of Color

For a century field naturalists have observed the close similarity between the colors of organisms and their environment, and have seen in this resemblance an adaptation for protection. Since the advent of the selection theory protective and aggressive resemblance, warning colors, recognition characters and mimicry have been prominent in biological literature, and are conceptions associated with some of the most honored names in biology. Yet almost all the evidence has been comparative, and attempts to determine empirically whether given color patterns are in the long run of vital significance are discouragingly few, and some biologists are now questioning whether the so-called protective adaptations have any value at all.

One of the simplest and most direct tests of the value of any character in determining the chances of survival of an individual is that of Di Cesnola for the protective value of color in Mantis religiosa[1] In Italy the green individuals of this species are found on green grass, the brown ones upon grass burnt by the sun. If the color has any protective value there should be a higher death rate from enemies when the insects are exposed on vegetation of a color unlike their own.

Altogether 110 insects, 45 green and 65 brown were secured and were exposed on separate plants as follows:

 Green insects on green plants 20 Green insects on brown plants 25 Brown insects on brown plants 20 Brown insects on green plants 45

The experiment began August 15 and observations were made daily for seventeen days. Of the forty individuals exposed on vegetation of similar color, every one survived throughout the entire experiment. All the green individuals exposed on brown grass were killed in eleven days; of the forty-five brown individuals exposed on green grass, ten survived to September 1, when a severe gale destroyed the experiment. This is all made clear by Fig. 1.

The biometrician would like to see this experiment carried out on a much larger scale, but when we consider that not one of the forty insects exposed on similarly colored vegetation was killed at the end of seventeen days, while sixty of the seventy exposed on dissimilarly colored vegetation were eliminated[2] by the end of the first eleven days. I think we must regard Di Cesnola's evidence as rather strongly indicating a real protective value in the color dimorphism.

For vertebrates two papers are available. Davenport[3] and Pearl[4] have made observations on the relative number of self-colored and of barred or pencilled birds killed by enemies. Davenport finds that out of 24 chicks from five to eight weeks old killed on one afternoon by three crows only a single one was other than self-colored, although twenty per cent, of the flock of about 300 chicks "had a pencilled or

Fig. 1. Number of Insects surviving day by day in series of Brown and Green Mantis exposed on Brown and Green Vegetation. Ordinates ${\displaystyle {{\ce {=}}}}$ dates; abscissæ ${\displaystyle {{\ce {=}}}}$ number of individuals. Similar color in insect and environment represented by heavy dots; dissimilar combinations by circles.

striped marking more or less like that of the female jungle fowl or ordinary game." He concludes, "this fragment, then, so far as it goes, indicates that the self-colors of poultry which have arisen under domestication, tend to be eliminated by the natural enemies of these birds, and the pencilled birds are relatively immune from attacks because relatively inconspicuous."

Photographs by Pearl show that the barred birds are much less conspicuous in their surroundings than are self-colored ones. Theoretically, their mortality from predaceous enemies ought to be lower, but from a rather large series of observations in his experimental poultry plant only negative results are secured. The actual figures are:

 Class of Birds Number in Original Flock Eliminated by Known Enemies, Almost all Rats Eliminated by Unknown Enemies, Chiefly Preda- ceous Birds All Eliminated Birds Self-colored 336 6 29 35 ${\displaystyle {{\ce {=}}}}$ 1.79% ${\displaystyle {{\ce {=}}}}$ 8.63% = 10.42% Barred 3007 68 222 290 ${\displaystyle {{\ce {=}}}}$ 2.26% ${\displaystyle {{\ce {=}}}}$ 7.38% ${\displaystyle {{\ce {=}}}}$ 9.64% Totals 3343 74 251 325

It will be noticed that when the chicks eliminated chiefly by predaceous birds are examined alone, the proportion of self-colored birds is a little higher, but without further statistics no significance could be attached to the difference.[5]

B. Structural Characters in Relation to Survival

The comparisons in the preceding section were drawn between well-marked color varieties. Many more experiments of this kind are desirable, but if natural selection be a factor of the potency required to account for the origin of specific characters by the accumulation of small variations, it must be shown that the peculiarities of form or color which separate one individual from another are of significance in determining the ability to more than hold its own in competition with its fellows. So far as I am aware pertinent data are available for structural characters only.

Experiments with Crabs

The pioneer in the measurement of the intensity of natural selection was W. F. R. Weldon. His first attempt to determine whether survival may depend upon definite physical characters was made with the common shore crab, Carcinus mænas.[6]

Fig. 2 will be recognized at once, even by the reader whose knowledge of marine biology is limited to the menu-fauna of the city restaurant as the outline of the solid upper portion of the crab's body known as the carapace. In measuring the frontal breadth[7] of Carcinus from a particular spot of beach near the Marine Biological Laboratory Fig. 2. Outline of the upper surface of the Carapace of the Shore Crab, Carcinus mænas. at Plymouth, Weldon and Thompson noticed a peculiar change from year to year. For crabs of the same length of carapace,[8] the frontal breadth seemed to be decreasing.

I have tried to make this clear by a diagram. In Fig. 3 the individuals are classified into twenty-five groups according to length of carapace and the proportional frontal breadth[9] for each class represented for the three years by the position of the circles.[10] The general slope of the connecting lines convinces one that the Plymouth Sound crabs, as observed by Thompson and Weldon, were undergoing a pronounced change in frontal breadth.

The two reasonable hypotheses to account for this decrease are: (1) A modification of the young individuals by the direct action of a changing environment, (2) a decrease in the average frontal breadth in the population due to elimination of the individuals with broader frontal dimensions.

A change in the environmental conditions of Plymouth Sound was undoubtedly in progress during the time when Professor Weldon's observations were made. The streams bring into the sound large quantities

of fine china clay, while the growing cities along the shore and the shipping have greatly increased the refuse thrown into the harbor. The construction of a huge artificial breakwater has minimized the scour of the tide and the waves of severe storms which formerly swept the fine silt out of the sound, so that this was constantly increasing in amount during the period. These physical changes affected the fauna and some organisms disappeared and were only to be found outside the breakwater.

Fig. 3. Change in Frontal Breadth of Carcinus. The slope of the lines shows the change in mean relative frontal breadth for crabs of different length of carapace.

As a first series of experiments Weldon put crabs in a large vessel of sea water in which a quantity of fine china clay was kept from settling by a slow automatic agitator. After a period of time both dead and living individuals were measured. In every case the crabs which died were on the whole distinctly broader than those which lived through the experiment, so that a crab's chance for survival could be measured by its frontal breadth. When the experiment was performed with clay coarser than that brought down by the rivers the death rate was smaller, and was not selective.

By washing the stones under which the crabs live along the beach Professor Weldon obtained a silt of a finer texture than the china clay he had been able to use. This was employed in experiments of the same kind, with identical results.

There seems no reason to suppose that the relation of the crabs to the mud on the beach is different from that in the aquarium. Whenever the fine sediment is stirred up a selective elimination of crabs must occur. It is this selective elimination which "Weldon regarded as furnishing the explanation of the decrease in frontal breadth observed in the measurements in 1903, 1905 and 1908.

Not content with these experiments, Weldon tried to obtain evidence of an entirely different kind. He arranged several hundred aquaria in each of which a young crab from the beach was kept in clear running sea water—and so entirely free from the influence of the mud. They were allowed to moult and grow and harden new shells. When measured they were found to be unmistakably broader than wild crabs of the same length. This is precisely the result to be expected if a selective elimination of broad-fronted individuals occurs in nature.

The source of this difference in capacity for survival seems to lie in the way in which the crabs filter the water entering their gill chambers. Professor Weldon found that a narrow frontal breadth renders one part of the process of filtration of water more efficient than it is in crabs of greater frontal breadth. The gills of the crabs which died during the experiments were covered with fine white mud, and this was not found in the gills of the survivors.

The labor of these experiments—the daily care of hundreds of animals, the thousands of measurements and the drudgery of calculation—was excessive. Most discouraging of all, perhaps, were the sterile and hostile criticisms which are so often the portion of a pioneer.

Observations on Other Invertebrates.

Besults which may be logically attributed to the action of natural selection but which by reason of the possibility of other explanations are not conclusive evidence for its potency, have sometimes been secured by biometricians concerned with other problems. For instance, Warren[11] adduces "the elimination of the physically unfit" as one of the factors to account for the difference in variability of the termites of the same nest at different seasons. Possibly this factor may also account in part for differences in variability from nest to nest, but of course much more extensive and direct evidence must precede any final conclusions.

Another study of variation in insects, social and otherwise, is that of Kellogg and Bell.[12] They decide against natural selection, but their evidence for the lady beetle, Hippodamia convergens, can not be regarded as conclusive since they have made no direct comparison of eliminated and surviving individuals. Their case for the honey bee where observations are made upon free flying individuals and those which have not yet left the shelter of the hive, is much better, but even here I must feel that their numbers are too small to give finally conclusive results in a problem so difficult as that of natural selection. Furthermore, they suggest that the more abnormal individuals may be made way with before they have the opportunity of leaving the hive.

A most suggestive result was obtained by Schuster in an investigation of deep and shallow water crabs of the genus Eupagurus[13] He finds that for both sexes, but especially for the males,[14] the individuals from deep water were more variable than those from shallow water.[15] Schuster wisely leaves the determination of the reason for this difference in variability to the time when more data, and data collected under the guidance of this first study, shall be available. He points out, however, that if these differences in variability are not those of deep water and shallow water local races, but arise anew in each generation, they must be due either to the direct influence of the environment or to selection. If elimination be the true explanation the less variable shallow-water forms would be regarded as a selection from the more variable deep-water population.

Turning again to studies carried out primarily to test the possible action of natural selection, we may mention the work of Browne on the medusa, Aurelia aurita[16] In this jelly fish the number of marginal sense organs, tentaculocysts, is definitely fixed in the larval stage commonly known as the Ephyra, and by a comparison of collections of Ephyæ and adults it is possible to determine whether variation in the number of the marginal sense organs affects the chance of survival from larval to adult life. Since all of the young and adult populations compared were sensibly identical, one must conclude that neither an increase nor a decrease in the number of tentaculocysts is so injurious that there is any selective elimination during development.

Crampton's study of pupal and pupal-imaginal elimination in the Ailanthus silk-worm moth[17] I will pass by without discussion, since I believe he is soon to publish further observations on the same subject.

Weldon's work on natural selection was not limited to Crustacea, but extended to the mollusca as well.

The shells of certain snails, such as Clausilia and Helix, is essentially a tube increasing in size as the animal grows older and wound in a spiral, or more properly a helix, around a central axis with the successive coils in contact. If one of these shells be cut longitudinally, the central cone, or columella, as it is technically called, will be laid open and will appear as a narrow conical tube extending the whole length of the shell, while the tube which contained the animal will be cut across twice in each complete revolution and will appear in cross section.

This point is made quite clear by an examination of the three diagrams (Fig. 4).

Fig. 4. Longitudinal Sections of Clausilia laminata (A), Clausilia italia (B), and Helix arbustorum (C), after Weldon and Di Cesnola in Biometrika.

By deft manipulation such sections can be prepared. A shell may be ground upon a soft stone until a plane which extends almost exactly through the central columella is exposed.[18] From such a preparation it is quite possible to make the measurements which determine the pitch and several other characteristics of the spiral.

The shell of these snails is a permanent structure. In the adult the whorls first laid down by the young animals can be measured. Now it is clear that one can compare the properties of the portions of the spiral already laid down in the shell of a young snail with the same portions in the shell of an adult. In snails, as in other animals, not all individuals survive to adult life. The problem is to find out whether the properties of the first-formed portions of the spirals of the shells of individuals which have survived to adult life differ from those portions of the shells of young snails, part of which will survive and part be eliminated. If differences do occur they are most easily explained as due to a selective mortality in the animals during their period of growth.

Details of measurement and calculation are entirely too elaborate and complicated for explanation here. In two studies, one by Weldon[19] on Clausilia laminata from Gremsmühlen in Holstein and one by Di Cesnola[20] on Helix arbustorum from the banks of the Isis near Oxford, the authors found that while there is no difference between the mean characters of young and old shells there is a distinct difference in variability. This kind of selective elimination which recurs every generation and by which the existing type is maintained (without necessarily giving rise to any progressive change) Weldon designates as periodic, in accordance with Pearson's terminology.[21]

The reader must not conclude from what has just been said that Weldon regarded variation in the peripheral radii as the direct cause of the selective elimination. "Such selection is, of course, 'indirect,' that is to say, the life or death of the individual is determined in each case by the value of a (probably large) number of correlated characters, of which the length of the peripheral radius is only one." With justice Weldon emphasizes the minuteness of the structural differences which seem to mark the boundary between fitness and unfitness for survival in Clausilia laminata.

The results of Weldon's investigation[22] of Clausilia italia did not agree with that of his former study of C. laminata. By some readers this fact will be interpreted as vitiating entirely any conclusion to be drawn from all this laborious work on shells. To my mind this attitude is quite wrong. Laying aside the fact that Weldon has suggested biological reasons which may explain why no change in variability was observed between young and old individuals in C. italia, we must bear in mind the fact that there is no justification whatever for assuming that natural selection, either secular or periodic, is to be observed at all times in all species. Naturally contradictory results call for repetition and amplification and for more refined control of conditions—but these are the things which make for the advancement of science. Only the merest beginning has been made in the study of selective elimination, but Weldon has shown us the way in which the problem may be attacked in two large groups of invertebrates. If other workers with his patience are ready to volunteer their service to this phase of evolution, the next ten years ought to see a material advance in our knowledge of the least investigated of the Darwinian principles.

Studies of Vertebrates.

The relative ease with which large numbers of individuals can be secured and observed is an ample explanation of the fact that practically all the studies of selective elimination have been made on invertebrates.

To Dr. H. C. Bumpus belongs the credit of the first effort to determine whether the death rate among vertebrates may depend in some degree upon the measurable physical characteristics of the individual. Indeed, in the attempt to apply quantitative methods to the problem of natural selection Bumpus[23] was a close competitor for priority with Professor Weldon. His statement of the problem, like that of Professor Weldon, is beautifully clear:

A possible instance of the operation of natural selection, through the process of elimination of the unfit, was brought to our notice on February 1 of the present year (1898), when, after an uncommonly severe storm of snow, rain and sleet, a number of English sparrows were brought to the anatomical laboratory of Brown University. Seventy-two of these birds revived; sixty-four perished; and it is the purpose of this lecture to show that the birds which perished, perished not through accident, but because they were physically disqualified, and that the birds which survived, survived because they possessed certain physical characters. These characters enabled them to withstand the intensity of this particular phase of selective elimination, and distinguish them from their more unfortunate companions.

From his measurements of various bodily dimensions Professor Bumpus concluded that the birds which perished were actually differentiated from those which survived. Some of the differences, however, are so small that to the cautious statistician this attempt to measure the influence of a selective death rate on the type of a population of birds living in a state of nature seems suggestive rather than finally conclusive.

This must not be read as a criticism of Bumpus's work, for he not only saw the problem and the possibility of applying the new methods to it, but he also gave us the full results from an unusual opportunity.

First and last, considerable has been written concerning natural selection in man. Most of the arguments are purely general or speculative, but Beeton and Pearson[24] have succeeded in obtaining more definite evidence.

From extensive researches on inheritance in man, Pearson and his associates have shown that duration of life gives much lower correlations—both parental and fraternal—than the substantial values found for other physical and psychical characters. If the duration of life of an individual were absolutely determined by physical and mental fitness, then one would expect it to show a correlation as high as that found for other characters. The fact that the values are regularly and conspicuously lower evidences for the existence of a non-selective death rate. The relative amount of the selective and non-selective death rate may be roughly estimated from the reduction in correlation as one passes from the inheritance of characters in general to that of longevity. By this means Beeton and Pearson calculated that from fifty to eighty per cent, of the death rate in civilized man is selective.

C. The Fitness of Organs

Except among the lowest forms of life, every animal or plant is made up of a large number of parts which are differentiated in form and function. The fitness of such a complex organism for self preservation and perpetuation probably depends not merely upon the degree of development of its several component members, but also upon the nicety with which they are coordinated.

Undoubtedly the proper way of taking up the study of Natural Selection is to compare by means of the measurement of particular organs series of individuals which survive with series which perish, but after this is done in a large number of cases we shall have considered only the first part of our problem.

This first phase consists in finding out whether variations in the form, size or other property of an organ affects its efficiency to such an extent as to prejudice the chances of survival of the individual possessing it.<ref>In actual work one is at once confronted by the difficulty that variations in the organ he is studying may have no real influence upon the chances of survival but merely an apparent significance due to its correlation with other organs. /ref>

Involving, as it does, questions of structural characteristics and functional efficiency this is at bottom a problem on the boundary line between morphology and physiology. For several years it has seemed tc me that we might, in the long run, make better progress in the study of the problems of evolution if we turned our backs on some of its more complex phases and devoted ourselves temporarily to the morphological and physiological problems upon which they rest.

While testing out this idea in studies of the relationship between structural characteristics and fertility in various fruits, certain inconsistencies in results were found which could be most easily explained by the assumption that there is a selective elimination in which ovaries

Fig. 5. Comparison of Matured Fruits with Fallen Ovaries for Twenty-eight individual Shrubs of Staphylea. Differences expressed In percentages of the mean of the eliminated series. Solid dots and broken lines ${\displaystyle {{\ce {=}}}}$ mean number of ovules; circles and firm lines ${\displaystyle {{\ce {=}}}}$ radial asymmetry.

of certain types are extensively weeded out Direct investigation proved the correctness of this assumption. The results will be set forth very briefly.[25]

In the selective elimination which occurs during the development of the ovary of the American bladder nut, Staphylea, into a fruit, the mean number of ovules is materially increased. For a small series of developing ovaries taken at the Missouri Botanical Garden in 1906, the oldest—the group from which the most elimination had taken place—had about eight per cent, more ovules per fruit than the youngest. In large samples taken in 1908, the fruits which matured had about seven per cent, more ovules than those which were eliminated. The same result is seen if the material is split up into twenty-eight individual pairs of samples, each from a separate tree. This is made clear by Fig. 5. The solid dots connected by broken lines show the percentage excess of the matured fruits over the fallen ovaries in the number of ovules. In twenty-seven cases out of twenty-eight the number is larger in the ovaries which mature!

Fig. 6. Percentage of Ovaries which are perfectly radially Symmetrical in youngest and oldest Collections, Missouri Botanical Garden, 1906.

The ovary of Staphylea is three-celled. If each cell contains the same number of ovules,e. g., 8-8-8, it may be regarded as radially symmetrical, while if the numbers differ from locule to locule, for instance 8-7-8 or 9-10-8, the ovary may be described as radially asymmetrical. If this radial asymmetry be expressed by a statistical constant such that a perfectly symmetrical fruit shall have a degree of asymmetry of 0, while the coefficient increases as the ovaries become more irregular, one can compare asymmetries in the ovaries which do and those which do not develop to maturity as easily as he can the means.

Fig. 6 shows the percentage of perfectly symmetrical ovaries in the youngest and oldest series of the 1906 collection. The conclusion that the conspicuously higher percentage of perfectly symmetrical ovaries in the oldest collection is due to a selective mortality by which the more irregular ones are weeded out is fully substantiated by the statistics of 1908.

The average asymmetry for the matured fruits for 1908 is about seventeen per cent, lower than the mean for the eliminated sample. For the individual trees the results are somewhat more irregular than they were for the mean number of ovules, but Fig. 5 shows that in twenty-one cases to seven the asymmetry of matured fruits is less than that of the ovaries which do not complete their development. Taken as a whole, the differences show an unmistakable tendency to fall far to the negative side of the 0 bar.

Not merely the degree of radial asymmetry of the ovary, but the number of its locules which have an odd number of ovules, seems to be of consequence in determining whether an ovary shall complete its development. Ovaries with even numbers—6, 8, 10, 12—of ovules in their locules have a better chance of developing to maturity than do those with one or more locules with an odd number. The question is too involved for adequate discussion, and I will leave the subject with a mere reference to Fig. 7 which shows that in the 1908 series the reduction in the percentage of "odd" locules is a very material one.

Fig. 7. Percentage of "Odd" Locules in eliminated and matured Ovaries of Staphylea—1908 collection.

The work just outlined has been rather drastically criticized on the ground that studies on the selective elimination of organs can never have any bearing on the problem of evolution. I think there is room for differences of opinion on this point, but at present the purely morphogenetic and physiological sides of the problem are of paramount interest; our knowledge of facts is too meager to justify speculations on so complex a problem as that of the origin of species.

IV. Concluding Remarks

In the paragraphs which have preceded these I have tried to set forth honestly the results which have been secured in attempts to ascertain by direct quantitative methods the intensity of the selective elimination which may occur in nature.

What is the general significance of these results? What claim have they to the special attention of scientific men?

First, let us premise that the measurement of natural selection is not synonymous with such an expression as the demonstration of the natural selection theory. Upon the application of biometric methods many supposedly valid biological theories have shrunken to nothing; possibly this may ultimately be the fate of the natural-selection theory. In approaching the problem our aim is not to "get positive results," but to find out the truth. Our object is not to bolster up a venerable and out-of-fashion hypothesis, but to test conscientiously that hypothesis against concrete data.

Like other theories, the Darwinian theory must stand or fall according as the evidence of quantitative biology shall be for or against it. If the micrometer scale and the calculating machine show that any given character has no influence in determining whether an individual shall survive, then for that organ, in that species at the time under consideration, evidence for the potency of selection is wanting.

The problem is a difficult one; a priori one would expect most generally to find no changes taking place in the characters of a species because of a selective death rate. If natural selection be actually at work in nature, it is likely that the ancestors of individuals collected in the open will have been subjected to the selective factors which one is trying to measure, and that the race will be held pretty close to the attainable limit of perfection. It is more likely that a selective elimination which recurs every generation will be observed than one of the kind that brings about changes in specific characters. Only in rare cases when a new territory is opened to organisms or some special modification of environment (inorganic or organic) has taken place can we reasonably expect to see the changing of types going forward. Possibly the very difficulties of demonstrating a selective death rate bear witness to its reality!

Taking all this for granted, biologists must, it seems to me, face the duty of determining whether natural selection is a fundamental factor in evolution—in short of actually measuring the intensity of the selective death rate. The calipers are ready and their efficiency has been proved.

The duty to use them is imposed by ideals of good workmanship. "Measure that which is measurable and render measurable that which is not," is the ideal which has hitherto separated the precise from the descriptive sciences. It is the duty as well as the opportunity of the biologist of to-day to break down this distinction.

The duty to use them is imposed by the history of our science. For nearly half a century natural selection has been one of the chief problems of biology, and it would be cowardly for naturalists of this generation to leave the problem until a definite solution has been secured.

The duty to use them is imposed by the inability of the biologist to construct for himself a philosophical theory of evolution without natural selection as one of the factors. Yet the philosophical necessity of a given factor does not relieve the scientist from the duty of finding out whether that factor be a reality, and of measuring its intensity.

1. Di Cesnola, A. P., "Preliminary Note on the Protective Colour in Mantis religiosa," Biometrika, Vol. III., pp. 58-59, 1909.
2. Most of the insects were destroyed by birds; five were known to have been killed by ants.
3. Davenport, C. B., "Elimination of Self-coloured Birds," Nature, Vol. LXXVIII., p. 101, 1908.
4. Pearl, R., "Data on the Relative Conspicuousness of Barred and Self-colored Fowls," Amer. Nat., Vol. XLV., pp. 107-117, figs. 1-4, 1911.
5. It would be very interesting if data could be obtained from flocks of young chickens in a diversified environment—i. e., one in which there is a variety of underbrush, weeds, stones, etc., giving wider opportunity for hiding. Davenport's chicks were on a "well-cropped pasture" and Pearl's birds "ran together on the same open, turf-covered range." Now it is quite possible that barring might afford no protection on open turf, and yet be most advantageous in a thicket. Some poultry man could do a very good service to science by appropriating a few hundreds of young birds to the hawks and crows, allowing them to have the run of a lot affording a diversity of shelter. Only where the habitat simulates closely the kind in which animals are found in nature can an experiment of this kind be really critical.
6. Weldon, W. F. R., "An Attempt to Measure the Death-rate due to the Selective Destruction of Carcinus mænas with Respect to a Particular Character" (Report of the Committee for Conducting Statistical Inquiries into the Measurable Characteristics of Plants and Animals), Proc. Roy. Soc. Lond., Vol. XLVII., pp. 360-379, 1894. Also, W. F. E. Weldon, presidential address to the Section of Zoology, British Association, Report of Bristol Meeting (1898), pp. 887-902, 1899. Interesting and valuable supplementary information concerning Weldon 's studies on selective elimination are to be found in Pearson's biographical memoir of Professor Weldon (see Biometrika, Vol. V., pp. 1-52, pl. I.-V., 1906).
7. The distance between the tips of the extra-orbital teeth, from the point A to the point A' in the figure.
8. There is no way of knowing precisely how old an individual beast is; if the specimens for different series are sorted into classes of about the same length of carapace, on a line from C to D, and if there is no reason to suspect any differences due to special environmental influences, dimensions of other parts of the shell can be compared in different lots with reasonable confidence that animals of about the same average age are being examined.
9. The frontal breadth is expressed in thousandths of the carapace length.
10. For 1898 the number of observations is not large enough for thoroughly satisfactory determinations.
11. Warren, E., "Some Statistical Observations on Termites, Mainly Based on the Work of the Late Mr. G. D. Haviland," Biometrika, Vol. VI., pp. 329-347, 1909.
12. 12 Kellogg, V. L., and Ruby G. Bell, "Studies of Variation in Insects," Proc. Wash. Acad. Sci., Vol. VI., pp. 203-332, 81 figures, 1904.
13. Schuster, E. H. J., "Variation in Eupagurus prideauxi," Biometrika, Vol. II., pp. 191-210, 1903.
14. 14 Deep water forms were those taken at a depth more than 35 meters; shallow water forms from a depth of less than 35 meters.
15. The males are more variable than the females, in both deep and shallow water.
16. "Browne, E. T., "Variation in Aurelia aurita," Biometrika, Vol. I., pp. 90-108, 1901.
17. Crampton, A. E., "Experimental and Statistical Studies upon Lepidoptera. I. Variation and Elimination in Phylosamia cynthia," Biometrika, Vol. III., pp. 113-130, 1904.
18. To be sure, the work is exceedingly tedious and many shells are accidentally spoiled, but four or five may be prepared in a day's work and in the course of time a number sufficient for statistical work may be secured.
19. Weldon, W. F. R., "A First Study of Natural Selection in Clausilia laminata," Biometrika, Vol. I., pp. 109-124, 1901.
20. Di Cesnola, A, P., "A First Study of Natural Selection in Helix arbustorum," Biometrika, Vol. V., pp. 387-399, 1907.
21. Pearson, K., "Grammar of Science," 2d ed., pp. 413-414, 1900.
22. Weldon, W. F. R, "Note on a Race of Clausilia Italia," Biometrika, Vol. III., pp. 299-307, 1904.
23. Bumpus, H. C, "The Elimination of the Unfit as Illustrated by the Introduced Sparrow, Passer domesticus," Biological Lectures from the Marine Biological Laboratory, Woods Hole, 1908, pp. 209-226, Boston, 1899. For further statistical constants calculated from Bumpus 's data see a note entitled, "A Neglected Paper on Natural Selection in the English Sparrow," Amer. Nat., May, 1911, p. 314-318.
24. Beeton, Mary, and K. Pearson, "A First Study of the Inheritance of Longevity and the Selective Death-rate in Man," Proc. Roy. Soc. Lond., Vol. LXV., pp. 290-305, 1889. Beeton, Mary, and K. Pearson, "On the Inheritance of the Duration of Life, and on the Intensity of Natural Selection in Man," Biometrika, Vol. I., pp. 50-89, 1901.
25. Those who care for a detailed account may find it in three papers: (a) "Is there a Selective Elimination in the Fruiting of the Leguminosæ?" Amer. Nat., Vol. XLIII., pp. 556-559, 1909; (b) "On the Selective Elimination during the Development of the Fruits of Staphylea," Biometrika, Vol. VII., pp. 452-504, 1910; (c) "On the Selective Elimination of Organs," Science, n. s., Vol. XXXII., pp. 519-528, 1910.