Annals of the Missouri Botanical Garden, 28: 287-292 (1941)
Geneticist to the Missouri Botanical Garden
Engelmann Professor in the Henry Shaw School of Botany of Washington University

1 Anderson, Edgar, and W. B. Turrill. Biometrical
studies on herbarium material. Nature 136: 986.

A slight extension of ordinary herbarium techniques promises to increase the accuracy of herbarium studies and greatly to widen their scope. Attention was called to the possibility in 1935.1 Since that time its potentialities have become increasingly evident and details of its technique have been considerably improved. The method consists in supplementing ordinary specimens by mass collections made as described in detail below. For small plants these mass collections may well consist of the entire plant. For larger plants considerations of space require the selection of some critical portion, as the leaves, the inflorescence, or the fruits. For instance, in studying sugar maples (Acer saccharum and its relatives) a mass collection will consist of one leaf per tree (carefully selected from the same kind of non-fruiting branch) from thirty to fifty trees, and complete specimens of the usual sort, from two or three of the trees. Properly made and filed (see below) mass collections require only a little more space than ordinary herbarium specimens.

Such a collection is a record of a population as well as of the individuals which make up that population, and it therefore gives the facts about variation which can be obtained from populations but not from individuals. In other words, it would bring into the herbarium information which now we can get only in the field. This information can be grouped under three different heads.

(1) Frequency of the variation.—The most important defect of the ordinary herbarium material, in biometrical terminology, is that, while it may give a reliable estimate of the range of variation, it does not allow an estimate of the frequencies within that range. That is to say, in non-mathematical language, that it may give a good idea of extremes but it does not indicate the relative prevalence of the extremes or of any particular intermediate. It is not enough to know that a variant exists; for its complete interpretation one needs to know how often it occurs in the places where it has been reported. As Dr. Fassett has put it in a homely analogy: "there are Democrats and Republicans in both Mississippi and Vermont but their comparative frequency varies significantly between these two regions."

(2) Discontinuity of variation.—This is potentially one of the greatest sources of error with present-day techniques. The chief criterion for separating taxonomic entities is the degree of morphological discontinuity between them. At the present time it takes good judgment and often field experience to decide if the discontinuity shown by a group of herbarium specimens is real or only apparent. This is particularly true for categories smaller than the species.

2 Statistical studies on two populations of Fraxinus. New Phytol. 37: 160-172. 1936.

(3) Correlation between variables.—While an estimate of this correlation can be obtained from ordinary herbarium specimens, it can be derived much more precisely from mass collections and can be perceived more readily and its perception requires less biological judgment. Those who have undertaken monographic work will have encountered complexes in which variation was so extreme and involved so many different characters that it was difficult to comprehend. Mass collections make it possible to study such complexes analytically and to determine precisely the extent to which the variation in different characters is correlated. Anderson and Turrill,2 for instance, by using mass collections, were able to resolve the variation in the Fraxinus Pallisae complex into two elements and to relate these elements to species of Fraxinus from southeastern Europe.

There are two problems in making mass collections: what part of the plant to collect and the selection of a random sample of the population. The first is not as difficult as it may seem to anyone who has not tried it. Taxonomic studies on the customary herbarium material are a necessary foundation for the making of mass collections intelligently, and the study of a taxonomic revision will tell what parts are significant and should be collected in quantity. The portion of the plant chosen for intensive collection should provide good morphological criteria; it should he easy to press and store in quantity; and its selection should be definable in precise terms. The following examples may make these points clear: Tripsacum, the terminal inflorescence of each plant; Monarda, an average flower-head from each plant, with its subtending bracts; Veronica peregrina, the entire plant. Wherever possible mass collections should be a series of one sample from each plant. Occasionally a single vegetatively reproduced individual (technically known as a clone) may cover a very large area, and it may be difficult or impossible to know where one individual leaves off and another begins. This is particularly true in such plants as Sanguinaria canadensis whose rhizomes grow and branch vigorously and the organic connection between two branches usually rots away after a few years leaving them physically separate. In most cases a careful study of flower and leaf variation will reveal the probable extent of each clone, and a careful collector can minimize the chances of gathering a disproportionate number of samples from a single clone.

The problem of a truly random sample is difficult, and bristles with difficulties which are not even suspected by the uninitiated. One should bear in mind that he is trying to make a record of a population of individuals and that the record will have the greatest significance if it is chosen at random from an actual inter-breeding population. Lacking the precise information as to what an ''actual inter-breeding population'' may be, one can only use his biological judgment in selecting for each case an area which presents uniform conditions for that species and make his collections from that area. Even for those species which grow in definite, more or less isolated colonies, the trained eye can often find evidence for distinctive neighborhoods within such communities, and it will be a matter of individual judgment whether these neighborhoods should be recognized or ignored in making a sample of the population. When the area has been chosen one may either collect a specimen from each individual within that area, if there be not too many, or make a random selection of thirty to one hundred or more individuals. One can walk across it in a straight line, making collections at every second or third step, or can use strings and collect every individual which is touched by the string or is close to it.

Figure 1.

If properly made, mass collections do not take up more space than a few herbarium sheets, and they give information which cannot be obtained from hundreds of ordinary specimens. If supplemented by complete specimens there should be no objection to their fragmentary nature. Technical improvements in storing the collections have been made by my assistant, Mr. Leslie Hubricht, and are illustrated in the accompanying figure. Since my collections are subject to intensive study but are not distributed in an ordinary public herbarium they are kept unmounted. To prevent the specimens from scattering, the genus covers are folded so that the edges meet down the middle rather than the side. One label is made for each collection and is pasted on the genus cover. The ordinary specimens made at the same time and place are mounted and labeled and are kept in the same genus cover. Each collection is given a geographical name and all are assembled alphabetically under the genus or the species, depending upon the nature of the particular problem.

3 Mayr, Ernst. Speciation phenomena in birds. Amer. Nat. 74: 249-278. 1940.
4 Kinsey, Alfred C. The gall wasp genus Cynips. A study in the origin of species. Indian Univ. Studies. 84-88: 1-577. 1930; The origin of higher categories in Cynips. Indiana Univ. Publ. Sci. Ser. 4: 1-331. 1936.

The information derived from a study of mass collections is useful in two ways. It will, in the first place, aid the systematist in cataloguing the various entities involved, species, varieties, forms, etc. While it may raise more new questions than it may solve old ones, it will aid in the production of monographs whose categories are more accurately adjusted to the variation patterns of their particular genera. Mass collections have for some time been customary in avian taxonomy (see, for instance, Mayr3), and Kinsey, in a series of brilliant monographs,4 has shown their superiority in insect systematics. If taxonomy were to be nothing more than cataloguing and if taxonomists were to confine themselves to the problems raised by their herbaria, mass collections would still be a useful adjunct to herbarium technique and in many critical groups would provide more efficient working material, even when their special difficulties of collecting and filing are considered.

There is no reason, however, why taxonomy should be content to cultivate such a narrow field. If collectors and herbarium administrators could he persuaded to encourage mass collections, critically made and carefully assembled, a second kind of problem could he investigated in herbarium material. The description and analysis of geographical trends in variation, the delimitation and interpretation of centers of variation, the establishment and analysis of variation patterns in different genera and families, are only a few of the problems that might well be investigated. It is already possible to correlate information from the field of taxonomy with that from cytogenetics. The time is not far distant when the biochemist of the germplasm will also turn to the taxonomist for morphological evidence derived from studying the products of the germplasms. To speak with authority on such questions taxonomists will need to refine their biological as well as their bibliographical techniques.

Annals of the Missouri Botanical Garden, 28: 293-297 (1941)
University Fellow in the Show School of Botany of Washington University

1 Rhodora 40: 178, 1938.
2 Ann. Mo. Bot. Gard. 28: 287-292. 1941.
3 Ann. Mo. Bot. Gard. 28: 299-374. 1941.
4 ILLINOIS, ST.. CLAIR CO.: in river-bottom woods, about 1 mile northwest of New Athens, May 7, 1940. MISSOURI, ST. LOUIS co.: on top of a ridge, 2 miles south of Allenton, 5. 10, T. 43N, 11, 3E, April 28, 1940; along a steep south slope below the Frisco railroad tracks, at Meramec Highlands, 5. 10, T. 44N, B. 5E, May 18, 1940. FRANKLIN co.: at the ''cliff,'' Missouri Botanical Garden Arboretum near Gray Summit, S. 17 & 20, T. 43N, B. 2E, April 21 and May 5, 1940; on a south slope at Spring Creek, 4 miles northwest of Stanton, May 11, 1940.

In the St. Louis vicinity, Camassia scilloides (Raf.) Cory exhibits a considerable amount of variability, which is apparent on looking at a number of plants even casually. This fact has been given some recognition by the description of C. scilloides forma Petersenii Steyermark.1 In the spring of 1940 an attempt was made to learn something about the statistical nature of the variation. "Mass collections" (Anderson,2 Fassett3), each consisting of 25-50 or more inflorescences, and in some cases leaves, were made at the places which are cited in detail in the footnote,4 and which will be referred to as New Athens, Allenton, Meramec Highlands, Gray Summit and Spring Creek, respectively.

The plants of the first collection were first studied to decide what features of the variation could be most satisfactorily submitted to measurement. The dimensions of the inflorescence seemed most promising. The "open" appearance of some of the inflorescences, as contrasted with the compactness of others, seemed obviously related to internode length and pedicel length, and these lengths were measured in plants of each of the collections. There is considerable difference in length between successive internodes, so that instead of measuring a particular one, the length of the lowest eight was used.

The lowest pedicel was measured in each inflorescence. Neither of these lengths changes appreciably after anthesis, and no inflorescences were measured in which at least half of the flowers had not bloomed. As a check, measurements of members of a few supposed clones were made, and were found to agree.

When internode length was plotted against petiole length in a scatter diagram, a simple picture of the variability within each collection was obtained. Not only was there variability within each of the colonies, but a striking difference was seen between colonies. The Meramec Highlands, Allenton, Gray Summit and Spring Creek collections, all of which were made within the Ozark region, produced essentially the same sort of scatter diagrams, while that for the New Athens material was quite different (see fig. 1). At New Athens the range of variation in these two characters is quite restricted as compared with Meramec Highlands and the other Ozark stations. Although the Meramec Highlands collection includes practically all the types represented at New Athens, most of the plants from Meramec Highlands lie completely outside the range of variation of the New Athens colony.

Figure 1.


5 Wright in Jour. Amer. Stat. Assoc. 21: 162-178. 1926.

The same sort of difference in variability is shown by the scatter diagrams in fig. 2, where sepal length and width, as measured from camera-lucida drawings of fresh material, are plotted. (The difference, however, is not so pronounced; or is it exaggerated in the case of the internode and pedicel lengths by the use of an inappropriate scale?5 )

Furthermore, the uniformity of the New Athens plants as compared with those collected in the Ozark localities is apparent in their general aspect, whether seen in the field or as dried specimens. See the photographs of dried inflorescences in pl. 8.

Figure 2.


Many of the features of the variation which is so apparent in the Camassias of this region cannot be studied statistically, but some of them will be discussed in general terms here. The color of the flowers varies somewhat. While the prevailing color is a pure, pale blue, there is an appreciable range in the depth of the blue color, and in many plants a slight reddish tinge is perceptible. The petals (and sepals) vary both in size and shape, one of the details of the variation being the presence or absence of a sagittate base. It is also easy to see variation in style length and form.

The general aspect of the inflorescence presents easily apparent differences, as mentioned above. There are inflorescences which are decidely cylindrical in shape, and others which appear more or less pyramidal, even when all allowances are made for the acropetal order of flowering. In some plants the inflorescence appears compact, with the flowers greatly crowded; in others it is open and spindly in appearance. The peduncle varies, the extremes being a stout and fleshy sort, and a quite slender and woody one. The bracts are usually inconspicuous and completely withered at anthesis, but plants can be found in which large green bracts, approaching in size the basal leaves, occur at the lowest few nodes of the inflorescence. The largest of these bracts do not subtend flowers. Incidentally, no such large, leaf-like bracts were found among the New Athens plants.

The leaves offer differences in size, and in the amount of bloom on the upper surface. There also appear to be differences in shape of the bulb, which are not connected with its size.

In all of the above-mentioned characters, the New Athens collection is less variable than are the other four.

6 Fenneman, Physiography of the eastern United States. New York. 1938.

Although the two collections which have been discussed in detail were made less than fifty miles apart, they are from quite different habitats, and for that matter from different physiographic regions. Meramec Highlands (as well as Allenton, Gray Summit and Spring Creek) is on the northeastern edge of the Ozark Plateau,6 and at these Ozark localities Camassia grows for the most part on steep slopes. New Athens is in the Till Plains section of the Central Lowland," and Camassia was there found growing in rich bottom-land soil. The former region is unglaciated, and has been occupied by plants continuously since preglacial times, while the latter was covered by the Illinoian ice sheet.

While it is not the purpose of this paper to offer a complete explanation of the facts presented above, one might suppose that the differences which have been demonstrated between Camassia of the Ozarks and of the Illinois bottom-lands are related to the different vegetational histories of the two regions.