USDA Yearbook, 1898

Special Agent of Division of Vegetable Physiology and Pathology


Selection is one of the most important factors in plant breeding, the natural capacity of all plants to vary furnishing the basis on which the breeder has to work. The prime factor of selection, or, as Darwin calls it, the "law of the preservation of the favorable individual differences and variations and the destruction of those which are injurious," consists in the skillful selection and propagation of plants showing desirable variations. Selection is frequently understood as meaning simply the use of large, vigorous seed, and a consequent slight increase in production. While such selection is no doubt beneficial, a factor of far more importance is that the seed be taken from vigorous, productive plants. The inherent potentiality of the mother plant, if it may be so expressed, is a more important consideration than the size of the seed, which is so materially influenced by the general productiveness of the plant.

The object of selection is to effect a complete transmission of the desired qualities and to augment them if possible, and the main factor with which the breeder has to contend is the varying degree of the power of inheritance possessed by individual plants.

Plant breeding includes two processes, largely distinct in their nature: (1) The methods of securing variation, and (2) the fixation of desirable variations by methodical selection. The first process has been discussed in previous Yearbooks, the variations arising naturally and supposed to be induced directly or indirectly by environment having been the subject of a paper by the writer in the Yearbook of 1896, pages 89-106, while the variations induced by crossing different varieties, species, or genera were treated of by the writer in conjunction with Mr. W. T. Swingle in the Yearbook of 1897, pages 383-420. In the present paper will be discussed the second process—the fixation of desirable variations by methodical selection and the gradual improvement of plants resulting from the cumulative effect of selecting through many generations those showing the very slight modifications which normally occur in all plants.


The unity of the individual taken as a whole is a factor of prime importance in selection and should be clearly recognized by every one striving to secure improved pedigree plants. If the object in selecting any fruit is to decrease seed production it is of prime importance that the seeds be selected from plants which have been found, through examinations of a number of fruits front different parts of such plants, to have a general tendency toward reduced seed production. The plants from such seeds are very liable to inherit the tendency to seedlessness in a greater or less degree, but seeds taken from selected individual fruits that produce but few seeds, cases of which can be found on almost any tree, are no more apt to produce seedless fruits than are seeds taken from unselected fruits. Henri de Vilmorin, whose extensive experience enables him to speak authoritatively on such points, says that "the unity of character of any single plant is the main factor in the work of pedigree or grade breeding." His experiments with chrysanthemums forcibly illustrate this point: ''I tried an experiment with seeds of Chrysanthemum carinatum gathered on double, single, and semidouble heads, all growing on one plant, and found no difference whatever in the proportion of single and double-flowered plants." Had the seeds been selected from plants on which all the flowers were double or semidouble a very large and probably even an increased proportion of double flowers would have been obtained.

Livingston, who has had such marked success in establishing improved sorts of the tomato, in explaining the origin of the Paragon tomato very graphically describes his independent discovery of this principle of the unity of the individual:

In passing over my fields of growing tomatoes, which were still of all sizes, sorts, and shapes, my attention was attracted to a tomato plant having distinct characteristics and bearing heavy foliage. It was unlike any other in the field or that I had ever seen. It showed itself very prolific. Its fruit was uniformly smooth, but too small to be of general market value. As I examined it closely, observing how alike every tomato was on the stalk, wishing they were larger, and meditating over its possibilities long, it came to me like an inspiration, Why not select special tomato plants instead of specimen tomatoes?

In selecting Indian corn for seed it is the common practice to collect the best ears at the time of husking, the main object being to secure ears of good size and shape, and having large, well-formed kernels and a proper proportion of cob to kernel. This method of selection, while good so far as it goes, does not take the vigor of the plant into account, and hence does not accomplish all that is intended. The largest ears may grow on comparatively unproductive or weak stalks, and therefore to obtain the best results seed corn should be selected in the field, and attention given to the habit, productiveness, general vigor, etc., of the plant, as well as to the characters of the ear, kernel, and cob, and uniformity in ripening. The same remarks apply to the selection of seed wheat, but the difficulty in selecting wheat from plants fulfilling all the requirements would probably be too great for this method of selection to he feasible as a common practice.

Individuals may be very unequally endowed with the power of transmitting their own characteristics to their progeny, and where several individuals are selected in breeding for the same feature it is best to keep the seed from each plant separate. The principal aim is to secure the complete transmission of the qualities desired, and in case the progeny of any plant selected shows a decided tendency to revert to the original unselected type it is best to make future selections from the progeny of some other plant which has more uniformly impressed its qualities on its offspring.

The instability of "bud sports" further illustrates the importance of attention to the unity of the individual when depending on selection to improve or fix a sort. Frequently a certain branch differs greatly in form of fruit or leaf from all other branches on the same tree. Such variations, termed "bud sports," are seldom reproduced true through the seed from fruit grown on the branch showing the variation. If such variations are desirable, they may of course be perpetuated by budding or grafting in the case of plants commonly propagated by such methods. The different branches on any plant almost invariably possess certain characteristic differences, and in selecting an individual front which to obtain seed with a view to improve any feature, it is important that this feature should be common to all branches of the tree and not limited to any particular one.


The variations which form the basis for selection and the formation of new and improved races of plants are the direct or indirect results of changed environment or of hybridization and cross fertilization. Ray Lankester says:

For the present I see no evidence of a production of new races on the face of the earth excepting by the method adopted by these men (breeders, nurserymen, etc.), viz, by the selection of congenital variations being produced as the result of (but without any direct adaptational relation to) a disturbance of the material of the reproductive particles of both sexes, that disturbance being increased if not determined by change of environment of the parental organisms or the coupling of remote strains.

Probably the most common way of obtaining initial variations is to select them from seedlings as they appear, but their advent can be greatly hastened by artificially changing the conditions under which the plants grow or by crossing different races or species. As soon as the desired variation makes its appearance its improvement and fixation by selection should begin.

In all cases where selection is the means adopted for transmitting a desired feature, large numbers of individuals should be grown in order to provide a greater scope for selection and thus increase the probabilities of securing the variation desired. The testing grounds and seed farina of large seed establishments are fruitful sources for variation. In such places a large number of individuals of the same race are grown together, and in the necessarily careful elimination of all variations from the type of the race (roguing) in order to insure the production of seed true to the type, attention is forcibly called to all variations, and if any of these promise to be desirable improvements, the intelligent seedsman almost invariably marks the plants showing such variations and preserves the seed for further trial and selection.

1 The degrees or "percentiles" here used indicate the variation above the mean average of any quality, which is considered as 50 degrees.

Another factor of great importance in selecting to improve races is the selection of individuals for breeding which exhibit in the most marked degree the feature to be improved. Galton has shown that the rate of racial changes which accompany different degrees of severity in selection call be determined with almost mathematical precision, and emphasizes the great necessity of using highly selected parents. As tile result of carefully conducted work, lie says: ''One generation of the 99-degree1 selection is seen to be more effective than two generations of the 90-degree selection, and to have about equal effect with the 80-degree selection carried on to perpetuity. Two generations of the 99-degree selection are more effective than four of the 95-degree and than a perpetuity of the 90-degree." Thus, in selecting wheat or any other plant to increase the productiveness, it is of the greatest importance that very many individuals grown under the same conditions should he examined and the seed taken only from those producing the largest yield.


The methods of selection pursued by certain growers of sea island cotton through many years are the most careful and painstaking known to the writer, and a description of these methods is given as an illustration of the actual procedure in continuous selection.

According to tradition and the reports of growers, sea island cotton when first introduced into this country from the West Indies was a perennial, unsuited to The duration of the seasons of the latitude of the sea islands off South Carolina and Georgia, where it seldom matured fruit. However, through the selection of seed from early-maturing individual plants and through better methods of culture there has been developed an improved race, which now seems to be thoroughly adapted to the conditions of growth in the region referred to. Furthermore, under the continuous and rigorous selection to which the plants have been subjected the fiber has been gradually improved (Pl. XXVI), and now that produced along the coast and on the islands lying off South Carolina and Georgia is considered superior to that grown in any other part of the world. The custom of carefully selecting the seed has grown with the industry and may be said to be inseparable from it, and it is only by such careful selection that the staple can be kept up to its present superior excellence. Several different strains have been developed and are maintained by different growers selecting with different ideals in view. The method described below is that which has been employed for many years by Mr. W. A. Clark, of Columbia, S. C., on his James Island plantation, and to him the writer is greatly indebted for the details. This method and similar ones employed by numerous other growers are applicable, with slight variations, to most of our common crops, such as corn, wheat, etc.


(1) The first selection is made in the general field, where there are a great number of individuals growing and consequently abundant opportunity for choice. Each plant in the field is somewhat hastily examined, special attention being given to the vigor and productiveness of the plant, the strength, silkiness, and general quality of the staple, etc., and a number of those which appear to be distinctly superior to the general crop are marked.

(2) The selected plants are then compared, and several of the best selected for more careful comparison, field notes on these being recorded and preserved for comparison with more critical notes to be taken later. The following is an illustration of field notes copied from Mr. Clark's selection notes of 1895: "No. 1, stalk medium, pod medium, bearing close, fairly double, lint fair; No. 2, stalk medium, pod medium, bearing good, lint fair; No. 3, stalk large, pod large, bearing close and double, lint fine and long; No. 4, stalk large, pod medium, bearing extra close and double, lint fine and long; No. 5, stalk medium, pod medium, bearing close and double, lint fine and long."

(3) Critical house examinations of the specially selected plants are now made at leisure, the fiber being "pulled" and carefully examined and graded according to (a) covering of seed; (b) size of seed; (c) length of staple; (d) fineness of staple; (e) uniformity in length (an important feature in preventing loss in manufacture). The following table, taken from notes made by Mr. Clark in 1895, the selections being based on critical pulling, shows how carefully the record is kept:

Results of examinations of specially selected plants.

Factors used in grading No. 1 No.2 No. 3 No. 4 No.5
Covering First Fifth Second Fourth Third
Size of seed Second Fifth Fourth Third First
Length Third Fifth Fourth First Second
Fineness Fourth Fifth Third First Second
Uniformity 1 1 1 1 1

NOTE.—All stalks present good appearance in field except No. 3, which is defective in middle of top.

"Valuing the first place as 5 and the fifth place as 1, and in like manner intervening positions, the general grade or rank of stalks, valuation of uniformity being omitted, as it is the same in all, would be as follows: No. 1, 14 points; No. 2, 4 points; No. 3, 11 points; No. 4, 15 points; No. 5, 16 points. Nos. 4 and 5 would therefore seem to rank first."

(4) The next step in the select ion is the comparison of the ginning quality, that is, the actual weight of lint to seed in the individual plants selected. The nearer the weight of the lint approaches the weight of the seed the better. In the early days of sea island cotton growing in the United States the proportion of lint to seed by weight stood about as 1 to 5, but under the influence of continuous selection the difference in the ratio has been gradually reduced until now it is frequently about as 1 to 3. In order to obtain values easy to compare, Mr. Clark weighs the seed and lint after ginning and determines the weight of unginned cotton necessary to produce a standard 300-pound bale. This is obtained by the following proportion: Weight of lint is to combined weight of lint and seed as 300 pounds is to X (X equaling the weight of the unginned product necessary). To illustrate this point, according to Mr. Clark's notes the five plants above referred to ranked as follows, as shown by their ginning qualities: No. 5 required 1,001 pounds to produce a 300-pound hale; No. 4, 1,001 pounds; No. , 1,038 pounds; No. 2, 1,060 pounds, and No. 3, 1,068 pounds.

(5) To secure further evidence as to the qualities of the selections and as a check on individual judgments, the ginned fiber is sent, labeled by number, to all expert cotton commission merchant known as a factor, who judges the sample from the standpoint of the expert marketer. Of the above plants, Mr. Clark's factor selected No. 5 as outranking the others. This number being on the whole superior to the others, it was finally selected and the seed retained for further breeding, the other numbers being discarded.


The seed of the individual plant selected the first year is planted in the spring of the second year, and as each cotton plant yields from 500 to 800 seeds, 500 or more seedlings will probably be produced. When these reach the proper stage of maturity, all are carefully examined, as in the preceding year's selections, and several chosen for further and more careful examination, These specially selected individuals of the second generation are put through the same careful tests as to covering, size of seed, length of staple, proportion of lint to seed, etc., as those of the first year's selection, and the plant found to be of particularly high grade is selected for further breeding. The seed from the remaining plants (about 500) resulting from the first year's selection are retained for planting the third year in order to obtain sufficient seed of a selected strain to plant the general crop.


The seed from the second year's selection is planted in the spring of the third year, and when the plants reach maturity each one is examined as in the first and second years, and an individual particularly good in all respects selected for further breeding, as in the previous years. The seed produced by the plants (some 500 individuals) resulting from the single plant selected the second year, and which are not specially selected for further breeding, are retained to plant in the spring of the fourth year in order to provide sufficient seed the fourth year to plant the general crop the fifth year. The seed from the 500 or more unselected plants of the second year's selections are grown this year, being sufficient to plant an area of 5 or 0 acres and furnishing enough seed to plant the general crop the fourth year.

Fig. 92.—Diagram illustrating method of selecting sea island cotton.


The seed from the specially selected plant of the third year are planted, and from the resulting 500 or more seedlings a particularly fine individual is again carefully selected for further breeding, as in the preceding years. The seed of the plants descending from the individual specially selected the third year is sown to obtain sufficient seed to plant the general crop of the fifth year. The seed used to plant the general crop of the fourth year is that from plants of the third year grown from the unselected plants of the second year, and thus the general crop this year is derived directly from the first-year selection, and so on through succeeding generations. (See fig. 92.)


The foregoing description and diagram show that after the selection work is under way special selections are made each year from the small plats of very select seed, and that the general crop is continually grown from stock descending from a single selected individual plant. In the method of selection above outlined the selection of a single individual each year only is considered. In practice, each grower generally selects several plants each year from which to breed: for example, two being selected for superior excellence of staple, one or two for general vigor and productiveness, etc. Each of these, however, is chosen from several selected individuals, the same care being exercised as in the first-year selection. It is always desirable to choose several special plants each year as breeders, as occasionally a selected plant may prove erratic and produce seedlings materially differing from the type, even after the selection has been carried on for a number of years with the same ideal in view.

Under this continuous painstaking selection the quality and length of the fiber has been gradually increased (Pl. XXVI) and the proportion of seed to lint gradually decreased. The fiber from unselected plants is only from 1 3/4 to 2 inches long, while that from the selected strain is about 2 1/2 inches long and is very strong and silky. The finest grades are used to adulterate silks. These high-bred strains are maintained only by continuous selection, and if for any reason the selection is interrupted, there is a general and rapid decline in the quality of the staple. The cotton produced by these rigidly selected plants commands a much higher price than the general crop and is sold direct to manufacturers for special purposes. The price of such cotton is governed entirely by the excellence of the crop, so no regular quotations for the product of the highly selected plants are given in trade journals. The finest grades from the selected plants, the writer is informed, sell for from 50 to 60 cents per pound, while the ordinary sea island cotton is quoted at from 15 to 30 cents per pound.

Different growers select with different ideals in view, and the crop of each plantation may differ greatly in quality and value from that of adjoining plantations. Mr. Clark selects mainly with a view to increasing the fineness and length of the staple, and this is done at the expense of quantity. His fine product, however, commands the very highest price, and this compensates for the small yield. Mr. W. G. Hinson, another careful grower of sea island cotton, selects with a different ideal in view and has produced a strain with somewhat coarser fiber, but yielding heavier; and although the coarser grade may not bring so much per pound, yet it may prove fully as remunerative because of the greater productiveness of the strain.


Gardeners believe that the maturity of the seed has considerable influence on the offspring, particularly as to time of ripening, plants grown from immature seed being said to ripen their fruit much earlier than those grown from mature seed. In 1885 Goodale and later Goff observed that certain of the early sorts of market vegetables indicate that they may have originated in this way. Arthur, who has given this subject careful attention, says: "Another feature of importance is the tendency to an increased earliness of ripening the fruit on plants raised from immature seeds. In the cumulative trials with tomatoes by Goff, the strain from green seed ripened from ten days to four weeks earlier in different years than the corresponding series from ripe seed." According to C. L. Allen, specialists on Long Island who give careful attention to growing cabbage for seed, always examine each plant carefully before cutting it when harvesting their stock seed, and "if the seed is of large size it is rejected, because they hold that such seed will make leaves instead of heads. Besides that, these men will not use seed until it is at least three years old, for the same reason." He states further that "gardeners with keen observation note the fact that the older melon, cucumber, and squash seeds are, without having lost their germinating power, the better, as the proportion of flesh to the seed is greater and the vines are more productive of fruit and less inclined to throw out branches." From the evidence at the writer's command it is not clear how great an influence the maturity of the seed may have in selection experiments for general features. From Arthur's experience, however, it is practically certain that while immature seed gives a tendency to earliness, its use commonly results in lessened vitality and smaller fruits, and therefore fully matured seed should commonly be used.


It is very desirable that plants for selection purposes be grown in a region well suited to the crop under consideration and especially one having conditions of soil and climate favorable to the development of the feature which the selections are intended to accelerate and render stable. According to Allen, cabbage, which is particularly sensitive to changed conditions of growth, furnishes a good illustration of the necessity of giving attention to these points. "All our improved varieties of cabbage have come from careful selections in different localities. We have our best early types from light soils, which are favorable for early growths, and our large, late varieties from heavy soils, which encourage continuous growth, consequently a larger head and one better adapted to wintering over." If it is desired to produce a bush beau from a twining or pole variety, the best place to conduct the selection experiment would be in some locality as far north as the sort will grow successfully, Burpee, Wood, and other seedsmen having observed that when seed pole beans for the trade are grown in the far North without poles, which is a common practice, they to a great extent lose their habit of sticking close to the poles. It must not be thought, however, that climatic conditions favorable to the best growth of the plant should always be secured, for in many cases exactly the opposite is desirable. In selecting with a view to obtaining a sort. suited to local conditions of soil or climate somewhat adverse to the best. growth of all existing sorts, the plants for selection must be grown in that location in order that they may be subjected to the adverse conditions, and those individuals selected which survive and prosper best.


The most experienced seedsmen and plant breeders claim that a clearly defined ideal of the type desired is of the greatest importance, as is also a rigid adherence to this type year after year in making the selections. Mr. W. W. Tracy, who has had extensive experience both in selecting seed-bearing plants to keep the variety true to type and in originating and improving new sorts by selection, says: "My success in seed breeding has always been in direct proportion to the clearness of my conception of the ideal I was striving to produce and the persistency with which I adhered to that ideal in my annual selection of breeding stock."

Before beginning a selection experiment, the variety from which the selection is to be made should be carefully studied and a definite ideal formed of a perfect type of the sort desired. Mr. Tracy described an interesting experiment in selecting corn, which forcibly illustrates the necessity of adhering to the type in such work:

The result of the work was that the sixth year I had on that same 5-acre field a crop over 50 per cent et which was within the limits of the variation established for breeding six years before when not one plant in a thousand came within the limit. * * * In the fourth or fifth year of this selection one lot showed a very remarkable ability to resist drought. The plants were fairly true to type, and this new quality was so desirable that the seed breeder was tempted to save his "breeders" from that lot, but resisted it, and saved them according to rule, but he also saved the ten best plants from the new departure and planted the seed in a lot by itself. The result was that only a very few of the plants showed the drought-resisting quality so noticeable the year before, and all of the ten blocks varied more and had a smaller proportion of plants true to type than had any single lot since the first year.


When a desirable variation of any race of plants is obtained by hybridization or by changed environment or otherwise it should be fixed, that is, it should be so stamped on the strain by selection as to render it hereditary and cause it to be produced trite through the seed. Many of our most valuable sorts of vegetables and agricultural plants are developed from individuals in the general crop which exhibit marked differences from the normal type of the race to which they belong and from other known races. The plants from the seeds of such individuals usually in greater part resemble the type of the original race, but by selecting seed through several generations from individuals which most nearly resemble the original variation, the changed characters maybe rendered hereditary and anew race created. In sorts propagated vegetatively by cuttings, stickers, slips, etc., such variations may be utilized without waiting for fixation, as in such cases simply the portions of the individual showing the variation are grown. The original cause of these fortuitous variations, or "seminal sports," as they are frequently called, is not definitely known, but many of them are probably chance hybrids of a first, second, or later generation, or cases of partial reversion to some ancestral type, while others may result from environmental conditions. It is difficult to race any connection between such marked variations and the conditions of environment, however, while they are exactly what would be expected to occur in the markedly unlike progeny of a hybrid. All the fine races and strains of tomatoes originated by A. W. Livingston, such as Acme, Paragon, etc., were selected from accidental variations, being simply variations found in large fields of growing tomatoes and improved and fixed into stable races by selection.

The smooth-seeded upland cotton Klondike is an interesting illustration of the fixation of seed races by selection. Some years ago Mr. W. A. Clark, whose methods of selection have been described, conceived the idea of producing a fitter grade of upland cotton suitable for the finer textiles. In view of the injurious effects of the saw gin upon a long-staple cotton and the difficulty of separating the lint from a tufted seed with the roller gin, he determined to produce by selection an upland cotton growing on a clean, black seed, which when once secured could be improved in length and quality of the staple by hybridization with the sea island cotton. In the ordinary sorts of upland cotton, smooth black seeds, similar to those of the sea island cotton (Pl. XXVII, S), are occasionally found mixed with the ordinary tufted or green seeds (Pl. XXVII, U U). Originally, certain upland sorts, such as Peterkin, had smooth seeds, and the production of such seeds in sorts commonly having tufted seeds may be due to hybridization of the ancestors of the plant with the sea island or some smooth-seeded sorts of the upland.

Mr. Clark selected at random and planted a quantity of smooth black seeds from the ordinary upland cotton, and the great majority of the resulting plants produced the ordinary tufted seed, but a few had mainly smooth black seed like those from which the plants were grown. Seeds were selected from the few plants which produced mainly smooth black seed, and were planted the second year. This season a much larger proportion of the plants produced smooth black seed, but still many produced the ordinary tufted seed. Seeds were again selected from the plants producing smooth seed and planted the third year, and soon through five generations, when the character was fully fixed, and all the plants came true, producing only the smooth black seed (Pl. XXVII, K K).

To avoid the introduction of any new disturbing elements in the fixation of hybrids it is usually necessary to inbreed or close fertilize them, and if this, together with careful selection, is carried on through several generations hybrids can usually be fixed so that they may be depended upon to reproduce themselves in the main true to seed, even under different conditions. Hybrids found to be sterile to their own pollen, which is not infrequently the case, should be fertilized with pollen from hybrids showing the same characteristics, and preferably front the same parents.

As generally understood, the so-called fixation of a seed race is simply the strengthening of the inherent stability of the individual so that it will impress its characteristics more strongly and surely on its offspring. It may well he asked whether this does not also necessarily include the character of more or less marked prepotency to its own pollen. Most species and natural varieties are more stable than cultivated races, principally because they are markedly prepotent to their own pollen. It seems quite probable that in some of our most stable cultivated races prepotency has also been developed to some extent. If, during selection, prepotency could be acquired, it would add greatly to the stability of the race.

After a race is fairly well fixed, as generally understood, it is probable that the character of prepotency could be acquired by growing plants of the race in close proximity to plants of nearly related races or strains and planting the seeds of each individual the second year in separate plats or rows and again close to plants of related races. By carefully examining the plants grown the second season it could probably be determined which individuals grown the first year were least. affected by crossing with the related races grown in connection with them, and in this way any tendency to prepotency detected. The second year the seed to be used for further planting should be selected only front the plats of seedlings resulting from plants showing this tendency to prepotency. Furthermore, the seed from such plats should be selected only front individuals which careful examination has shown to be true to the type of the race. By continuing such selection through several generations it is probable that the race could be rendered largely prepotent to its own pollen.

Selection experiments, as normally conducted, and the rigorous ''roguing" practiced by all good seed firms, have a tendency to produce prepotency, all individuals varying from the type (such variation being caused largely by hybridization) being in this way rejected. The difficulty in such cases, however, is that no attention is given to the unity of the individual so far as prepotency is concerned. Belt says: "Artificial selection is more rapid in its results, but less stable than that of nature, because the barriers that man raises to prevent intermingling of varieties are temporary and partial, whilst that which nature fixes when sterility arises is permanent and complete." By selection, man can render variations hereditary through the seed and establish new races as markedly different from each other in visible characters as different natural species, but unlike these they cross easily and thus are swamped immediately when abandoned by man's fostering care.

If some attention were given to securing prepotency in new maces, before their introduction to the trade, it is probable that there would be less complaint of lack of fixity of type. In sorts which present marked characteristic features of value even such an expensive way of securing stability might be justifiable, for by such care the cultivated race could possibly be made to approach natural species and varieties in stability.

Improvement of Sea Island cotton by selection: OO, Ordinary Sea Island cotton—original type from which selection was made: SS, Selected Sea Island Cotton. Seeds of Klondike, Sea Island, and ordinary Upland cotton: KK, Seeds of the Klondike, with and without lint: S, Seeds of Sea Island; UU, Tufted seeds of ordinary Upland. The variety from which the Klondike was developed, with and without lint.


The effect of cross fertilization on plants grown for selection is a factor seldom carefully considered. It is well recognized that the greatest source of variation among plants is the crossing of individuals, which, though very similar, always show slight differences, and doubtless imperceptibly differ in constitution and structure also.

In the fixation of hybrids inbreeding is apparently very necessary, but doubtless results in lessened vigor. In the case of the cotton selection described above, however, and in almost all similar cases of gradual improvement by selection, no attention is paid to the crossing of different individuals further than growing the selected seed by itself in a plat isolated from other plants of the same species, to prevent the selected individuals from crossing with the unselected. In cases of this sort, and in all cases of gradual improvement by selection of slightly superior individuals, it is probable that more is gained in vigor by allowing the free crossing of the different selected plants than is lost by the greater variation introduced thereby into the selected strain. Furthermore, in selecting cotton, as described above, it is impossible to decide at the time of flowering which of the 500 seedlings resulting from the single selected individual will prove the best, and it is obviously impracticable to carefully inbreed all the flowers on each plant, or even a small per cent of them. If ten to twenty plants having the best qualities could be selected from the 500 and bred together, it is probable, considering the greater vigor of cross-fertilized plants, that the result would be much better than that from the most careful inbreeding. The 500 seedlings resulting from the selected individual being planted together and allowed to cross freely, it is highly probable that some of the numerous seeds developed on the individual plant finally selected from them for further breeding will have been fertilized with pollen from some similar high-grade individual among the 500. In selection experiments of this nature, therefore, it seems desirable to plant the seedlings resulting from a single plant close together in a square plat, rather than in a single long row, as by so doing mixed cross fertilization is favored. It is of course highly desirable that plats of selected plants be some distance from the general crop to prevent crossing with unselected individuals.


1 Yearbook for 1897, Pls. XVIII and XIX and fig. 12.

Many horticulturists believe that selection has had the greatest influence in the development of the various races and sorts of cultivated plants, and some go so far as to assert that all other factors are of minor importance. The skillful plant breeder, however, takes all elements into consideration, in order to bring about the amelioration: desired, Both hybridization and selection have their definite and distinct places in every rational system of plant breeding. As explained above, hybridization and changing the environment artificially are the principal means of securing desired variations, and selection is the means by which a variation when once secured is augmented and fixed. When used alone in the improvement of plants, selection depends upon the adding up of small, unimportant variations through many generations, which in the end may possibly result in marvelous differences; but by this method the breeder has no way to force the change, and must be satisfied with slight variation and long-continued selection. However, when marked changes and new creations are desired, it is to hybridization or to chance sports that attention must be turned. In the words of Henri de Vilmorin, ''Cross breeding greatly increases the chance of wide variation, but it bakes the task of fixation more difficult. It, however, gives the raiser the only means in his possession to unite in one the qualities of two different plants while discarding their weak points. All the different qualities of the two parents seem to unite in the most varied coin combinations fiat ions in the crossbred products." It would hardly be possible to obtain in a lifetime by selection a markedly hardy orange or rose, a fragrant pansy, or a new creation like Burbank's hybrid walnut or raspberry-blackberry hybrid "Primus,"1 although it is just possible that such changes could be ultimately secured by this means. The most feasible and by far the quickest way to secure such decided variations and new creations is by hybridizing different species or sorts. Where it is desired to render a sort hardier, it should be crossed with a hardy relative, and where it is desired to render an odorless flower fragrant, it should be crossed with a scented related sort.


The improvements effected by selecting a variation may be slight in one generation, but, as before explained, if these slight improvements are continued year after year, very marked improvements may result in the course of time. Bailey says: ''It is the slow and patient care and selection day by day which permanently ameliorate and improve the vegetable world. Nature starts the work; man may complete it."

The origin of our various cultivated plants is doubtless due to the cumulative effects of more or less unconscious selection through centuries. The wild progenitors of important cultivated plants are in any instances unknown, and in but very few cases is there any knowledge of the early stages in their development. They came down to historic times in an advanced stage of development. In some instances the fact of the development of cultivated forms from wild plants has been proved by experiment. The experiments of Buckman in developing the wild parsnip, those of Louis de Vilmorin in developing the wild carrot, and those of Carrière in improving the wild radish have become classical. After several years of selection, Buckman developed from the wild parsnip an improved form which he called the Student. This was further improved by Messrs. Sutton Son, and was finally sent out. According to Henslow, "it still remains, after more than forty years, the best parsnip in the trade." The changes effected by Louis de Vilmorin in the wild carrot clearly how what can be accomplished in this manner, he sowed seed of wild plants, and found that the offspring flowered continually through the summer. By collecting seed from plants producing the latest flowers and sowing them late the following season he encouraged the enlargement of the root. In this way the carrot was induced to flower uniformly in the second year of growth, and hence is now a biennial instead of an annual, the acquired habit having become hereditary. The selection of wild radish or jointed charlock seeds, carried on for some time by Carrière, resulted in the production of several varieties of radish similar to those commonly cultivated.

A still more interesting case of development from cultivation and selection is described by Henri de Vilmorin:

I may relate here, in a few words, au unpublished experiment which I have been conducting for more than twenty years, from 1872 to the present year [1893]. it has consisted in cultivating one of our parsley-worts (Anthriscus sylvestris), an European weed, in order to change its slender and much-forked roots into fleshy, straight, and clean roots, say like those of the parsnip. Among the first batch of roots raised from wild seeds a dozen were selected with a tendency in their roots to larger and straighter bodies. Each root was planted separately and its seeds harvested separately. Of the dozen lots obtained eight or nine were discarded at once, and roots were selected only in such lots as exhibited some trace of variation. Again, a dozen roots or so were chosen (a drawing made of each root), which were afterward planted separately. I have sketches of all the roots selected, so that it is possible to follow all the stages of variation of each plant living at his day. For the first ten years the changes were slight, but now they are more and more marked with every generation, and in some of the lots the straight and smooth roots are the most numerous.



The improvement in the texture, shape, and flavor of the tomato wrought by selection is a good illustration of what can be accomplished in this is direction. Livingston, who originated so many of our improved sorts of tomatoes, depended entirely rely on selecting from large fields of growing tomatoes certain individuals which showed desirable variation and selecting from their progeny through several generations to improve and fit the variation. In describing the origin of the Perfection tomato, Livingston says: "I selected a plant from a field of Acme (a purple tomato) and secured what is known everywhere as Livingston's Perfection tomato (a blood-red tomato), which I introduced in 1880. The stalks and foliage are tighter than those of Paragon, but stronger than those of Acme." All his numerous new sorts of tomatoes were from accidental variations selected in this way, and the same can be said of Bailey's Ignotum tomato. The underlying cause of such accidental variations is unknown, but is doubtless the result of accidental hybridization, partial reversion to some ancestor, or to the variations known as ''seminal sports" (fortuitous variations), the production of which seen entirely accidental or depends upon causes not yet understood.


Increased size and productiveness are among the most common and important features resulting from selection. The increased length and quantity of fiber of the sea island cotton, previously described (Pl. XXVI), are good illustrations of this, and doubtless all common agricultural crops could be similarly improved. Allen cites an interesting ease of increased yield in corn as a result of selection, as follows: "Four years ago my foreman, at my earnest request, began the selection of field corn for seed purposes. He grew the white dent red-cob variety. Before harvesting the main crop he went over the field and selected the lowest-growing, stocky stalks, with two perfect ears each. He has followed the same plan ever since, with an increase of fully 25 per cent in productiveness."


Louis de Vilmorin's classical experiments in selection, which resulted in increasing the richness of sugar in the sugar beet, shows what exceedingly important results can be obtained by careful attention in selecting the seed-producing plants. These experiments in fact saved the beet-sugar industry of France and established it on a paying basis. His method consisted simply in testing the individual roots to determine their richness in sugar, and selecting for seed production, or "mothers," as they are termed, only those showing the largest percentage. In his early experiments the quantity of sugar was estimated by the specific gravity of the root, but later a more accurate method was devised, consisting in cutting out a small cylinder from each root and testing the richness of the juice by polarization. This method does not injure the roots for seed purposes and furnishes accurate data as to richness in sugar content on which to base the selections.

Louis de Vilmorin's careful methods have been continued by the firm of Vilmorin, Andrieux & Co., and by many others engaged in the production of sugar-beet seed, and upon these methods depends the success of the industry. The White Improved Vilmorin beet, which is Vilmorin's selected strain of the old White Silesian, has a worldwide reputation, and immense quantities of its seed are distributed annually. It has been the subject of careful and persistent selection for over thirty-five years, and its quality of richness in sugar has become fixed and constant. Mr. Ernest Clarke says that "some thousands of analyses have shown it to yield up to as high a proportion of sugar as 16 per cent of the weight of the roots, 1 3/4 pounds of sugar per gallon of juice being a very ordinary yield with it." Similar results wave been obtained from the same kind of selections of seed-producing plants of the Kleinwanzlebener and other races of sugar beets the following statement by Mr. Ernest Clarke in regard to selections made by a single firm will give an idea of the extent to which the election work is carried on: "Messrs. Rabbethge & Geisecke, the famous cultivators of the Kleinwanzlebener variety, who have been growing beets for seed for upwards of thirty years, state that in 1889-90 they tested 2,782,300 roots, from which they selected 3,043 roots for seed-growing purposes." The percentage of proteid matter in wheat, peas, etc., and of starch in potatoes and barley, etc., could doubtless be increased by similar methods of selection.


There is evidence to show that a decided change of form may be brought about by selection carried through numerous generations. The origin of the Blanche Ferry sweet pea, described by Tracy, is an interesting illustration of modifications of this sort:

Some forty years age a woman in northern New York noticed and saved the seed of a particularly bright-flowered plant of the old Painted Lady. She planted them in her garden and each succeeding year saved and planted seed of what she thought were the best plants. She did not raise many, some years not more than a dozen plants and never more than could be grown in 3 square yards. She was the wife of a quarryman and her garden was always over limestone ledges, where the soil, though fertile, was very thin, often not over a foot in depth, and gradually her plants became more compact and sturdy, until after some ten or twelve years she ceased to "bush" them, simply letting them support themselves. After she had raised them in this way for some twenty-five years a seedsman noticed their beauty, obtained about one hundred seeds, and from them has come the Blanche Ferry.

In many instances the form of flowers has been greatly modified by selection. Darwin says:

Williamson, after sowing during several years seeds of Anemone coronaria, found a plant with one additional petal. He sowed the seeds of this and by perseverance in the same course obtained several varieties with six or seven rows of petals. The single Scotch rose was doubled and yielded eight good varieties in nine or ten years. The Canterbury Bell (Campanula medium) was doubled by careful selection in four generations.


Many very valuable improvements in this direction testify to the importance of the results secured by selection. For instance, the decided shortening of the period required for sea island cotton to mature, as previously described, has fitted it for cultivation in certain portions of the United States, in which it is now all important crop.

The uniformity of heading or ripening of lettuce obtained in the forcing business is also, as the writer is informed by Mr. P. H. Dorsett, of the Division of Vegetable Physiology and Pathology, an interesting and valuable illustration of improvements of this nature obtained by selection. Careful growers of this crop, particularly in the vicinity of Boston, where the industry has reached its greatest perfection, always raise their own seed, claiming that it is impossible to purchase seed suitable for their requirements. In growing lettuce under glass it is of the utmost importance that all the plants be ready for cutting at one time, so that the house may be immediately reset with other plants, in this way constantly utilizing the available space. Several selected strains have been developed with this idea in view, and also with sonic reference to habit and vigor of plant, quality, shape, firmness of head, etc., and, as the writer has been informed, the plants have become so perfectly uniform as to time of maturing that frequently an entire house may be cut in the morning and replanted the same day.


Some doubt has always existed as to whether plants normally propagated vegetatively by cuttings, slips, buds, etc., could be improved or permanently modified by a selection of these parts from certain individuals or parts of individuals by methods similar to those used in seed-propagated plants. Each joint of a plant, with its bud and leaf attached, possesses in most cases the faculty of growing into a new plant much like the parent. These new plants vary the same under environmental influence as do individuals produced from seed, and while parts of the same seedling, still they are in a physiological sense distinct, each possessing individual characteristics and constitutions depending upon the conditions tinder which they were grown. As is well known, even branches on the same tree may differ in many characters, and while such variations are not commonly recognized when slight, yet in certain cases they become very marked, and are then easily distinguished, being known as "bud sports." Mr. B. T. Galloway says:

Every one who propagates plants by cuttings knows that hardly any two of them possess exactly the same characters. Starting with two rooted cuttings from the same plant and growing them under as nearly the same conditions as possible, one may give a plant that will bloom freely, forming flowers of large size, and its leaf development may also be perfect, while the other may be a vegetable runt, lacking in vigor of leaf and utterly unable to give anything but small and imperfect flowers.

Bailey also cites cases showing the change produced by different climatic conditions on different individuals produced by budding:

We know, too, that the same variety of fruit tree takes on different characters in different geographical regions, so that in the West and South the Greening apple is no longer the Greening of Rhode Island. So it is apparent that even when we divide a plant into many parts and distribute the members far and wide, and when there is no occasion for concerning ourselves with fixing the type—even here there is variation.

Darwin says:

Mr. Salter brings the principle of selection to bear on variegated plants propagated by buds, and has thus greatly improved and fixed several varieties. He informs me that at first a branch often produced variegated leaves on one side alone, and that the leaves are marked only with an irregular edging or with a few lines of white and yellow. To improve and fix such varieties he finds it necessary to encourage the buds at the bases of the most distinctly marked leaves and to propagate from them alone. By following with perseverance this plan during three or four successive seasons a distinct and fixed variety can generally he secured.

For several years Messrs. B. T. Galloway and P. H. Dorsett, of the Division of Vegetable Physiology and Pathology, have been carefully selecting violet cuttings to determine to what extent the plants can by this means be improved in productiveness, vigor, and ability to resist disease, etc. The results already obtained show that productiveness is remarkably increased, and they also clearly demonstrate that violets can be gradually improved by a continuous selection of the cuttings used in propagation and of the plants from which these are obtained. The method consists in selecting a number of the finest-looking plants before they begin to bloom, placing beside each a stake to which a blank tag is attached, and carefully recording on each tag the daily pick of salable flowers from the plant, so that at the end of the season the number of flowers produced by each plant is known. The cuttings for the ensuing year are taken only from the plants producing the greatest yields and which are known from continual observation through the season to be desirable in other ways. The pedigree cuttings thus obtained are again subjected to selection, and only those which root well and form good, vigorous young plants are finally used (figs. 93 and 94).

Fig. 93.—Selected rooted cutting of Lady Hume Campbell violet 
(from a photograph by P. H. Dorsett).
Fig. 94.—Unselected rooted cuttings of Lady Hume Campbell violet
(from a photograph by P. H. Dorsett).

Mr. Dorsett has recently described some very striking results obtained in 1897 by growing comparison beds of selected and unselected cuttings of the Lady Campbell violet. In this experiment three beds were compared, the first containing 356 selected plants, the second 516 unselected plants, and the third 352 selected plants, and all received the same treatment throughout the season. In beds Nos. 1 and 3, containing selected plants, comparatively few of the plants had to be reset during the summer and fall, but in bed No. 2, containing unselected plants, fully 15 per cent were replaced. Besides this, the selected plants made a more vigorous and a healthier growth and produced a greater number of better-colored and longer-stemmed flowers than the unselected plants. The following is a record of the production of each bed during the season: Bed No. 1, 30,000 flowers, average per plant 84.2; bed No. 2, 31,200 flowers, average per plant 60.4; bed No. 3, 30,980 flowers, average per plant 88.

It will be seen by a comparison of the above data that ''the average yield from the beds of selected plants is greater in both cases than from the bed of unselected plants, the plants in bed No. I producing on an average 23.8 and those in bed No. 3, 27.6 more flowers per plant than were produced in bed No. 2, a combined average of 86.1 from the selected plants as against 60.4 from the unselected, or a gain of 25.7 flowers per Plant, and this with practically no additional expense."

The following table, from Mr. Dorsett's paper, gives the yield per month during the season of five selected plants from a parent which the previous season yielded eighty-five flowers:

Yield per month of five selected violet plants.

Number of flowers produced in—
October November December January February March April Total
36 a 2 6 5 16 10 16 26 84
38 b 7 12 4 7 23 26 27 109
38 c 1 13 6 18 27 23 33 127
38 d 3 9 4 9 23 26 27 103
38 e 3 7 2 11 13 16 30 82

"The table shows that three of the five plants gave a much greater yield than the parent. The average of the five plants is 101, or sixteen flowers more than the original plant produced. The average yield is thirteen flowers more than that given by Mr. Galloway from ten selected plants during the previous season."

In some selection experiments with strawberry cuttings Goff obtained very beneficial results by taking cuttings from plants which had not yet been weakened by fruiting, and the ancestors of which had been free from leaf blight (Sphaerella fragariae). Cuttings taken from plants which had been weakened by leaf blight were tender, and many of them were winterkilled.

In the case of the so-called Ripley spike, a disease which attacks the Ripley Queen pineapple and causes it to "go blind," that is, advance to the end of its growing period and sucker from below without fruiting, the writer observed very marked results from selecting suckers from healthy plants. In September, 1894, a careful grower in south Florida took a number of suckers from diseased plants and planted them in a bed by themselves, and in an adjoining bed planted a number of suckers taken from apparently healthy plants. In March, 1896, the writer examined the plants in each bed, and found that in the bed of 348 plants grown from suckers taken from diseased plants 219 of them, or about 63 per cent, had contracted the disease, while in the bed of 244 plants grown from suckers taken from healthy plants only 9 of them, or slightly less than 4 per cent, showed the disease. This result from but one selection indicates that the disease could probably he entirely controlled by a continuous selection of stickers from healthy fruiting plants.

Prof. L. C. Corbett, of the West Virginia Agricultural Experiment Station, informs the writer that he is succeeding to a great degree in breeding out the tendency in certain roses to produce blind branches by continuously selecting cuttings from flower-producing branches.

Another interesting illustration of modifications obtained by the selection of vegetative parts is the breeding out of thorns front citrus trees by bud selection. Seedling oranges and lemons are almost invariably very thorny, but nevertheless the majority of the standard varieties cultivated are now largely thornless, owing, it is said, to the continuous selection of buds from thornless branches. According to the testimony of orange nurserymen, it is quite certain that thorns can be bred out in this way in every case, and usually to do so requires but three or four bud generations. It is probable in the case of other fruit trees that by selecting buds or cuttings from branches that are thornless, or which have fewer thorns than usual, that the thorns could be entirely bred out, or at least the number greatly reduced.

Sufficient examples have been given to show conclusively that selection may play a very important part in the improvement of plants propagated by vegetative parts. Doubtless extremely valuable improvements, particularly increased productiveness and resistance to disease, can be secured by a careful selection of slips, cuttings, buds, etc. Equally as careful attention should be given to the selection of the mother plants from which these parts are taken for reproduction as is given to the selection of the plants used for seed production.


In the preceding pages attention has been directed to some exceedingly valuable results obtained by careful selection methods, for instance, the increased productiveness of cotton, corn, sugar beets, etc. The common methods of select ion are simple and inexpensive and should become general practices in agriculture. Every farmer and horticulturist should devise for each crop a systematic method of selection similar to that described in the case of sea island cotton, so that the general crop maybe grown continually from selected pedigree stock. All common agricultural crops respond to skillful selection, and in every case valuable results will doubtless reward the agriculturist's attention to this principle.