1. The number of seminal roots varies significantly with position of the seed on the ear and with differences in temperature and moisture during germination. 2. Each ear has relatively the same average number of seminal roots under different conditions. 3. The average number of seminal roots per ear is constant within the limits of sampling error under uniform conditions of germination. Samples of 20 seeds each were found to be adequate for determining the number of seminal roots under uniform conditions. 4. Number of seminal roots proved to be independent of length, diameter, or shape of ear, number of rows of grain, and length, width, degree of denting, appearance, or specific gravity of the seed. Ears with no discoloration of the shank had a slightly higher number of seminal roots than ears with discolored shanks. 5. No correlation was found between seminal roots and yield in four tests. A barely significant correlation of 0.21 was found in a fifth test. Coefficients of determination indicate that from 0.76 to 2.29% of the variation in yield is accounted for by variation in number of seminal roots. 6. Number of seminal roots was positively and significantly correlated with vigor of seedlings when the seeds were germinated between blotters and transplanted to soil, but negatively and significantly correlated when the seeds were planted directly in soil. 7. Number of seminal roots proved to be independent of date of silking, height of plant, height of ear, number of tillers, percentage of fallen plants, and percentage of smut-infected plants. A slight association between number of seminal roots and number of nodes of the progeny plants was indicated. 8. Five generations of self-pollination in four strains resulted in a marked and progressive reduction in number of seminal roots. 9. It is concluded that number of seminal roots is of no value as a criterion for selection of productive seed ears under the conditions represented by this experiment.

A genetic study was made of the F3 progenies of a cross of pure lines of Odessa X Sevier wheats. The characters in which these parents were contrasted and upon which a study was made included (a) awns, (b) spike density, (c) grain color, (d) length of culm, (e) number of culms per plant, and (f) number of spikelets per spike. From this cross three classes of awns were obtained, viz., (a) a homozygous group resembling the Odessa parent, (b) a homozygous group resembling the Sevier parent, and (c) a heterozygous group segregating for awn classes. In the homozygous hybrids there is a greater variation than is found in the parents. In both groups of true-breeding strains there is a tendency for the awns to be longer than those of the corresponding parental group. A good 1:2:1 ratio was obtained in awn segregation. There was a 1:2:1 ratio in spike density segregation. In the two homozygous groups there were found spikes considerably more compact than the spikes of either parent and some considerably more lax than was found in either parent. The true-breeding dense hybrids have, on the whole, shorter internode lengths than either parent, and the true-breeding lax hybrid had, on the whole, longer internode lengths than has the lax parent. In a few cases both the dense and the lax parental spike forms were recovered. Although a poor F3 ratio was obtained in grain color segregation, observations in F4 and F5 indicate that inheritance of this character is probably due to a one-factor difference. No genetical data for chaff color were obtained because of the similarity of the chaff color in the parents. Black markings occurring on the Odessa parent and on some of the progeny were too strongly influenced by environment to permit a determination of the ratios. There was little difference in the mean length of culms for the parents. The mean length for the hybrids as a group was slightly longer than the mean for the taller parent. There was considerably greater variation in the mean length of the hybrid families than there was in the means of the parental families. The variation reached both above and below the extremes of the parents. Too few parental rows decrease the weight of conclusions on culm length, number of culms, and number of spikelets. The average number of culms per plant in the hybrid was between the averages for the two parents. The average of the hybrids, however, was considerably toward the average of the parent bearing the greater number of culms. The lowest number of spikelets per spike in the hybrid is fewer than the least number in the parent with the lower average. The greatest number of spikelets per spike in the progeny is in excess of the greatest number of the high parent. The average number of spikelets per spike in the hybrids is between the average of the parents but is very near the average of the high spikelet bearing parent. None of these differences are statistically significant. There is perhaps segregation for the number of spikelets, but its nature was not determined.

The bulked-population method of handling cereal hybrids consists essentially of creating populations by hybridization, growing the hybrids in bulk for six or eight generations until they have become homozygous or nearly so, and then making head or plant selections for comparative testing in the usual way. Nineteen crosses were handled by this method in an experiment at University Farm, Davis, Calif., in all or part of the years from 1923 to 1926, inclusive. Two generations were grown in each year. Head selections were made in 1926 and were grown in head rows in 1927. The best head rows were sown in replicated 16-foot triple rows in 1928. The average yields of 33 of the 45 selections grown in 1928, or 73.3% of the total number, were above the average yield of all (33) check rows. As a group the selections showed marked resistance to lodging and shattering. Hard-kerneled types predominated. The number of generations required before selection depends on the number of character differences involved. Ordinarily, seven or eight generations are sufficient. The area used for a bulk population should be large enough so that, at the rate of seeding employed, all combinations expected in a cross maybe included. On the average, wheat contains 10,500 kernels per pound. From 1 to 2 pounds should be sown when dealing with crosses of ordinary complexity. The method is adapted for the development of strains possessing such characters as winterhardiness, rust resistance, smut resistance, etc., in the close-fertilized cereals.

The writer's belief concerning pasture improvement may be summarized thus: 1. Cleaning up the pasture is the first big job on the majority of farms. 2. Apply some manure on pastures where the hauling is reasonable and the pasture not too rough. 3. Improve with lime and superphosphate. For the majority of pastures the first two recommendations are the horses and number three the cart. Let us not get the cart before the horses.

Experiments have been conducted to determine the value of supplementing the legume bacteria which the soil naturally supports with an artificial culture for alfalfa, red clover, beans, and peas. Limed and unlimed field plats, representing only one type of soil, have been used. Yields from one season only are presented. A few points may be emphasized. Alfalfa which was grown on limed soil produced about 11% more dry weight when supplementary bacteria were applied at seedtime, although roots 46 days old failed to indicate any value of the extra bacteria as measured by the number of nodules. Supplementing the legume bacteria on the unlimed soil which had a pH of 5.3 produced an increase over the plats not receiving such a treatment not only in the number of nodules and in the dry weight of the crop, the latter being equal to 18%, but also in the number of plants that survived throughout the season, this being equal to several hundred thousand plants per acre. Red clover which was grown on limed and unlimed soils produced 39.9% and 32.2% more dry weight, respectively, when supplementary bacteria were applied at seedtime. Red kidney beans which were grown on the limed plats showed an increase of 12.8% in oven-dry crop and 9.2% in shelled beans per acre in favor of supplementary legume bacteria. On the acid plats there was a slight gain of total dry crop and a slight decrease in shelled beans where the artificial culture was used. An average of both total crop and of shelled beans from the six plats receiving the supplementary bacteria compared with that from the six not receiving such treatment shows an increase of only 9.2% in total crop and 4.98% in shelled beans in favor of supplementary bacteria. Peas which were grown on both limed and unlimed plats had a larger quantity of nodules on their roots than plants from similar plats which did not receive supplementary bacteria. The total crop and the dried peas from these plats were also considerably larger, being 15.1% and 35.8% greater in total crop and 14% and 25.7% greater in dried peas, respectively, than those from similar plats not receiving supplementary bacteria. The total yield from the six plats which received supplementary bacteria was 14.3% larger than the total yields from the six plats not so treated. In dried peas it was 29.3% in favor of supplementary bacteria.

Experiments with kudzu regarding the effect of number of cutting treatments on yields and on the production of reserve foods in the roots brought out the following facts: 1. The fewer the number of cuttings the greater is the production of root reserves. The roots of plants receiving six cuttings per season decreased in weight during a period of two years, those from plants receiving four cuttings increased about 150%, those from plants receiving two cuttings increased approximately 400%, and those from plants receiving one cutting increased about 1,250%. 2. Yields of top were found to be dependent on the amount of reserve food stored in the roots. The greater the amount of root storage, the greater is the yield of top. 3. The greatest yield of tops was secured with the two-cutting treatment when the roots were of equal size at the beginning of the experiment. Due to the greater amount of storage material formed in the roots during the first year from plants cut only once, however, the greatest yield secured the second year was from the one-cutting treatment. 4. The percentage of reserve starch and nitrogen was found to be less than one-half as much in the roots from plants receiving six cuttings as in the roots of plants receiving four or a less number of cuttings. The percentage of total sugars, however, was found to be greater. This is taken to indicate that a change from starch to sugar is taking place in the roots of plants receiving six cuttings in order to produce new top growth. A study of the changes during the season in the reserve food storage of the roots of plants receiving no cuttings and the effect of late cutting on these reserves gave the following results: 1. The percentage of sugar did not change materially during the season. 2. The percentage of nitrogen was high at the beginning of the season and gradually decreased, while growth was rapid until sometime in August. Then, it again increased until the end of the season. 3. The percentage of starch and dextrins remained very constant until sometime in September and then increased very markedly from about 30 to 45%, calculated as dextrose. 4. Cutting the tops as late as September 5 caused a marked reduction in the percentage of starch and dextrins and an increase in the percentage of sugars and water found in the roots for the remainder of the season. Studies made in order to determine the effects of planting various sized roots on the vigor of the plants showed that the top growth from large roots was much more rapid than from small roots. This was taken as additional evidence that top growth is dependent to a large extent on the reserve storage materials in the roots. The practical field applications of these results are discussed.