PLANT-BREEDING AT CANTERBURY AGRICULTURAL COLLEGE.

New Zealand Journal of Agriculture, Volume XXXVI, Issue 3, 20 March 1928, Page 156

PLANT-BREEDING AT CANTERBURY AGRICULTURAL COLLEGE.

WORK ON CEREALS, GRASSES, AND RED CLOVER.

F. W. Hilgendorf,

M.A., D.Sc. ; F.N.Z.lnst., Canterbury Agricultural College

I. HISTORICAL OUTLINE. By the year iB6O wheat-growing was an established industry in Canterbury, and the area devoted to this crop expanded. continuously until 1880, when the average annual area amounted to 300,000 acres. During the earlier part of this period very numerous varieties of wheat were introduced, but by 1870 the commonly used varieties had become narrowed down to Tuscan, Hunter’s, and Pearl, and this has governed the millers’ and merchants’ classification of New Zealand wheats from that time to the present day. Individual farmers, of ' course, still made occasional importations, but none of these (except Solidstraw. Tuscan) was able to establish itself, largely owing to the fact that the old varieties had become wind-resistant by natural selection, while new introductions were badly shaken by the nor’westers.

During all this period there was no attempt at improvement of seed. New strains of the old varieties could not be introduced from England because they had gone out of cultivation there, so that even the names were lost, and there was no idea of systematic improvement from within. An occasional farmer rogued a few acres for seed, and a fairly clean crop was eagerly bought up by neighbours, but on the whole the different varieties were extremely mixed and impure. In 1909 an average crop contained 20 per cent, of impurities, and in many cases the wheats in a field were so mixed that it was impossible to guess which variety the farmer had intended to sow.

In 1910 Mr. R. E. Alexander, present Director of this College, suggested that an. attempt be made to improve the wheats of Canterbury by the method of pure line selection, which had proved so successful with barley in Ireland, and the work was entrusted to the present writer. It was carried on in a small way out of the slender resources of the College, but in 1915 the first success had been achieved and the pure line known as College Hunter’s had been distributed. In 1920 the Government, through the Department of Agriculture, provided for the continuance and extension of the work by making an annual grant of £5OO. Selection of oats was then undertaken in addition to further work on wheat. In 1921 the limit of improvement by selection among our New Zealand wheats appeared to have been reached, and crossing among pure strains was commenced.

In 1923 an extensive series of crosses was made between New Zealand wheats and the most promising of those from other parts of the world, and at the same time a very modest start was made in the selection of grasses. This work looked so promising that in 1925 the Department of Agriculture increased its annual grant to £l,OOO, and an assistant, Mr. J. W. Calder, was appointed to help in the work.

Selections of cocksfoot and rye-grass were made, not from the vicinity of the College alone, but from all the chief sown-grass areas

throughout the South Island, from Nelson to Invercargill. In 1927 a start was made with selections of red clover, and in the same year Parliament. made a statutory grant towards the experimental work at the College, the share of the plant-breeding department coming to about /i,600 a year. ■

11. SELECTION WORK IN CEREALS. (a) Methods. A factor that has influenced the methods used is the kind and area of ground on the College farm. This yields an average of 45 bushels of wheat per acre. It is almost all absolutely flat, extraordinarily even in chemical and mechanical composition, and perfectly drained. It is divided into fields of about 25 acres in extent, and between 400 and 500 acres are under the plough each year. Any one of a dozen fields is available for plant-breeding plots, and there is always available a field that has borne a crop of rape or Italian rye-grass in the preceding season. Thus there is ample space available, the same ground has not to be used more than once in ten years, unevenness of soil is reduced to a minimum, and self-sown cereals do not occur among the plots. Owing to the size of the farm and the presence of working students at the College, both horse, and hand labour are available whenever they are wanted, so that the facilities for breeding-work are probably as good as at any place in the world.

The unrestricted area of suitable ground has had a considerable effect on the choice of methods of trial. Single-ear selections are made from commercial crops, and each year and of each variety about one hundred head - to - row plots are sown in a permanent “ bird - cage ” enclosure. Elaborate trials have been made throughout the years to find which arrangement of plots would give the smallest probable error, whether chess-boards, or rows singly or in groups, and with replications up to ten in number. The result has been to abandon the expectation of any yield results from the bird-cage, and to rely on various forms of trial in the field.

About 25 per cent, of the rows are rejected on appearance or handling. Rows are eliminated if they show weakness of straw or too great length of straw, but especially if they show looseness of grain in the chaff, this character leading to shaking of the grain in the heavy hot winds that are common during the ripening season. Disease has little significance here. Rows are cut by sickle, and the bundles hung in a large shed until threshing. The thresher used consists of an old peg-drum machine from which the concave has been extracted, and on to the pegs of whose drum wooden beaters have been tied. Each sheaf is put in a cotton clover or timothy seed bag, and its heads held in the bag, against the beaters. When it is threshed the straw is thrown away, and grain and chaff emptied into a wire-gauze riddle,. whence the chaff is blown away by an electric fan. The method is not as quick as some others, but it is cheap and efficient, and the maintenance of purity of the grain is absolute. Another 20 per cent, of the strains are rejected on grain sample.

In the second year the seed is sown in the open field, areas being carefully chosen for freedom from feerings or finishes in the last or any

traceable preceding ploughing. The strains are sown in three-row plots, each row about 15 ft. (or a. rod) in length ; the three-row plots of each strain are replicated three, four, or five times, as the seed permits. Every fifth three-row plot is a check. A few strains are rejected on sight or handling ; the middle row of each of the survivors is harvested separately, and its two outside rows together. At first only the middle rows are threshed, in bags as before; then the seed is weighed and examined, and the worst yielders and those of worst quality are eliminated.

Certain strains remain, usually about thirty out of the hundred heads originally selected. Of these thirty, the sheaves made up of their two outside rows are now threshed, and all the seed of the same strain is bulked, dressed, and pickled ready for sowing.

The strains remain only one year in the bird-cage and one year in the rod-row plots. After that all trials are by field methods. Beaven’s* half-drill-strip method is employed for testing strains against the parent variety, because these nearly always sow at the same rate. A fifteencoulter drill is used ; the middle coulter is blocked, the seed-box divided into two by a partition, and one-half receives the seed of the strain and the other half that of the parent variety. The drill is driven wheel on wheel mark, and the result is that one has pairs of plots separated by 14-in. spaces (our drill coulters are in. apart), the number of pairs being limited only by the amount of seed available, or by the number of times that one chooses to drive up and down the field. The plots are drilled obliquely (usually at 45°) to the last ploughing, which means, in rectangular fields,, obliquely to all preceding ploughings as well.

The plots are about 49 in. wide, and they are cut by a binder having a 54-in. knife. This allows perfectly clean cutting, and each row of sheaves as it lies on the ground represents a plot. Threshing is done in the field by a specially adapted locally made threshing-machine. This will thresh up to 50 bushels of wheat per hour, and the six men that are used for threshing can, when they become skilled, clean the machine out to the last grain in seven minutes. All the plots of one strain are threshed in succession, and weighings (but not cleanings out) take place after threshing the last sheaf of each plot. The yields are compared by Student’s method.

The amount of seed available from the three-row rod-long plots replicated four or five times is about 4 lb., and this sows four to five strips 2 chains to 2| chains long. That this method can be used with so small a quantity of seed is due, first, to the large area at our disposal, and, second, to the structure of colonial drills, which will sow to the last grain, and which can be completely cleaned out in a few minutes. The small number of replications made with the seed available does not, of course, allow of the best strains being selected, but it allows an elimination of a few of the worst.

Next year the same method is adopted, but an unlimited number of replications is possible. It has been found that 3 chains is a suitable length of plot, and that twenty replications give a result that makes a difference of 2 per cent, in the yield that is, 1 bushel per acre —, significant. This number of plots is therefore sown, and the resulting three or four best strains are selected.

These few strains are grown by the half-drill-strip method for three or four more years ; one is finally selected as the best available from the original hundred selections, and if it has for four or five years given a significant increase over commercial seed it is considered good enough to distribute to farmers.

The seed produced from the -drill strips is unavoidably contaminated by seed from adjacent plots, and therefore this seed, while good enough for yield trials, is not good enough to sell as a. pure strain. The pure seed is obtained by the following device: Just before the first half-drill-strip harvest —that is, before the strain has ever been touched by binding or threshing machinery a few hundred heads of each strain are gathered and stored, it may be for one or two years. When it becomes obvious which strains are likely to be in the running for the final selection, the hand-selected sheaves of these strains are threshed and sown in a solid block well isolated from other strains. Each of these blocks is harvested and threshed with special care, and grown in a solid block again next year. Thus, when the final selection has been made and tested for a sufficient number of years, there is somewhere else on the farm a held of 20 or 25 acres of wheat of the same strain in a high state of purity and ready for immediate distribution. SUMMARY OF METHODS. First year: One hundred head-to-row plots in bird-cage. Threshed in bags. Second year : Three-row plots hand-sown in held ; three to five replicates; rows rod-long. Comparison by middle rows. Threshed in bags. Third year : Field methods ; half-drill strips, three or four replicates of blocks six coulters wide by 2 or 3 chains long. Threshed by machine. (Before harvest, hand selection of heads to provide pure seed.) Fourth year: Twenty replicates of half-drill-strip plots about 3 chains long. After threshing, strains reduced to three or four. Fifth year : Twenty replicates of the three or four strains, which, are thus reduced to two or one. (Hand-selected heads of the three or four strains threshed and sown.) Sixth year : Twenty replicates of two or one strains, allowing final selection. (Pure seed of two or one sown for multiplication). Seventh year : Twenty replicates of final selection for confirmation, (Pure seed of that selection sown in a field, giving about 1,000 bushelsfor distribution.) (b) Results to Date. • (l) WHEAT. With wheat in the condition described earlier (page 156), plantbreeding at the College had every chance of easy success. Following is a brief account of the several varieties selected • 1915: College Hunter s.—A pure line of the variety locally known as Hunter’s was distributed. after what we should now consider very, imperfect trials. For three years it had yielded an average of 9 bushels per acre over bought seed. Owing to its purity as compared. with the then available seed, it was favourably' received by farmers, and the

accidental circumstance of its having a striking and beautiful colour was a fortunate advertisement of the beginning of our plant-breeding work. Its yield under all trials in various localities showed that it was probably 4 bushels per acre better than unselected Hunter’s, which it rapidly replaced. In 1916 there were nineteen fields of Hunter’s on the road between the College and Christchurch ; two of these were the College strain. In 1917 there were twenty-two crops along the same road, and twenty of these were the pure strain. By 1918 the unselected seed had entirely disappeared. Twenty per cent, of the wheat grown in New Zealand is College Hunter’s, the average yearly area being about 50,000 acres. This' success, coming so early in the history of the trials, was an important factor in the public interest and support of the work of plant-breeding in New Zealand.

igiB: College Solid-straw Tuscan. This is a pure line of the variety locally called Solid-straw Tuscan. It yielded during five years •of trials about 3 bushels per acre better than bought seed, but on other farms more adapted to this particular variety its superiority was more pronounced. Owing to its lack of any distinguishing feature, its reputation among farmers has never equalled that of College Hunter’s, but it gradually permeated all Tuscan areas ; and it is now certain that practically all the Solid-straw Tuscan in New Zealand is descended from this particular selection. “ College Tuscan ” probably accounts for 60 per cent, of the wheat grown in New Zealand —say, about 150,000 acres annually. 1918: College Purple-straw Tuscan. — A selection from a mixed crop of White-straw Tuscan, this wheat had very strong straw and a fair-quality grain. However, it filled no special niche in Canterbury farming, and could not stand the competition of Solid-straw Tuscan. It therefore was used for only two or three years, and then went out of cultivation. 1918 : College Pearl.— A selection from the variety locally known as Pearl. The strain was about 2 bushels per acre better than commercial seed, and the quality of the grain was very good and even. The wheat shook badly, however, and was soon replaced by the strain next recorded. It is now nearly extinct. 1920 : College Velvet. — A selection from the variety locally known as Velvet. The strain yielded only 1 bushel per acre better than the commercial variety, and this was the best strain procured after ten years’ trials. It was therefore decided to distribute it, because Velvet was a variety much desired by millers and the then available seed was very impure. Apparently the strain filled a want, and all the Velvet grown in Canterbury is now. descended from it. ' Its wind-resistance is good for the variety, and its milling-quality is the best among all New Zealand wheats. College Velvet now accounts for 10 per cent, of the wheat grown in New Zealand-say, about 25,000 acres.

(2) OATS.

1923 : College Algerians.— This is a selection from Algerian seed, originally brought from Australia, which had undergone natural selection in New Zealand for 'about fifteen years. The strain tillered well, recovered quickly after feeding off with, sheep, gave a grain with a somewhat reduced proportion of husk, and during five years of trial at the

College beat commercial seed by n bushels per acre. It was an immediate success, increased yields of 20 bushels per acre being not infrequently recorded. By 1925 the ordinary market reports quoted College Algerians at 6d. a bushel above unselected seed, and two Christchurch firms between them sold 20,200 bushels of the seed, , of which 3,000 bushels went to the North Island. Practically all the Algerian oats now grown in New Zealand are descended from this strain. It is grown in the South Island for threshing, for chaff, and for green feed for sheep in autumn and winter, and in the North Island for chaff .and green feed. There is also a certain export trade in the seed to Tasmania and other Australian States.

2925 : . College Danish. A selection from a yellow-skinned oat locally known as Danish. The strain yielded 3-5 bushels per acre better than commercial seed, but it never attained any popularity; the variety had largely gone out of cultivation, as it has no merits different from those of Garton's Abundance, the standard white oat of the country. College Danish is grown in only a few isolated localities, and will doubtless soon die out.

1925 : College Duns.— A selection from . the variety known locally as Dun —probably the English Winter Grey. The strain is of good quality, and outyielded commercial seed by 4-4 bushels per acre during four years of trial. The variety, however, is not very widely grown, and the strain is infested with or sports into a fatuoid impurity that has checked its free distribution. A purified- line is now in process of multiplication. (c) Detailed Results for Harvest of 1927. Most of the work done during the season consisted of variety trials which this article does not record —or of crossbred trials, which are recorded under Section 111. A few strains were under trial as follows : (1) bell’s selection from college hunter’s wheat versus college hunter’s (see page 159). John Bell was a student at the College in 1920. In a crop of College Hunter’s he noticed one plant whose general characters were those of the main crop, but whose heads were noticeably more dense. The plant was saved, threshed, and grown as follows : 1921- : A single row for observation. 1922- : A small multiplication plot. 1923- In field—Five plots of about acre each; total weight of Bell, 238 lb. ; total weight of Hunter’s, 236 lb. In small plots—a ten-replicate chess-board Bell better than Hunter’s by 12 per cent. Odds in favour of significance, thousands to 1. 1924- :In field Twelve replicates of half-drill-strip plots. Bell 7 per cent, better than Hunter’s. Odds in favour of significance, 117 to 1. 1925- : In field Fourteen replicates of .half-drill-strip plots College Hunter’s, 50 bushels per acre ; Bell, 54-6 bushels per acre; advantage in favour of Bell, 9-2 per cent.; odds in favour of significance, 2,000 to 1. In small —Eight replicates of three-row plots; increase over College Hunter’s, 13-9 per cent.; odds in favour, thousands to 1. In small plots—Single rows, Hunter’s and Bell alternating sixteen times; increase over Hunter’s, 6-5 per cent.; odds in favour, 30 to 1.

1926-27 :In field-Twenty-three replicates of half-drill strips. Increase over Hunter’s, it bushels per acre ; odds in favour of significance 103 to 1. The figures of this last trial are shown in Table 1.

It will be seen that on the whole series of years Bell has outyielded College Hunter’s.

(2) ALGERIAN OATS — PURE LINE B. 49 IN COMPETITION WITH COLLEGE Algerians (see page 160). The strain was originated from a head picked at random from a commercial crop in 1920. 1920- : One of the best ten out of one hundred. 1921- : Small plots—One of the best three of the above ten. 1922- :No trial —multiplication only. 1923- : Five replicate plots, each of about 3%.acre. Yield, 9 bushels per acre better than College Algerians. 1924- : Twenty-four replicate half-drill strips. Yield, 7 bushels per acre over College Algerians. , e 1925-26 : Similar trial. . Advantage for B. 49 ----- 3-1 bushels per acre 1926- : Eight similar replicates. Advantage in favour of B. 49, 5-6 bushels per acre ; odds in favour of significance, 600 to 1. Details of this last trial are given in Table 2. By mistake in drilling, guard plots were omitted, so that Plot pairs 1 and 13 (the outside ones)

could not be used in calculation. The small mill used to thresh B. 49 did not remove the awns at all completely, so that the strain was difficult to sow. In Plots 2,9, and 10 there were obvious misses in the rows, so the pairs involved in those plots were also omitted from the calculations.

111. CROSS-BREEDING OF CEREALS. (a) Methods. The methods adopted for selecting among the progeny of crossbreds are only slightly modified from those already described (page 157), and are briefly summarized as follows : First year : Cross made and seed sown. Second year: All F. 1 sown. Third, fourth, and fifth years: F. 2, %c., sown in plants to rows. Selection of promising strains by inspection. Testing for homozygosity for visible characters, and starting a new family with any individual variant now appearing. Sixth and seventh years : Three-row plots, 1 rod long, replicated four or five times, the standard variety being inserted as check every fourth plot. From one hundred to two hundred families reach this stage. Elimination (largely arbitrary) of most strains, leaving twenty or . thirty. (Single - ear selection started within the best families to secure greater degree of fixity, and to start supply of seed unmixed by subsequent half-drill-strip trials.) Eighth to eleventh years : Half-drill-strip trials with constant elimination of families, and increasing exactness of trials of the survivors. (Building up pure supply of seed of the final selection separate from the half-drill strips as already explained.)* (b) Results to Date. Improvement by selection having been exploited until the rate of improvement was seriously slowed down, crossing was started in 1921 between pure strains. Only one cross has reached any advanced stage. This is the cross between College Hunter’s and College Solid-straw

Tuscan, and the object is to obtain the good grain quality and yield of Hunter’s in combination with the solid straw and tightness of chaff of Tuscan. Many of the families show the combination to a promising degree. *

In 1925-26 some of the families that Had early’ proved homozygous were tried in 5-chain rows with Hunter’s as alternate checks. The winter was abnormally wet, which may have given the crosses a special advantage. The yields for the first five families as compared with Hunter’s were 140, 150, 140, 100, and 108 per cent, respectively. Many other families were tried in i-rod rows.

In 1926-27, again, many families were tried in rod rows, and others of which most seed was available were sown in half-drill strips in competition with Hunter’s. The results for the same five families as above were 114, in, 103, 100, and 108 per cent, respectively.

In the present year families are being tested in half-drill strips, the number of replicates varying from five to thirty. IV. SELECTION IN GRASSES. The two most important sown grasses in the South Island of New Zealand are rye-grass and cocksfoot. The commercial seed consists of enormous numbers of strains all mixed upland our first step in 1923 was to demonstrate this fact. We went into a field that had been sown in commercial cocksfoot, and grubbed up plants here and there, broke each down till it consisted of a single tiller, planted that in good soil till it grew into a clump, and broke the clump up into a row. When the rows were compared they showed that some plants produced three or four times as much feed as others. A single field produced an almost infinite number of types of leafage, habit of growth, date of flowering, resistance to drought, resistance to frost, earliness of spring growth, and any other economic characters that were considered. Two rows that happened to be growing side by side are shown in Fig. 1.

It was then argued that if there are ’different types of cocksfoot (or rye-grass) and a mixture of these types is sown in any special locality, then some will be more suited to the soil and climate of that locality than others are. Those that are least suited to the environment will die out soon, and those that are best suited will survive longest, so that an old pasture where the cocksfoots have almost all died out will contain only those types selected by nature as suitable for that special environment. For instance, on our dry shingle plains a reasonable stand of cocksfoot in its second or third year will in ten years -be reduced to a few scattered plants, each many yards from its neighbour, and with the intervening spaces filled with weed grasses. Now, the assumption is that these scattered plants are those selected by nature as suitable to the dry shingle plain, and so we go to an almost exhausted field on such land and collect a hundred or two of the surviving plants. These are then broken down until one is sure that there is only one plant in the tuft. The tuft is grown in a garden until it becomes a fairsized clump, and then all the clumps are taken back to the dry shingle plain, and there divided and planted out in rows for observation. Out of the hundred rows, all composed of plants assumed to be permanent on shingly country, one (theoretically) can be selected after a few years’ trial as showing the best combination of economic characters ; and then

we shall be able to say, “ Here is a cocksfoot permanent on your country, and highly productive, frost resistant, drought resistant, &c., there.” As a matter of practice, not one but a few rows are likely to be thus finally selected.

In the belief that nature must make this selection among the enormously heterogeneous rye-grasses and cocksfoots she has to work on, we have gathered some of the strains she has approved from all the chief types of sown grassland in the South Island : from dry shingle plains near Nelson, Blenheim, Christchurch, Timaru, and Lumsden from wheat-growing land at Blenheim and Lincoln, on the Timaru Downs, and near Tapanui ; from cattle-grazing land with high rainfall in two places- on the West Coast — Koiterangi and the Grey Valley — from Omimi near Dunedin, and from Waimahaka in Southland ; from the clays of the Moutere Hills near Nelson; from the lucerne land of Seddon ; from the limestone faces of Waikari; from 2,500 ft. up on the front ranges of the mountains ; from the almost desert area of Central Otago— fact, from every region where we found a considerable and important area of any type of sown grassland. The localities of our plots are shown on the accompanying map.

Whether nature actually has produced such physiological ecotypes as our mode of work assumes we have never put to direct trial, for we have so far almost invariably grown each selection in its own environment. In the few cases where selections from two environments have been grown together distinct evidence of these physiological ecotypes has been noticed, though it appears that the formation of structural ecotypes has not gone as far as it has in England.* This, indeed, was

to be expected from the comparatively short time that the English grasses have been growing here, and from the fact that it is not yet •the custom to sow seed from one’s own locality. When and where that custom prevails the formation of ecotypes in the same grasses will probably proceed with considerable rapidity. Of course, all the work of selection ' would be useless unless there ■could be obtained from the selections seed that would reproduce the characters of its parent. That there is good hope of this being .achieved we have gained some satisfactory evidence. In April, 1924, we chose several distinct types of rye-grass and cocksfoot, and planted

small clones* of each type on a yard-square patch hoed out of the middle of a 30-acre field of oats and tares. Each clone was 5 chains from any other grass, and the lines of clones ran at right angles to. our only hot and dry wind. The oats were Algerians, which when sown in March can be relied on to shoot by 1st November, so that at the time the grasses flowered the oats were 18 in. or 2 ft. taller than the grass flower-stalks. The tares had climbed high up the oats, so that each clone of grass was practically in a well of oats and tares, screened from drifting pollen to a considerable degree. Seed produced under these circumstances we have called “ shelter-fertilized ” seed.

A large quantity of seed was produced from each clone, and it was sown immediately after harvest in a greenhouse. Germination was generally fair, but in some cases poor. Growth was rapid, so that the seedlings were ready to plant out in the open, with the first autumn rains, and by the following November were sufficiently far advanced for comparison. The seedlings from each clone were, planted in a double row containing about sixty plants.

It is . clear that if the parent clones were heterozygous for a considerable number of important characters, and if they . were open pollinated, all the rows of seedlings would be alike.. The measure by which the seedlings in any row are alike inter se, and different from those in other rows, is a measure of the efficiency of shelter pollination, and of the prospect of obtaining seed producing different types of the same grass. It was evident on inspection that each row consisted of seedlings very similar to each other in habit of growth and other economic characters, and that the separate rows were very different from each other. Two such rows are shown in Fig. 3. ..

To express the similarities and differences in figures, ■ counts were made of certain easily observed characters in two adjacent rows of rye-grass, with the following results : — /

It will be seen that C 2 shelter-fertilized seed produced plants that were significantly fewer-flowered (or later flowered), more spreading, longer-leaved, and wider-leaved than the plants similarly produced from C 15. • ' '

In the case of the cocksfoot, at the same time as the shelter-fertilized seed was produced, self-fertilized and' open-fertilized seed of some of

the same clones was also obtained. 1 All three lots of seed, were sown at the same time, and when the resultant seedlings were a year old both they and their parent plant were broken down into clones, each clonal plant consisting of a single tiller.

In a typical case there are forty-seven rows, each row .a clone, and -every clone given the same start in its present location. There are four rows of the parent, plant; three rows, each of which is a clone of a plant sprung from a self-fertilized seed ; twenty rows similarly produced from shelter - fertilized seed ; and twenty rows of open - fertilized seed. If shelter fertilization is a useful practical procedure, then the twenty shelter rows ought to be nearly as similar to the parent rows as the self-fertilized rows are, and much more nearly' similar to the selfed rows than the open-fertilized rows are. Unfortunately, the plants are not sufficiently far advanced to make measurements of particular characters, but inspection makes it. clear that the above-demanded similarities

.and differences do indeed exist. Most of the shelter-fertilized rows have at present the same growth-form as the parent, and can be easily, recognized as that parent’s offspring, while among the open-fertilized rows there are all the differences that would be ordinarily expected among an equal number of chance selections. It is dear that this assumes that the parent .plant was homozygous for many outstanding •characters, and that this is so is proved by the similarity inter se of the three rows of selfed plants obtained. In other cases the character of the selfed plants shows that the ■parent plant was heterozygous for the . most obvious characters, but .as far as we are yet able to judge the proportion of plants homozygous for outstanding characters is by no means small. ' One further piece of evidence of the efficacy of shelter fertilization is available. In two cases we have rye-grass clones of parent and their .shelter-fertilized offspring, and as these plants are further advanced

towards maturity than are the cocksfoot, some measurements were possible. The length and width of certain leaves were measured; the differences in these characters between the parent were not significant, and the differences between the offspring were non-significant also. In one feature, however, there was a marked difference between the parent plants, and this was completely reproduced in the offspring, as shown in the following table :

It thus appears that shelter fertilization is a means of producing from seed fairly homogeneous crops of rye-grasses similar to their parent strains. In the case described only i sq. yard was given to each clone to be shelter-fertilized, and enough seed was produced to sow thinly a plot of 48 sq. yards. If we found a rye-grass of sufficiently outstanding merit, and proved it homozygous for a sufficient number of characters of economic importance, we could raise a hundred times that quantity of seed by the same process —that is, enough to sow a field of about an acre. Interfertilization of most of the plants of such a field might be assumed, and so seed would be ready for immediate distribution.

V. SELECTION IN RED CLOVER.

This work was begun only in the early part of 1927. A field in the wheat-growing area was chosen which had been sown in grasses and clovers five years ago. A few scattered -clover plants still survived, and these were dug up and examined for size of roots. Only those that appeared to be really old were chosen, and any that appeared to have arisen by reseeding were rejected. We thus hoped to secure the most permanent strains. Cuttings of the selected plants were made after the method of Sylven, of Svalof,* nipping off pieces of the stem about an inch below each node, and sticking the pieces into garden soil in a greenhouse. About 50 per cent, of the cuttings struck roots, and these were planted out in- rows in early spring. Each original plant has thus produced a row of plants by vegetative propagation, and the rows can be compared. There are great differences in the habits of growth, date of flowering, and quantity of leafage among the various rows (see Fig. 4), as Williamsf shows there are among the clovers of various nationalities. There seems to be a prospect of our being able ,in course of time to. select a strain of red clover prolific and reasonably long-lived under our conditions of soil and climate.

* Beaven, E.S.: “Trials of New Cereals,” Jour. Ministry Agric. XXTX, 4, 1922

Note.—Odds in favour of significance of result, 103 to 1

Note.—Odds in favour of significance of result, 600 to 1.

* Conf. Love, H. H. : A Program for Selecting and Testing Small Grains, Jour. Amer. Soc. Agronomy, XIX, 8, Aug., 1927.

' * Stapleton, R. G.: “ Value of Selection, &c.,” Report Imperial Botanic Conference, London, 1914. " ,

* A “clone” is a group of plants produced vegetatively (not by seed) from, a single parent.

* Sylven, Nils : Sveriges Uopenings Tidskrift, Hefte 5, 1925. f Williams, R. D : “ Red Clover Investigations,” Welsh Plant Breeding Station, 1927. '

Plot No. Kilos per Plot. Hunter’s. Bell. Difference in Favour of Bell. A 14 .. . . Lost in threshing Lost in threshing 15 16-05 I5-3O -0-75 16 . . . . 135 16-40 0-05 17 17-03 17-25 0-22 18 16-81 17-20 o-39 19 17-80 16-85 -o-95 20 15-84 16-25 ' 0-41 21 , 15-40 15-40 0-00 22 14-90 i5-oo o-io 23 15-40 15-00 — 0-40 24 16-00 16-45 o-45 25 15-90 16-85 o-95 B 14 16-85 18-35 1-50 15 17-5O 18-70 1-20 16 .. .. 18-35 20-20 1-85 . 17 21-10 20-50 — o-6o 18 17-35 19-20 1-85 19 18-25 19-82 i-57 20 16-90 16-55 -o-35 21 15-8O 16-55 o-75 22 .. . I5-65 15-90 0-25 23 17-95 18-00 0-05 24 17-20 18-00 o-8o 25 I7-7O 18-50 . o-8o Means .. I6-82 i7-3i o-44 = i-i bush, per acre

Table 1. — Comparison of Bell’s Strain and College Hunter’s, 1927.

Plot No. Kilos per Plot. College Algerians. B. 49. Difference in Favour of B. 49. Plot 3 40*1 45'5 5'4 4 • • 35'3 46-8 n-5 5 . . 43‘7 45-o i-3 6 . . 42'3 50-2 7'9 7 • • 42'5 50-4 T9 8 5 1 ’ 1 49-3 — i-8 IX 37'4 49-1 117 12 36-0 42-1 6-i Means 4 i-i 46*6 6-2 = 5-6 bush, per acre.

Table 2. —Yield of Algerian Strain B. 49 in Competition with College Algerians.

— ■ C 2. C 1.5. Difference. Number of ■ flowering stems per ■ plant on 23rd November in first year 30-7 A i-66 43’4 ± 2-38-12-7 fl; 2-86 Angle of flowering stems to the. horizontal' ‘ 44-8 L i-oi° , 79-0 fl- o-66° 34-2 ± 1-17° ; Length to top cauline leaf 20-5 -- 0-25 17-4 4- 0-27 3-1 -fl 0-37 cm. Width of top cauline leaf at 1 in. from base 7-20 4 ■ 0-08 6-26 fl; 0-07 0-94 fl; O'1 mm.

Table 3. —Showing Differences between Seedlings of Two Clones Shelter-fertilized.

Angle of the Flowering Stems to the Ground — Angle of the Flowering Stems to the Ground C 2. ' ' C 5. Parent 52'5 ± 1-2° 75'0 ± °’9 Offspring . : ,. , 5 2- 5 ± I-2 ° 44-8 i i-oi° 75'0 ± 0’9 75-9 ± o-8

Table 4. -Showing that Offspring by Shelter Fertilization resemble Parents.