THE BEST STAGE FOR CUTTING WHEAT.

New Zealand Journal of Agriculture, Volume XXVI, Issue 6, 20 June 1923, Page 354

THE BEST STAGE FOR CUTTING WHEAT.

EXPERIMENT UNDER THE STATISTICAL METHOD AT . LINCOLN.*

F. W. HILGENDORF,

D.Sc., and J. W. CALDER, B.Agr., Canterbury Agricultural College, Lincoln.

If wheat is cut too early it shrivels, and if cut too late there is a risk that grain will be lost by shaking. The question . that has to be answered many times each harvest, therefore, is, How soon can the wheat be cut without losing weight by shrivelling? ”

To answer this question by a definite experiment had long been in mind, but it seemed difficult to arrange. If one were to take the yields of adjacent plots cut early and late, the mere removal of its neighbour would give the late-cut plot an advantage ; and if the plots were taken large enough to obviate this difficulty they might run into different soils, and so would need numerous repetitions to correct the variations due to this cause. So the question was left as requiring more time than was then available.

Some time before last harvest, however, it occurred to one of us that it may confidently be taken for granted that no new grains of wheat can be formed during the last fortnight before maturity, and consequently that the weight of, say,' 100 grains of the early and latecut crops would be an accurate reflex of their final yields. An experiment was therefore arranged on these lines. An even crop of Solidstraw Tuscan was taken, and a block 14 yards by 11 yards was marked out therein. Strings were run lengthwise and across the block at each yard, so that 154 plots, each 1 yard square, were marked out. Since it was designed to cut the wheat at five different stages, the plots were named Ai, Bi, Ci, Di, Ei ; A2, B2, C2, &c., so that there were thirty or thirty-one plots bearing each naming letter, and each letter was distributed evenly over the . block. ;

When it was judged that the crop was about a fortnight from maturity cutting commenced, twelve straws being cut from the middle of each of the square yards named Ar, A2, A3, &c. The straws cut were always adjacent, they were cut without looking at their heads to see if large or small, and were all cut at ground-level. Just before cutting elaborate notes were taken as to the stage of ripeness that the crop had reached, and after cutting each bundle of twelve straws was tied and labelled and hung up -indoors to harden and dry.

After three days new observations were taken, and twelve straws were similarly cut and tied from each of the thirty-one plots marked B ; and so on at three-day intervals until plots E were cut, twelve days after the cutting of plots A. The first cutting was made while the

general aspect of the crop was still decidedly green, and the last one when about half a bushel of grain per acre had been shaken on to the ground, so that the various cuttings embraced the whole period during which any one would be likely to harvest a crop on a farming scale. Complications possibly arising from differences in weather after cutting were reckoned with in a subsidiary experiment. The actual stage of maturity at which each cutting was made will be described later. ' •

Before proceeding to state the weights of the 100 grains cut at the various stages, it is perhaps necessary to explain why so large a number as thirty plots were taken from which to cut representative straws at each stage. It was because of the generally unsatisfactory nature of averages as commonly determined in agricultural . experiments. An average alone gives a scanty amount of information. An average of 20 may mean that there were 100 observations between 19 and 21, with a mean of 20 ; or it may mean that there were two observations of 1 and 39. Again, in tossing 20 coins 20 times we may find that an average of 9 heads will turn up—a result that is obviously not strictly reliable. Therefore mathematicians have invented a device to indicate the reliability of any average. They add to the figure indicating the mean another figure indicating what is called the "probable error.” Thus, if an average is stated to be 20 i 1 the 1 is called the probable error, meaning that if another average were computed with equal care the chances would be even that the new average would be between 19 and 21, or outside these limits. Not the clearest of indications, one might think, but one that becomes quite easily understandable with use. It is clear that the smaller the probable error the more reliable the average,

Returning to the Lincoln experiment, the twelve heads of each of the thirty A plots and the twelve from each of the thirty B plots, &c., were hung up for a couple of months until they were all dry, and assumed to have the same water .content — assumption that was checked and found correct later on. ' They were all threshed on the one day, and from each of the 154 lots 100 grains were counted, and weighed within two days. The averages with their probable errors were as follows :— Average Weight (from thirty or thirty-one Plots) of 100 Grains of Solid-straw Tuscan Wheat cut at Three-day Intervals. Stage A (the earliest) . . . . 4-047 ± 0-024 grammes. ,,. B cut three days later .. 4-060 A 0-026 ,, „ C ,, ' . . 4-299 ± 0-028 ,, D „ . .. 4-269 0-032 „ E „ ■ .. . 4-299 ± 0-045 These figures are of considerable interest. The smallness of the probable errors (0-5 to i-o per cent, of the averages) is satisfactory proof that the weighings were accurate and their number sufficient, but no explanation occurs to us of their gradual increase with the advance in the date of cutting. There is a general similarity between the A and B cuttings, then a distinct break between B and C, and then a similarity between C, D, and E. The variations between the last three — i.e., the’ falling-away at D to less than C or Emight:

occasion some concern at first, until the meaning of probable error" is recalled. The weight of C is 4-299 ± 0-028 grammes i.e., it is equally likely to be between 4-271 grammes and 4-327 grammes or outside these limits. At the same time the weight of D is 4-269 i 0-032 grammes — i.e., it is equally likely to be between 4-238 grammes and 4-301 grammes or outside these limits. Thus the lower limit of C (4-271 grammes) is much lower than the upper limit of D (4-301) — that is, the two overlap- —so that there is no real difference between C and D, but only a difference caused by the special set of samples here used to compute the means. Such a difference is said to be “ non-significant ”i.e., it does not signify any real difference.. Mathematicians have computed from their study of probability that any difference less than three times the probable error of that difference has only a 20-to-i chance of being a real difference—that is, one not caused by mere fluctuations of sampling. Thus, on comparing averages, the probable errors of the differences are of great importance. They have been calculated for the present set of figures as follows :— Difference between A and B = 0-023 4 6-035 grammes. ,, B and C = 0-239 T 0-038 . ,, ■ C and D = 0-030 4; 0-042 ,, ,, D and E = 0-030 4 0-055 >• Thus, the differences between A and B, C and D, and D and E are less than the probable errors of . these differences, and so are nonsignificant. Therefore we cannot say that there is any difference in weight between grain cut at stages A and B and between those cut at stages C, D, and E. On the other hand, the difference between B and C (0-239 grammes) is 6-3 times the probable error of the difference, 0-038. This is the outstanding fact of the investigation, and consultation of tables of probabilities shows that the chances are 25,000 to 1 that such a difference is significant —odds that most of us would be willing to accept as certainty. Thus we conclude that if we cut at stage C there are 25,000 chances to 1 that we will get a higher yield than cutting at stage B, but that if we wait longer no increased weight need be anticipated.

Two more points must be mentioned. Firstly, a further calculation will enable us to estimate the gain obtained by waiting from stage B to stage C as 2-5 bushels per cent., or exactly 1 bushel per acre on a 40-bushel cropthe probability of at least this gain occurring being 40 to I,. which may be taken as practical certainty. Secondly, we may now define the degree of maturity to which the crops had attained at the critical stage C, cutting before which results in a loss of weight, and cutting after which results in no gain. The stage will be more closely defined by describing those before and after it as follows :■ —-

Stage B, second cutting: General aspect of cropnearly ripecoloured, in some patches nearly all green has disappeared. Straw practically all straws now yellow, but a few still green just below the head and just above top knot. Headsone-third still green and twothirds white or turning white. Grainthat in white heads when squeezed between the fingers produced a dry dough. (This stage proved too early to cut profitably).

Stage C, third cutting, three days after B : General aspect of ripe-coloured, but close scrutiny showed green tinge. Straw all yellow except about i per cent., which showed in. of green above the top knot ; all knots green. Heads —ripe-coloured except about r per cent, still green. Grainthat in ripe heads would not squeeze out any kind of dough, but cut easily with thumb-nail. (This proved the earliest . stage at which cutting meant no loss of weight.) Stage D, fourth cutting, three days after C : General aspect — quite ready to cut, a slightly green head and stalk 'being found only after close search. Strawnearly all ripe to the head, but about 5 per cent, still yellow below head. Knotsmost knots still green, but about 25 per cent, half brown and a few quite brown. Heads — ripe. Grain—will not squeeze but will still cut with thumb-nail. (No gain in weight resulted from leaving the crop till this stage.) . SUMMARY AND CONCLUSION. A crop of wheat was cut at five different stages of ripeness at three-day intervals, and thirty plots were cut at each stage so as to reduce the experimental error of the average. The relative crops at various cuttings were computed by weighing 100 grains of each of the thirty plots cut at each stage. A calculation of probabilities showed 25,000 chances to 1 that an increased weight was securable by waiting till stage C, and that there was no further gain but only risk of loss by waiting longer. There is a 40-to-i chance — i.e., a practical certaintythat the gain from waiting till stage C is 1 bushel per acre on a 40-bushel crop. . ■ ' Stage C is thus shortly defined The green has been replaced .by yellow in the top internode of 99 per cent, of the straws ; all knots are still green ; no dough can be squeezed out from the grain, but the grain is still soft enough to cut with the thumb-nail.

Raising Cherry-plum and Olearia Forsteri for Hedge Purposes. — Both these plants root quite freely from cuttings if properly inserted in suitable soil. It would be in nearly all cases useless to plant the cuttings in a hedge-row, as conditions suitable for rooting are rarely obtainable. For the cuttings select a plot of good friable soil where the drainage is good. Dig or deeply plough the soil, throwing ' aside any bulky weeds there may be, as these will not have time to rot down before the cuttings are put ' in. When planting the cuttings dig with a spade till there is room for a row ; then stretch a line where the first row is to go. Make a trench in front of the line by inserting the spade, held quite upright, against the line and dragging the soil forward. The trench will require to be of a depth to suit the length of the cuttings, which may be placed with about two-thirds of their length in the ground. Place a row of cuttings in position, and then push some soil against them and tread it firm. Finish by digging to the row of cuttings, leaving the surface soil loose, until there is room for another row, then bring the' line forward and continue as before. The plants will be ready for setting out in hedge-rows the following year. Cherry-plum cuttings may be 12 in. to 15 in. long, and should be made from straight pieces of young wood, not thin spray. Make and plant them as soon as possible after the leaves fall, placing them about 3 in. apart. For cuttings of Olearia Forsteri take the ends of branches cut about 12 in. long, trim the soft tops off, cut the base square across just under a joint, and trim off side growths so as to leave a head with two or more branches. Insert so that the heads just touch each other in the row. These cuttings may be put in at any time up to the middle of August. To facilitate rooting and after-growth, the space between the rows of cuttings should be kept free of weeds and the surface frequently loosened. —Horticulture Division.

* This article is of note as recording what is probably the first application of the “method of statistics” and the “ the'ory of probability ” to an agricultural experiment in New Zealand. The system has been increasingly adopted of late in experimental work in —Editor.