in the contest. After December, the standard deviation declines to reach a low point in April, when this group of fowls is more homogeneous than at any other time, after which it rises gradually to a second high point in October.

COEFFICIENT OF VARIATION. These changes are better shown when measured by the coefficient of variation, which takes into account the changes in mean egg production from month to month. Here we find an enormously large variability in the fowls in November, which is lowered as differences in the rate of maturity and in pre-contest conditions are extinguished, and reaches a low point in April when practically all of the birds are laying. Even in this month when every fowl is more nearly like every other one than at any other time in the year, a considerable amount of variation still exists; indicating that even after six months under similar conditions of care, and under a common reproductive impulse which reaches its height at this time, a high degree of individual diversity in fecundity asserts itself. Such diversity can only be regarded as inherent or innate, and probably relates chiefly to individual variations in the rate of egg production, rather than in maturity or other similar characters. The changes in the coefficient after April are more gradual, but variability regularly increases as more individual traits assert themselves, such as variation in broodiness, time of molting and other physiological factors which affect egg production in the summer. The birds never again become as variable as they were in November, and variability at the close of the contest year is only half as great as at the beginning. In no single month, it is to be noted, is the coefficient of variability as low as it is when the whole year's record is used as the unit of variation. This shows that individual differences are much more striking when considered over a short period of time, since many of the variable traits affecting egg production, operate only in a limited time; such as rate of maturity, susceptibility to broodiness, etc. Their effects on the whole year's record appear to be much less important than on the record made while they are acting.

OF THE DOMESTIC FOWL

EGG PRODUCTION IN THE FOUR MAJOR SEASONS.

The changes in mean egg production and in variability in egg production we have found in general to follow a single orderly sequence indicating that for the whole sample of Barred Rocks, the period from November first to October 31st forms the principal part of a single natural cycle. It is not difficult to see from the diagram in Fig. 5, however, that while every month differs to some extent from every other, there are several groups of months in which either mean egg production itself or the direction of changes in egg production are very similar. Thus the months from November to February form a period in which egg production increases at an almost uniform rate; from March to May it is practically stationary: from June to September it is decreasing at a rate which varies somewhat from month to month, but is roughly similar in the several months of the period. These natural periods of egg production have long been recognized and the annual cycle of egg production has been subdivided into periods corresponding roughly to the seasons of the year: Winter (November 1 to the end of February); Spring (March 1 to May 30); Summer June 1 to August 31) and Autumn (September 1 to October 31).

Changes in the average egg production of Barred Rocks in these periods may be obtained by adding together the averages for the months as given in Table 7. These totals are given in the table below, to which has been added the proportion of the annual production which occurs in each season (from Table 11).

The periods plainly differ in the number and proportion of eggs laid. The winter months, as in all breeds, contribute least to the annual total, while the spring period contributes most. In addition to these obvious differences, the periods dif

fer in other more important and interesting ways. In order to determine the significance of these differences, we must now examine them more closely, and note (1) the salient features of egg production within each period and (2) the character of the changes which take place in egg production as we pass from one period to another. We shall be particularly concerned in deciding whether these transitions are abrupt or gradual, and whether in this breed, the seasonal periods represent distinct and separable reproductive cycles. Pearl and Surface (1911) in their analysis of fecundity in this same variety, found some indications of the existence of a distinct winter cycle of fecundity, which was separated from the succeeding TABLE 8

SEASONAL DISTRIBUTION OF EGG PRODUCTION BY YEARS. Percentage of Total Eggs Laid in the Season Indicated.

or spring cycle by a period during which egg production declined, indicating a period of reproductive rest inserted between two cycles of greater activity. Goodale (1918) likewise, found distinct evidence of such a rest period in many Rhode Island Red pullets, but was too well aware of the complexity of the factors concerned to conclude that it was a general characteristic of fecundity. Evidence that such separate cycles are either non-existent or are extinguished in the contest data by variations in the time of hatching and rate of maturity, was obtained for Wyandottes and Rhode Island Reds (Bulletins 117 and 118 of this series) and the hypothesis advanced that in each of these breeds as a whole, the typical reproductive cycle consisted of a whole year's activity, which cen

OF THE DOMESTIC FOWL

tered about the spring or natural egg laying period, and that the fecundity of other periods was the result of the extension of the reproductive activities of the spring into other seasons by gradual and continuous increments of increase. For this breed our efforts will be directed toward deciding between the above hypotheses. The evidence on egg production in the separate seasons follows:

WINTER. This is the period of first egg production, and during the four months from November 1 to the last day of February, most of the contest fowls attain sexual maturity and begin to lay. (See p 244). Once having begun to lay, the rate of egg production for individual fowls increases gradualTABLE 9

MONTHLY VARIATION IN THE EFFICIENCY OF EGG PRODUCTION.

(1) This column gives the percent yield of the flock in each month

6.59 +16

2.67

-19.07

It is to be read: in November the flock (or average bird) produced 11.27% of the total number of eggs which could possibly be produced in November, etc. The average Barred Rock is therefore 11.27% efficient in November.

(2) Two months of 29 days and seven months of 28 days each.

ly, although this varies considerably with individuals. These two processes are reflected in the steady upward trend of the seasonal curve of egg production during this period (Fig. 5). The advance is uninterrupted from November to February, and there is no pause in the flock as a whole to indicate the completion of a cycle.

These facts are plainly evident in the table of monthly means (Table 7) and in Fig. 5 which illustrates this table. The rate of increase of egg production during these months is better shown by the data of Table 9. The numbers of days in the months of the winter season are not equal and in order to have a true measure of the changes in the rate of egg production in successive months we have taken as the rate for any month the percentage relation between the actual number of eggs laid and the maximum number of eggs (as limited by the number of days in the month) which it would be possible for a fowl to lay. This rate is the "per cent production" now familiar to poultrymen. Its values for each month are given in the third column of Table 9. The changes in rate with which we are here concerned are given in the last column of Table 9. The changes for the winter period are all increases. The increases for November and December, as we have shown elsewhere, are probably too high, because of assuming production previous to November 1 as zero and because of the change in conditions at the opening of the contest. The increases in January and February probably give a true measure of the normal conditions and these show that a nearly uniform rate of increase is maintained through February.

Although we have not attempted to examine individual records with a view to detecting separate cycles, we have divided the total number of fowls into groups differing by fifteen eggs in annual production. In Table 10 the monthly distribution of each of these classes is given. With only a few unimportant exceptions, the number of eggs laid by the birds in each class increase in each month from November to March. A comparison of the seasonal distribution of egg production in the high, low and medium egg producing groups is given in Table 11. Although there are differences between these groups which will be discussed later, it must be noted that in the four months of the winter period egg production increases in each month in each group. There is, therefore, no cessation of egg production or abatement of the rate of increase to mark the close of a separate reproductive cycle.