as in others, is in spite of the attempted selection, an extreme, ly variable characteristic even when conditions are kept as uniform as possible. The great variability of these fowls under the identical conditions of a single year indicates that the bases or causes of variation in egg production are innate rather than environmental.

It is not easy to find comparable data on Rhode Island Reds from points outside of the United States, since this has been chiefly an American breed. Some representative figures compiled from laying contests in England and the British Dominions, and from one other in the United States, are given. in Table 3. The conditions of these contests are similar to those from which our data are derived, but the numbers are too small to afford a very valuable comparison. At the Vineland contest, the average production and the variability of Rhode Island Red pullets were similar to those at Storrs. In recent Australian contests the average egg production of Rhode Island Red pullets was about twenty eggs higher than at Storrs; while the fowls in the two places were about equally variable. The chief differences are to be noted in the extremely high average egg production of the Rhode Island Reds in the last three contests at the Harper Adams Agricultural College at Newport, Salop, England. There the average egg production was some thirty eggs higher than at Storrs, while the variability of the fowls was much lower. This last difference is due to the almost complete elimination of low producing fowls, i. e. those laying about 100 eggs and less, leaving a rather homogeneous, high producing group.

The numbers of Rhode Island Reds in the South African and Canadian contests are too small for use. The records were in general rather low, while the average production for fortyeight weeks in the Bentley, Suffolk contest was only slightly less than the full years average for the nine years at Storrs.

YEARLY DIFFERENCES IN EGG PRODUCTION
(a) MEAN

The differences between the average egg production of different years is plainly shown in Table 2 and in the scattered points in Fig 2. Two years were obviously different from

all others-1913 and 1918-when the average egg production was quite low. These were also the years of lowest production in the Wyandottes (Bulletin 117 p 46) and in the other breeds as well. These differences were probably due to unfavorable conditions of health, weather and care as noted in the previous bulletin. The years of highest average production were 1915 and 1919, but in neither of these was an extremely high average attained. 1919 was not the year of highest production for the contest as a whole and the conditions which produced the higher production in this breed were probably peculiar to this breed alone, such as chance choice of more fecund fowls, or more skill in selecting entries.

The egg production of other years was similar. Many of the deviations are probably due only to the inherent error of sampling and are hence not real or significant. In Appendix Table III, the magnitude of the differences between different pairs of years is given in relation to the probable errors of the differences. The ratios of differences to their errors of three and over may be taken as indicating differences other than those due to sampling alone. Of such differences there are relatively few and they occur only in the comparison of 1913 and 1918 with other years.

The differences in mean production between 1913 and all other years except 1918 are undoubtedly significant. The differences between 1918 and 1915, 1916 and 1919 are significant while the egg production of 1918 was lower (though not significantly so) than egg production in 1911, 1912, 1914 and 1917. If we turn to the actual frequency distributions for the years 1913 and 1918 (Table 1a) we find that in these years of low production the range of variation has not been restricted. In 1918 on the other hand the range is greater than in any other year. A normal proportion of birds laying 210 eggs and more is present together with an abnormally large propor tion laying 104 eggs and less. The lowering of the mean is due not so much to the, prevention of high records as it is to the increase of the low classes at the expense of the medium. This then is possibly the explanation of the high variability which accompanies the low production of 1918 (and 1913): that unfavorable environment, instead of eliminating the highest producing individuals and thus shortening the whole distribution, acts differentially on the high and mediocre producers and with greater severity on the latter. A high producing hen is apparently one which is less susceptible to changes of environment because of a greater inherent vitality coupled with the inherent capacity to lay a large number of eggs.

Differences in egg production in the other years are not statistically significant, and we may say that with the exception of two years in which bad conditions were encountered, egg production was fairly similar in all years.

(b) VARIATION CONSTANTS.

The yearly changes in the measures of variability-the standard deviation and the coefficient of variation-are similar

to the changes in the mean. The only notable deviations in the nine years were the high variabilities of the fowls in 1913 and 1918 due to unfavorable conditions. The constants for the other years are roughly comparable, the coefficient of variation ranging only from 24 to 28 per cent. About the same kinds of fowls were submitted each year.

THE GENERAL TREND OF EGG PRODUCTION THROUGH THE NINE

YEARS

CHANGES IN AVERAGE EGG PRODUCTION

Although it is evident that many of the differences in mean egg production between different pairs of years (without regard to sequence) are probably due only to sampling and are therefore not statistically significant, it is possible that by their arrangement in a sequence, a significant direction of change may become evident. Are the means for the different years arranged in a descending or an ascending order? Is there a marked tendency toward increasing or decreasing egg production in the different lots of pullets tested between 1911 and 1919?

The means of egg production in the different years as given in Table 2 have varied irregularly. The course of these changes is shown in Fig. 2. As an approximate method of determining whether the tendency exhibited by these changes is toward higher or lower production, we have by the method of least squares fitted to the points designating mean production. in the separate years, that straight line which most nearly describes the changes in the nine years (solid line, Fig 2). Assuming the origin of this line on the ordinate for the year preceding the first year of the data (Yo-1910) the equation of the line is found to be Y-147.753+1.001x. The slope of the line is slightly upward; the tendency toward increase is small, amounting to about one egg per year, and if statistically significant, it indicates that egg production in the Rhode Island Reds entered in this contest has shown some improvement. In analysing the egg production of Wyandottes in the previous bulletin of this series a similar tendency was observed.

This describes changes due to both environmental and other causes. If we wish to measure the changes in the innate egg laying ability of the fowls submitted over a period of years, it is hardly fair to include the records of two years in which conditions were probably abnormal. We have, therefore, omitted the records of 1913 and 1918. and have recalculated the slope of the line which most nearly describes changes in production in the seven normal years. The equation of this line is y=157.427+1.172x (broken line Fig 2). The slope is up and the tendency is toward an increase in egg production. The rate is low and is only slightly greater than that obtained by including all years; while the level of the line is raised by omitting the two lowest years. There is, therefore, an indication of a slight tendency toward an increase in the potential egg production of the Rhode Island Red pullets submitted to the contests. It must be emphasized that this method of measuring the trend of changes in production is descriptive only in a general way of the direction of change in the past and cannot be made the basis for wide generalizations or for prediction of future egg production. The mean production is affected in each year not only by the innate fecundity of the fowls submitted, but by the conditions under which they make their records. A combination of unfavorable conditions in any future year is entirely beyond prediction and would render unreliable any estimate of increases to be expected or even a prediction of the fact of increase itself.

CHANGES IN VARIABILITY

There appears to be no need of applying refined methods to determine the trend of the changes in the measures of variability, since with the exception of the two abnormal years, these have remained fairly constant from year to year. Whether the years of low average production are included or, omitted, the tendency is for the coefficient of variation of the fowls entered to remain stationary.

CHANGES IN THE PERCENTAGE OF HIGH AND LOW PRODUCING FOWLS

Of greater interest, perhaps, than the trend of average production, is the question whether the proportion of high and low producing fowls has changed during the nine years. Pro