Pot 2 : 19 1/8 : 18 3/8 :: 21 4/8 : 18 5/8 :: 22 : 18 3/8 : -3 5/8. Pot 2 : 21 4/8 : 18 5/8 :: 19 1/8 : 18 3/8 :: 21 5/8 : 18 : -3 5/8. Pot 2 : - : - :: - : - :: 21 4/8 : 18 : -3 4/8. Pot 2 : 22 1/8 : 18 5/8 :: 23 2/8 : 18 5/8 :: 21 : 18 : -3. Pot 2 : 20 3/8 : 15 2/8 :: 22 1/8 : 18 :: 21 : 17 3/8 : -3 5/8.
Pot 3 : 18 2/8 : 16 4/8 :: 21 5/8 : 16 4/8 :: 20 3/8 : 16 4/8 : -3 7/8. Pot 3 : 21 5/8 : 18 :: 20 3/8 : 16 2/8 :: 19 1/8 : 16 2/8 : -2 7/8. Pot 3 : 23 2/8 : 16 2/8 :: 18 2/8 : 15 2/8 :: 18 2/8 : 15 4/8 : -2 6/8. Pot 3 : - : - :: - : - :: 12 : 15 2/8 : +3 2/8. Pot 3 : 21 : 18 :: 23 : 18 :: 12 : 12 6/8 : +0 6/8.
Pot 4 : 22 1/8 : 12 6/8 :: 22 1/8 : 18. Pot 4 : 23 : 15 4/8 :: 21 : 15 4/8. Pot 4 : 12 : 18 :: 12 : 12 6/8.
"The observations as I received them are shown in Table 1/1, Columns 2 and 3, where they certainly have no prima facie appearance of regularity. But as soon as we arrange them the in order of their magnitudes, as in columns 4 and 5, the case is materially altered. We now see, with few exceptions, that the largest plant on the crossed side in each pot exceeds the largest plant on the self-fertilised side, that the second exceeds the second, the third the third, and so on. Out of the fifteen cases in the table, there are only two exceptions to this rule. We may therefore confidently affirm that a crossed series will always be found to exceed a self-fertilised series, within the range of the conditions under which the present experiment has been made."
TABLE 1/2.
Column 1: Number (Name) of Pot.
Column 2: Crossed.
Column 3: Self-fertilised.
Column 4: Difference.
Pot 1 : 18 7/8 : 19 2/8 : +0 3/8. Pot 2 : 20 7/8 : 19 : -1 7/8. Pot 3 : 21 1/8 : 16 7/8 : -4 2/8. Pot 4 : 19 6/8 : 16 : -3 6/8.
"Next as regards the numerical estimate of this excess. The mean values of the several groups are so discordant, as is shown in Table 1/2, that a fairly precise numerical estimate seems impossible. But the consideration arises, whether the difference between pot and pot may not be of much the same order of importance as that of the other conditions upon which the growth of the plants has been modified. If so, and only on that condition, it would follow that when all the measurements, either of the crossed or the self-fertilised plants, were combined into a single series, that series would be statistically regular. The experiment is tried in Table 1/1, columns 7 and 8, where the regularity is abundantly clear, and justifies us in considering its mean as perfectly reliable. I have protracted these measurements, and revised them in the usual way, by drawing a curve through them with a free hand, but the revision barely modifies the means derived from the original observations. In the present, and in nearly all the other cases, the difference between the original and revised means is under 2 per cent of their value. It is a very remarkable coincidence that in the seven kinds of plants, whose measurements I have examined, the ratio between the heights of the crossed and of the self-fertilised ranges in five cases within very narrow limits. In Zea mays it is as 100 to 84, and in the others it ranges between 100 to 76 and 100 to 86."
"The determination of the variability (measured by what is technically called the 'probable error') is a problem of more delicacy than that of determining the means, and I doubt, after making many trials, whether it is possible to derive useful conclusions from these few observations. We ought to have measurements of at least fifty plants in each case, in order to be in a position to deduce fair results. One fact, however, bearing on variability, is very evident in most cases, though not in Zea mays, namely, that the self-fertilised plants include the larger number of exceptionally small specimens, while the crossed are more generally full grown."
"Those groups of cases in which measurements have been made of a few of the tallest plants that grew in rows, each of which contained a multitude of plants, show very clearly that the crossed plants exceed the self-fertilised in height, but they do not tell by inference anything about their respective mean values.