I now want to turn to a set of experiments which in their own way, are as beautiful as Mendel's, and important, although not as important as Mendel's, for genetics. Before introducing them let me briefly recall that well before Mendel's work, people were aware of the possibility of changing the mean value of a QT in some group of organisms by selection. I showed the plots of average oil content for the Illinois Station corn experiments begun in 1896, selecting for high and low oil content each year. We saw that mean yield moved steadily up or down, according to the direction of selection, and that this could continue for many generations. Of course the many breeds of dogs, cats and other domestic animals would have been achieved through generations of selection, so this is ancient human knowledge.
Here is what Johannsen did. He experimented with beans, specifically Princess beans, Phaseolus vulgaris nana. Recall that, like peas, beans are naturally self-fertilizing. He began his series of experiments in 1900 by planting a number of bean seeds from an 8 kg lot, always permitting self-fertilization. In 1901 he harvested 287 plants from selected seeds of very different sizes and known weights.
In 1902 he planted 524 seeds from his 1901 crop, and harvested 5,494 seeds. Each of the 524 seeds originated from one of a set of 19 single beans (which he designated A, B, ..., T). He kept track of the weight (in mg) of the mother and the grandmother beans (being Princesses, all these beans are being viewed as female!).
Finally, he selected the smallest and largest beans (seeds) in each generation, and planted these (1902-1907).
What did he find?
Firstly, the distribution of the 5,494 weights followed a normal curve pretty closely, at least that appeared to be the case from his histogram with 7 bins. This was of some interest, but was far from the most interesting thing he found. His really striking results are illustrated by the following, referring to line J. I selected this line because there were the most weight classes for the mother bean in this one of his 19 lines. The results from all other lines are similar.
Weight of grandmother bean (1900): about 800 mg.
Average weight of mother beans (1901) about 400 mg. [We do not know how many.]
What do we see? Clearly the mean weights of offspring from mothers in different weight classes do not seem to differ greatly, and probably not significantly, given the SDs. Is this surprising? With hindsight, the answer must be no: the seeds from each of the 1901 mother beans are the result of two generations of selfing from a single grandmother bean, which in turn is probably the result of many generations of selfing. To use the terminology Johannsen introduced, they are pure lines of beans, and as such, the only variation is environmental...they are genetically identical. Indeed Johannsen introduced the terms genotype and phenotype to make just this distinction. He found that each of the sets of beans from a given pure line fitted a normal distribution about the line mean, and that selection within a pure line had no effect. In fact we might predict that there would be a small reverse effect, for selecting seeds that are heavy (or light) for planting for the next season, might be expected to produce seeds that are lighter (or heavier) in the next generation, simply by regression to the mean. However, any effect like this is likely to be masked by the large year to year variation in average weight of beans within the same pure line.
Looking at Johannsen's 19 pure lines overall, we find that there is a range of 300 mg between their mean weights, whereas the SD within a pure line was typically about 75 mg.
It is not hard to see the importance of Johannsen's results for selection: successful selection requires genetic variation on which to act. The corn selection experiments I used to set you up were successful precisely because corn is outbreeding: the male part of the plant (tassel) is at the top, and the female part (ovules) are in ears lower down, and under natural conditions, corn outbreeds.