Equally important are the results that Jennings has obtained with certain protozoa. Paramecium multiplies by dividing across in the middle, each half replacing its lacking part. Both the small nucleus (micronucleus) and the large nucleus (macronucleus) divide at each division of the body. Jennings found that while individuals descended from a single paramecium vary in size (fig. 89), yet the population from a large individual is the same as the population derived from a small individual. In other words, selection produces no result and the probable explanation is, of course, that the different sizes of individuals are due to the environment, while the constancy of the type is genetic. Jennings found a number of races of paramecium of different sizes living under natural conditions. The largest individual of a small race might overlap the smallest individual of other larger races (fig. 89); nevertheless each kind reproduced its particular race. The results are like those of Johannsen in a general way, but differ in that reproduction takes place in paramecium by direct division instead of through self-fertilization as in beans, and also in that the paramecia were probably not homozygous. Since, however, so far as known no "reduction" takes place in paramecium at each division, the genetic composition of parent and offspring should be the same. Whether pseudo-parthenogenesis that Woodruff and Erdmann have found occurring in paramecium at intervals involves a redistribution of the hereditary factors is not clear. Jennings's evidence seems incompatible with such a view. [Illustration: FIG. 90. Stylonychia showing division into two. (After Stein.)] More recently one of Jennings's students, Middleton, has made a careful series of selection experiments with Stylonychia (fig. 90) in which he selected for lines showing more rapid or slower rates of division. His observations seem to show that his selection separated two such lines that came from the same original stock. The rapidity of the effects of selection seems to preclude the explanation that pseudo-parthenogenesis has complicated the results. Nevertheless, the results are of such a kind as to suggest that they were due to selection of vegetative (somatic) differences and that no genetic change of factors was involved, for his conclusion that the rapidity with which the effects gained by long selection might be suddenly reversed when selection was reversed is hardly consistent with an interpretation of the results based on changes in the "potencies" of the factors present.