An experiment of this kind can only be carried out in a case where the groups of linked gens are known. At present Drosophila is the only animal (or plant) sufficiently well known to make this test possible, but this does not prove that the method is of no value. On the contrary it shows that any claim that factors can themselves be changed can have no finality until the claim can be tested out by means of the linkage test. For instance, bar eye (fig. 31) arose as a mutation. All our stock has descended from a single original mutant. But Zeleny has shown that selection within our stock will make the bar eye narrower or broader according to the direction of selection. It remains to be shown in this case how selection has produced its effects, and this can be done by utilizing the same process that was used in the case of truncate. Another mutant stock called beaded (fig. 84), has been bred for five years and selected for wings showing more beading. In extreme cases the wings have been reduced to mere stumps (see stumpy, fig. 5), but the stock shows great variability. It is probable here as Dexter has shown, that a number of mutant factors that act as modifiers have been picked up in the course of the selection, and when it is recalled that during those five years over 125 new characters have appeared elsewhere it does not seem improbable that factors also have appeared that modify the wings of this stock. [Illustration: FIG. 84. Two flies showing beaded wings.] (3) The third method is one that has been developed principally by East for plants; also by MacDowell for rabbits and flies. The method does not claim to prove that modifiers are present, but it shows why certain results are in harmony with that expectation and can not be accounted for on the basis that a factor has changed. Let me give an example. When a Belgian hare with large body was crossed to a common rabbit with a small body the hybrid was intermediate in size. When the hybrid was crossed back to the smaller type it produced rabbits of various sizes in apparently a continuous series. MacDowell made measurements of the range of variability in the first and in the second generations. _Classification in relation to parents based on skull lengths and ulna lengths, to show the relative variability of two measurements and of the first generation (F_1) and the back cross (B. C.)_ CHARACTER|GENERATION|-13|-12|-11|-10| -9| -8| -7| -6| -5| -4| -3| -2| -1| ---------+----------+---+---+---+---+---+---+---+---+---+---+---+---+---+ Length of{ F_1 | | | | | | | | | | | | | | skull { B.C. | | | | | | | | | | | | | 3| Length of{ F_1 | | | | | | | | | | | | | | ulna { B.C. | 1| | | | | 1| | 2| 3| 1| 2| 4| 4| _same table continued_