There is no way from Mendelian data by which this difference between a true case of multiple allelomorphs and one of complete linkage (as just illustrated) can be determined. There is, however, a different line of attack which, in a case like that of _Drosophila_, will give an answer to this question. The answer is found in the way in which the mutant factors arise. This argument has been fully developed in the book entitled "The Mechanism of Mendelian Inheritance," and will therefore not be repeated here. It must suffice to say that if two mutant {13} types that behave as allelomorphs of each other arise separately from the wild form, one of them must have arisen as a double mutation of two factors so close to each other as to be completely linked--a highly improbable occurrence when the infrequency of mutations is taken into consideration.[1] The evidence opposed to such an interpretation is now so strong that there can be little doubt that multiple allelomorphs have actually appeared. On _a priori_ grounds there is no reason why several mutative changes might not take place in the same locus of a chromosome. If we think of a chromosome as made up of a chain of chemical particles, there may be a number of possible recombinations or rearrangements within each particle. Any change might make a difference in the end-product of the activity of the cell, and give rise to a new mutant type. It is only when one arbitrarily supposes that the only possible change in a factor is its loss that any serious difficulty arises in the interpretation of multiple allelomorphs.