We have already mentioned that before the maturation division occurs the homologous maternal and paternal chromosomes fuse--the so-called synapsis of the cytologists--and afterward separate again. It had been observed by Janssens that in this stage of fusion and subsequent separation a partial twisting and a partial exchange between two chromosomes may take place. Morgan assumes that this exchange accounts for certain deviations in the ratio of linkage. If in Fig. 40 the white and black signify two homologous chromosomes I and I₁ containing the two pairs of homologous factors AB and ab respectively, the synapsis state would be as in Fig. 41. If the separation were complete, either I or its homologue I₁ might be lost in the maturation division of the egg. If, however, the synapsis is slightly irregular, as in Fig. 42, where the chromosomes are slightly twisted, I and I₁ will not separate completely but an exchange will take place, part of I₁ and I becoming exchanged. This would result in the formation of two mixed chromosomes Ab and aB (Fig. 42). This partial exchange of homologous chromosomes, which Morgan calls “crossing over,” occurs, as he found in _Drosophila_, in the egg only, not in the maturation division of the sperm. He informs me that in the silkworm moth Tanaka found that it occurs only in the male, while in _Primula_ it takes place both in the ovules and in the pollen as shown by Gregory. [Illustration: FIG. 40] [Illustration: FIG. 41] [Illustration: FIG. 42] Morgan and his fellow-workers have put this theory to numerous tests by breeding experiments and the results have fully supported it. According to the chromosome theory linkage should occur only when factors lie in the same chromosome. Hence it should be possible, on the basis of this linkage theory, to foretell how many linkage groups there may occur in a species; namely, as many as there are chromosomes. In _Drosophila_ there are four pairs of chromosomes, and Morgan and his fellow-workers found only four groups of linked characters.[207] This agreement can be no mere accident. [207] The number of hereditary characters examined to test the theory was over 130. Carrying the assumption still farther, these authors were able to show that each individual character has in all probability a definite location in the chromosome, so that it seems as if each individual chromosome consisted of a series of smaller chromosomes, each of which may be a factor in the determination of a hereditary character which is transmitted according to Mendel’s law of segregation. Biology has thus reached in the chromosome theory of Mendelian heredity an atomistic conception, according to which independent material determiners for hereditary characters exist in a linear arrangement in the chromosomes. _II_