REDUCTION OF VARIABILITY. 115 



Every time a small group of animals or plants start a colony, 

 the resulting group will have a limited potential variabil- 

 ity. Such animals as have colonizing habits will perpetually 

 start new, isolated colonies from small beginnings. Such ani- 

 mals will be found in groups of which every one is relatively 

 pure. (Lloyd). 



The case of the house-rat is a good example. It is not given to 

 much wandering, like the Norway rat. Each farmhouse in 

 France has its own Mus rattus population, and each of these 

 populations is relatively stable. We found the same state of 

 things in Holland, and in Java, as Lloyd observed in Brit- 

 ish India. In one house all the rats will be grey-bellied agouti, 

 in another house a small colony of white-bellied rats is found, or 

 all black rats. 



We saw that in a common, well-established species, where 

 the number of individuals remains constant from year to year, 

 the fact, that the number of plants or animals having off-spring 

 is only a fraction of the number produced, will result in an 

 automatic reduction of the potential variability, an automatic 

 purification of the type. 



Let us therefore imagine what would happen in a case, where 

 in reality every individual had only one descendant. What be- 

 comes of the total potential variability if in every generation 

 every plant has one daugter- plant ? Let us take the case of 

 self-fertilized plants, to avoid complications. 



A plant will produce as many gametes with, as gametes 

 without a gene for which it is heterozygous, and in respect 

 to every gene for which a plant is impure, it will produce three 

 kinds of daughters 25 % pure for it, having it, 25 % pure, 

 lacking it, and 50 % heterozygous. In other words, half the 

 number of the daughters of a plant heterozygous for a gene 

 will be impure in respect to this gene, and the other half will 

 be pure in respect to it. As half the number of daughters will 

 be pure for every gene, the chance of one daughter to be pure 

 is equal to its chance to be impure. If a plant is impure, hetero- 

 zygous for a number of genes, and it produces one daughter, 



