It is, however, not quite accurate to say that the organisation persists unchanged from generation to generation. The offspring is similar to the parent--that is, the organisation has been transmitted unchanged. But the offspring also differs just a little from the parent--that is to say, the organisation is modified by each transmission. In these two statements we formulate in the simplest manner the law of organic variability. Organisms may obviously be arranged in categories in such a way that the individuals in any one category resemble each other more closely than they resemble the individuals belonging to another category. We may, by experimental breeding, produce an assemblage of organisms all of which have had a common ancestor, or a pair of ancestors. Now the individuals composing such an assemblage would exhibit a close resemblance to each other, such a resemblance as our categories of naturally occurring organisms are seen to exhibit. We should also find that the individuals of our naturally occurring assemblage would be able to interbreed among themselves, just as in the case of the experimentally produced population. It may be concluded, then, that the naturally occurring population is also the product of a pair of ancestors. This inter-fertility, as well as the close morphological resemblance of the individuals, are the facts on which the hypothesis of the common origin and unity of the assemblage, or species, is formed. The morphological resemblance between the individuals, either in the natural or the artificial populations, is not absolute. If we take any single character capable of measurement we shall find that it is variable from organism to organism. This important concept of organic variability may be made more clear by a concrete example. Examination of a large number of cockle shells taken from the same restricted part of the sea-shore, and therefore belonging presumably to the same race, will show that the number of the radiating ridges on the shell varies from 19 to 27, and that the ratio of the length to the depth of the shell also varies from 1 : 0.59 to 1 : 0.85. In the former case the most common number of ridges is 23, and in the latter case the most common ratio of length to depth is 1 : 0.71. These are the characteristic or modal values of the morphological characters in question, and the other or less commonly occurring values are distributed symmetrically on either side of the mean or modal value, forming “frequency distributions.”[29] The value of the first character changes by unity in any distribution: obviously there cannot be a fraction of a ridge; and this kind of variation is called “discontinuous.” The value of the second character may change imperceptibly, and it is therefore called “continuous,” a term which is not strictly accurate, since in applying it we assume that the numerical difference between two variates may be less than any finite number, however small.