181

Classification of the Sci
10

then is no human purpose, there must be some agency, such as natural selection, for example, which acts like a purpose; or else no natural classification is possible. If we do not know what this agency is, or only know it in a general way, we may nevertheless by close study of large bodies of fact be able, with more or less approach to accuracy, to determine where the exigencies of this quasi-purpose have caused the lines of demarcation to be drawn between the different classes.

It is, however, a general rule that natural classes, as just defined, are apt to merge into one another, so that it is impossible to say in every case whether a given individual belongs to one or to another; or, at any rate, if this can be ascertained it will only be by the aid of inessential characters. This is susceptible of mathematical demonstration; and it can also be copiously exemplified by instances drawn from artificial objects whose natural classification is