The view of the lack of phonetic differentiation is based on the fact that certain sounds of primitive languages are interpreted by the European sometimes as one of our familiar sounds, sometimes as another; they have been called alternating sounds. A better knowledge of phonetics has shown in all these cases, however, that the sounds are quite definite, but that owing to the manner of their production they are intermediate between sounds familiar to us. Thus an _m_ produced by a very weak closing of the lips, and with half-open nose, sounds to our ear a little like _m_, a little like _b_, and a little like _w_; and according to slight accidental changes, it is sometimes heard as one of these sounds, sometimes as another, without, however, being in reality more variable than our _m_. Cases of this kind are quite numerous, but it would be a misinterpretation to adduce them as proof of lack of definiteness of the sound of primitive languages (Boas). In fact, it would seem that limitation in the number of sounds is necessary in each language in order to make possible rapid communication. If the number of sounds that are used in any particular language were unlimited, the accuracy with which the movements of the complicated mechanism required for producing the sounds are performed, would presumably be lacking; and consequently rapidity and accuracy of pronunciation, and with them the possibility of accurate interpretation of the sounds heard, would be difficult or even impossible. On the other hand, limitation of the number of sounds brings it about that the movements required in the production of each become automatic; that the association between the sound heard and the muscular movements, and that between the auditory impression and the muscular sensation of the articulation, become firmly fixed. Thus it would seem that limited phonetic resources are necessary for easy communication. The second point that is often brought up to characterize primitive languages is the lack of power of classification and abstraction. Here, again, we are easily misled by our habit of using the classifications of our own language, and considering these, therefore, as the most natural ones, and by overlooking the principles of classification used in the languages of primitive people. It may be well to make clear to our minds what constitutes the elements of all languages. It is a fundamental and common trait of articulate speech that the groups of sounds which are uttered serve to convey ideas, and each group of sounds has a fixed meaning. Languages differ not only in the character of their constituent phonetic elements and sound clusters, but also in the groups of ideas that find expression in fixed phonetic groups.