nor hooted down. It may be shown within the limits of the most strict and sober argument, that the theory of the origin of language in natural and directly expressive sounds does account for a considerable fraction of the existing copia verborum, while it raises a presumption that, could we trace the history of words more fully, it would account for far more.
In here examining interjectional and imitative sounds with their derivative words, as well as certain other parts of language of a more or less cognate character, I purpose to bring forward as far as possible new evidence derived from the languages of savage and barbarous races. By so doing it becomes practicable to use a check which in great measure stops the main source of uncertainty and error in such enquiries, the habit of etymologizing words off-hand from expressive sounds, by the unaided and often flighty fancy of a philologer. By simply enlarging the survey of language, the province of the imagination is brought within narrower limits. If several languages, which cannot be classed as distinctly of the same family, unite in expressing some notion by a particular sound which may fairly claim to be interjectional or imitative, their combined authority will go far to prove the claim a just one. For if it be objected that such words may have passed into the different languages from a common source, of which the trace is for the most part lost, this may be answered by the question, Why is there not a proportionate agreement between the languages in question throughout the far larger mass of words which cannot pretend to be direct sound-words? If several languages have independently chosen like words to express like meanings, then we may reasonably suppose that we are not deluding ourselves in thinking such words highly appropriate to their purpose. They are words which answered the conditions of original language, conforming as they do to the saying of Thomas Aquinas, that the names of things ought to agree with their natures, 'nomina debent naturis rerum congruere.' Applied in such comparison, the lan-
