members of classes. No doubt if it were possible to bestow upon individuals convenient names significant of all their qualities-past, present, and to comesuch names would take the place of 'Proper' Names. But such names are obviously impossible, both because there never is such knowledge of individuals, and also because, if there were, names conveying the knowledge would be quite unadapted for use. What is indispensable, and at the same time possible, in the case of persons or things distinguished by Proper Names, is to attach to them names which indicate definitely and easily which of certain known individuals it is that in any given case is being referred to; and this function is fulfilled by Proper Names.

There are certain Names composed entirely of Attribute Names, Adjectives, or Common Names, and having a maximum of Signification, which have necessarily, or actually, an unique Application-e.g. The longest river in the world, The noblest friendship of antiquity. And in the case of such Adjectives as Shakespearian, Rembrandtesque, which are potentially general, it is quite possible that there may never exist anything to which those terms can be applied except the productions of Shakespeare and Rembrandt respectively. But the majority of Adjectives and of Attribute and Common Names have an application both actually and potentially general; and it is of such words that Definitions are ordinarily most useful. In these cases—that is, where we are concerned with

classes, and connections of characteristics-Definitions may both furnish guidance in application, and also help to bring to mind the characteristics of the things we are referring to.

Certain rules for the framing of Definitions are commonly provided in logical handbooks, of which it may be said that though a Definition which conforins to them may be bad, a Definition which does not conform is certainly not good. These rules are to the effect that a Definition must not be tautological, that it should be expressed in clear and simple and (preferably) affirmative terms, that the word defined and the Definition of it must have identical application, that the Definition must state the Attributes included in the Signification, and those only. It may be added that it is generally desirable that a Definition should be brief; hence the old rule that a Definition (of any Class Name) should be by Proximate Genus, and Differentia is a useful one. When we define Man as Rational Animal or Triangle as Plane figure enclosed by three straight lines, we are defining by Proximate Genus and Differentia. These definitions are both economical and adequate, because the terms Animal, Plane Figure are so significant; and they are obviously in accordance with the other rules given above.

Some of the most important Definitions are of ClassNames; and, as remarked in the previous Section, Classing has a close connection with Definition-for

while Classing consists in grouping together a number of numerically distinct things in virtue of their possessing similar characteristics, those characteristics constitute the Signification which is unfolded in the Definition. And the connection between Classing and Definition on the one hand, and Induction on the other, is also very intimate. For it may perhaps be said that the majority of Class-names are a result of Induction, and may be unfolded into a statement of the interdependence, or inseparable and uniform coexistence, of attributes-since it is by a combination of attributes, and not by merely one attribute or kind of attribute, that we know the objects called by those Class-names. Consider, for instance, such names as Violet, Oak, Squirrel, Water, Air, Circle. From our knowledge of the application and meaning of the word Circle, we may extract, e.g., the proposition that any closed plane figure having every point of the circumference equidistant from a point within it, is a figure of which the diameters are equal. Similarly from a knowledge of the meaning and application of any of the other names instanced, we may frame Universal Propositions which assert a co-existence of characteristics. And every fresh Induction that is summed up in the Signification of a Class-name is, of course, expressed in the Definition of the name.

It is easy to define Definition by saying that it consists in giving the Signification of Names; but we require to know further by what criterion to

decide which characteristics of a thing should be included in the Signification, and the settlement of the Signification is the most difficult and important point in defining. A Definition may give a Proximate Genus and Difference; it may be clear, simple, affirmative, and not tautological, Definition, and word defined may be exactly equivalent; but owing to a mistaken choice of Signification, it may be a very bad Definition. For instance, the Definitions of Man as A featherless biped, or A bartering animal, break no rules, and yet for ordinary purposes are absurd Definitions. Perhaps the only useful general rules that can be given for the choice of Signification are the following: (1) The Signification ought to be as far as possible conformable to usageas regards non-technical words, ordinary usage, and the authorities generally recognised (that is, current speech and writing, standard authors and accepted dictionaries); in the case of terms which are technical or quasi-technical (Slang, Scientific Terms, Provincialisms, etc.), the usage of those recognised as the most competent judges. (It is in an analogous way that we come to know-in as far as we do know-who are the best lawyers, physicians, orators, artists, and so on.) Signification ought to be (2) consistent; (3) appropriate to the purpose in hand-(cf. Sidgwick, Principles of Political Economy, bk. 1. ch. ii. p. 54, 1st ed.); also (4) the characteristics comprised in the Signification should be, if possible, impressive and

distinctive. In all cases, of course, limits are set to the variations of the Definition of any word by its Application. And with regard to the great body of words in any important language, their application is practically fixed, and a person who does not know what this application is does not know the language.

Since any Definition is framed with some definite end in view, and every class of objects has a multitude of common characters, and may be regarded from different points of view, every Class-name is susceptible of a plurality of Definitions, application remaining fixed -e.g. Man may be defined (as by Cuvier, in order to indicate his place in a certain classification of animals) as A mammiferous animal having two hands; or as A rational animal; or An animal capable of speech; or as An animate creature responsible for his actions. There is, however, even with reference to Application regarded as fixed, often a 'ragged edge of usage'—a margin of inconsistency which admissible Definition must exclude. In the case of a 'dead' language, e.g. Greek, there is complete fixity. In a 'living' language with a literature, though there is practical fixity at any given time, yet as manners and customs and life altogether change, and as knowledge increases, and fresh discoveries, fresh analyses, and fresh syntheses are made, some old words have to be modified, and some new words have to be adopted-it is not possible to confine the new wine in the old bottles, to keep a growing, changing body altogether