always find the one I want at a single glance. But if I have a library full of books and pamphlets, and I want to be sure of finding one of them at a moment's notice, I must arrange them alphabetically, or topically, or in some other fixed order. When the things themselves cannot be arranged according to any plan, the next best thing is to arrange their names in some fixed order-usually alphabetical-and after the name write where the thing is to be found. In this way

we construct directories and gazetteers and the indexes of books.

6

A second purpose served by classification is to give an easy means of identifying an object when it is found. A suspicious looking person is arrested by the police and they wonder who he is; so they turn up their classified list of criminals, looking up (say) the class Eyes, blue'; then the subdivision Height, five feet eleven inches'; then the further subdivision Fingers, tapering'; and so on, till at last they find the photograph or thumb-mark of the individual in question with his name and record. If the descriptions were not classified the work of identifying the man might be endless.

A function similar to these two, yet distinguishable from them, is to enable us to find or identify a given kind of object. Often we do not know what book or man we want to find, but the classified list can tell us. We want some book or other about colonial furniture, and the subject-catalogue names several; we want a man who can mend china, and the classified list of tradesmen at the end of the city directory gives a number of names and addresses. In the same way if

and want to know

we find a new kind of plant in the woods what it is, we use some botanical 'key,' in which kinds of plants are identified by a series of obvious characteristics, and discover the name (not of that individual as such, but) of that kind of plant. In the same way also a classified list of symptoms might save a young physician much valuable time in diagnosing a disease.

A fourth object of classification is to make it easy to deal at the same time with things that bear any special relation to each other. To this end we put the things themselves or their names together. A grocer puts in the same basket all the parcels that are to go to the same house; a lady writes on the same list the names of the people or shops to be visited on the same afternoon. Here the classification is made for an immediate practical purpose-something must be done about each of the people or things in question, and the classified list helps us to do it. But often, and this is the object of classification in science, things or the names of things are put together because it is desirable to think of them not only at the same time but in relation to each other.* Thus chronological tables, maps, and astronomical charts are made not only to show what happened in a given year or the location of a given place, but also to show the general sequence of events at any given period, the general conformation of a country, the general arrangement of a planetary system. In these cases the relations that determine the classification are those of time and space, but they may be anything: degree of scholarship, in arranging a list of students; degree of strength, in a list of acids; durability, in a list of fabrics or dyes; cause and effect, when the writer on medicine puts together all the causes and symptoms of a given disease; means and end, when he adds a list of remedies.

Scientific classifications are most concerned with relations of resemblance and contrast. When a naturalist, for example,

* It is rather common to speak as though all classification were of one's ideas or concepts of things, but this is a blunder. A psychologist classifies ideas and feelings when he points out their resemblances and differences; e.g., the difference between a feeling of terror and a feeling of contempt; but a naturalist classifies things. It is the business of a psychologist to observe the difference between thoughts as such; but every other scientist is concerned with the difference between the objects that we think of, not the thoughts themselves. To think of the difference between things is not the same as to have different thoughts of them.

classifies all animals into vertebrates and non-vertebrates, he merely asserts that each of a certain long list of animals has a backbone, and in this respect and certain others that follow from it resembles each of the others and differs from all of those not on the list.

But what points of resemblance and contrast must be regarded, and in what order, if we are to make a classification scientific? The answer to this is that no basis of classification-no fundamentum divisionis-is any better than any other in itself. The only general rule is to choose and arrange fundamenta divisionis in the way that best shows the points of resemblance and contrast in which we are interested or likely to be interested. It is just as scientific to classify books by their size, publishers, date, color, type, or language as by their author or subject-matter, and a great deal more so if we are interested in the former and not in the latter.

If we are not interested in any special characteristic of the objects classified the only thing left is to try to get into the same class objects that have a great many things in common and into different classes objects that have very little in common, so "that we shall be enabled to make a maximum amount of aggregate assertion with a minimum number of propositions." To do this we must choose as our first fundamentum divisionis that point on which the greatest number of other noteworthy points of difference depend. It is better to divide all living things into animals and plants than into those which weigh more than a pound and those which weigh less; because the possession of sensation and the power of spontaneous motion† which distinguish animals from plants involve innumerable other points of difference, while the size involves little or nothing more to say that a

66

* Venn's Empirical Logic," Chap. XIII., to which the reader is referred. (Macmillan.)

This of course is the popular distinction. If we wish to be more scientific we should say 'the fact that they require protoplasmic foodstuff, or cannot decompose carbonic acid gas, which distinguishes,' etc.

thing is a plant tells a great deal about it; to say that it weighs less than a pound tells hardly anything.

It is this principle of trying to get into the same class those objects which on the whole are most alike that prevails in the classification of animals and plants in natural history and of books in the subject-catalogues of libraries.

Any system of classification that regards general resemblance is liable to be upset by an advance in knowledge or a change in scientific interests. The common man calls a whale a fish, the zoölogist says it is not a fish but a mammal, because he has found out that on the whole living in the sea and looking like a fish involves fewer other noteworthy characteristics than suckling the young. In Dewey's Library Index illusions are classified with witchcraft and fraud because they involve deception, and for the general reader this is perhaps the best classification; but a psychologist would classify illusions along with ordinary perceptions, because both are interpretations of sensations made in precisely the same way, and he knows that the correctness or incorrectness of the interpretation, which strikes the layman and upon which Dewey's classification depends, is a mere accident and does not involve any further differences in the mental processes. To illustrate the influence of wider knowledge or deeper insight upon a whole system of classification it is only necessary to point out how the old hard and fast lines between genera and species in natural history have been wiped out by the theory of evolution, which shows how new species are being created continually though slowly by the inheritance and consequent accumulation of a vast number of small individual variations.

The fact that a system of classification is likely to be upset by wider knowledge is no reason why it should not be constructed, for a bad classification is better than confusion ; and however bad it may be, it is likely to contribute something toward the attainment of the wider knowledge by which it can be corrected.

When a classification has been made, the resemblances and differences which it indicates can be still further marked and more easily remembered and talked about by Naming. the use of class names. The words vertebrate, mammal, radiate, batrachian are all scientific terms invented for this purpose. But the use of class names was not invented by scientists, for every common name marks out a class of things which it serves to distinguish from all others. Animals and plants, trees and grasses, hawks and doves, were distinguished and contrasted long before science existed. The scientist, as Venn says, usually finds the highest and lowest classes of things already made and named. His business is only to state their most important differences distinctly (which sometimes involves a correction in the classification of an ambiguous kind, such as the whale) and to arrange the intermediate groups.