LIBKABLES. 201 LIBAABIES. impracticable because of frequent reeopying, and even with the small sheets 10 X 25 cm. holding only 20 titles each the labor has led many li- braries lo take the greater risk of cards. In the relative system, now almost universally preferred, shelves require no numbers, the class numbers being the sole guide. For greater legi- bility these are often printed large on movable label-holders. If shelves are numbered, the plan should be so comprehensive that numbers signify position as well as sequence. Here as every- where numbers must run from top to bottom and left to right, to read as the columns of a news- paper are read, e.g. 2435.8 might mean second lioor, fourth room or face, third tier, fifth shelf from the top. and eighth book from the left-hand end. The unit figure should uniformly represent a height the same distance from the floor, so that one seeking a shelf ending in 5 knows on what level to look. •J.>IT0RIAL Depaktmext. Feather dusters merely redistribute dust. Damp cloths can be used in many places and moist sawdust scattered over the floor in sweeping collects the dust in little balls. Books are cleaned by slapping them sharp- ly together over a shallow pan of water which catches the dust. Some libraries have an air-shaft with strong exhaust, so that books may have the dust jarred off and carried away by an air-cur- rent. A recently invented portable exhaust with rubber hose sucks out all dust thoroughly by run- ning the nozzle over the tops of the volumes. The best protection against fire is an ample supply of fibre or metal buckets kept filled with water, and kept in plain benches with hinged covers. Raising the cover exposes the entire line of full pails, which are less likely to spill than when taken down from high racks. Hand grenades lose their power with age. As books are easily damaged by water, dry powder extinguishers are specially adapted to libraries. Insurance au- thorities have tested and approved three of the scores of appliances for mixing sulphuric acid quickly with soda and water, thus making the best of chemical extinguishers. A pattern should be chosen without rubber pipes, which harden with age, or valves, which are apt to stick. But prevention of fire is more important than extinc- tion. The greatest danger is from imperfect electric wiring. For unlighted corners candles are safer than oil lamps. Steam or hot water heat with a single fire and flue is much safer and cleaner than stoves, furnaces, or fireplaces. Library Classification. Books alone are no more a library than boxes of type or dictionaries of words are poems. A collection of books must be classified before it deserves the name library. Classification is putting like things together. Each book, pamphlet, clipping, map. or other item goes with any others like it on a carefully sys- tematized plan, so that matter most closely allied and oftenest used with it will precede or follow closely. Only thus can all reasonable demands of readers be met fully and promptly. Tlie vital importance of classification has long been recog- nized. Alexander Bain says. "To leain to clas- sify is in itself an education." But practical difficulties were so grave as largely to neutralize advantages. It is an almost endless work to pre- pare a complete scheme, and when done it never wholly suits the maker, much less any one else. To avoid the inevitable delays and confusion of these elaborate systems, some libraries were ar- ranged in order of acquisition, some by authors like a directory. Usually there was coarse classi- fication by subjects, and librarians and readers did the best they could to find their resources by aid of bibliographies, subject catalogues, and in- dexes. Some classifications had no indexes. Others referred to pages of scheme on which sub- jects might be found. The book number itself indicated a fixed location on a particular shelf and had to be altered as often as growth made it necessary to move that subject. The great desideratum was a system that would do away with the expensive necessity of renum- bering books whenever their location was changed. This was provided in 18TG by the publication of the Decimal Classification and Relative Index, which showed by the same number both subject and location. This scheme divided the field of knowledge into nine main classes, numbered 1 to 9. Encyclopsedias. periodicals, etc., so general in character as to belong to no one of these classes, are marked and form a tenth class. Each class is similarly separated into nine divisions, general works belonging to no division having naught in jjlace of the division number. Divisions are sim- ilarly divided into nine sections and the process is repeated as often as necessary, the full tables covering some 20,000 topics. Books on the shelves and cards in the subject catalogue are arranged in simple numerical order, all class nimibers being decimal. Since each sub- ject has a definite number, it follows that all books on any subject must stand together. The tables show the order in which subjects follow one another. 512 algebra preceding 513 geometry, and following 511 arithmetic. Of this E. C. Richardson says in his Classification, Theoretical and Practical (pages 199-200), published, by Seribners in 1901 : "This system lias probably had more vogue than any other bibliographic system ever pub- lished save possibly that of Brunet. Taken as a whole and regarding the substantially unchang- ing form and notation, among the multitude of derived systems with minor variations, it is un- doubtedly true that no system ever invented has been applied to as many libraries (probably at the present day several thousand) as this. In many libraries considerable changes have been made, but in the majority it remains practically unchanged. It is now beijig adopted very gen- erally on the Continent of Europe by booksellers even as well as libraries, and is of late, through its adoption by the Brussels Institute (for inter- national bibliographic work) , having a very zeal- ous propaganda by its converts, especially in France and Italy. Many of the most noteworthy jiartial classifications of the present day are avowedly founded on and are enlargements of this system. The system itself is supposed to be in some way an adaptation of Bacon, but the relation is hardly to be discovered and it really should be counted as independent. The reasons for its deserved poiuilarity are to be found : (1) In an intelligent and consistent application of the decimal notation (not new with Dewey, but first by him vigorously and consistentlv applied) ; (■2) in the grasp of mnemonic possibilities of this situation: (3) in the practical, intelligent, and often up-to-date management of the remoter sub- divisions of the, in some places, somewhat arti- ficial, larger subclasses: (4) in the fully printed schedules with their 'relative index,' which more
