DICTIONARY 181 after a certain period. Richardson gives only those useful for etymology, which is Littre^s rule for patois. Grimm admits all words at any time belonging to High German or its dialects. The great German dictionaries generally admit dialects, and in this respect are more complete than the French and English. The Chinese give in their standard dictionaries every character known to exist, though many are erroneous, corrupt, vulgar, or local, or are merely improvements proposed by some eminent person. Of the ancient characters, sometimes the pro nunciation, and occasionally the meaning, are unknown, while both one and the other are in some cases completely lost. Johnson omits all words relating to proper names, but they, as well as proper names, often as really belong to a language as any other words. The Philological Society propose that their new dictionary of English, begun in 1856, shall contain "every word occurring in the literature of the language," and " admit as authorities all English books," unwisely excepting " such as are devoted to purely scientific subjects, as treatises on electricity, mathe matics," &c., beginning " with that definite appearance of an English type of language distinct from the preceding semi-Saxon," about the year 1250. Their vocabulary of words beginning with the letter B, printed in 1863, con tains 17,729. The practice of universal admission of words is becoming more generally adopted iu standard dictionaries of all languages. Words can be most surely and quickly found when arranged alphabetically in a single series. Other arrange ments, though sometimes more useful, are not so generally convenient. When it is thought desirable to separate any class of words, they should still be also inserted in their proper places iti the general alphabet. In a large dic tionary a small separate additional alphabet is almost lost, and is usually overlooked by searchers. According to Grimm, the alphabetical arrangement not only facilitates reference, but makes the author s work quicker and surer; " for he who would insert rich contributions must have the places for them before his eyes, and not have to search about undecidedly to find whether the word is already there or not." The order of the alphabet should be that commonly used in the language. Any other makes re ference more sluw and uncertain. Grimm says that the order of the Sanskrit alphabet, adopted by Diefenbach and others, brings confusion rather than light to European languages. The etymological arrangement under roots has been generally condemned by experience. It places all words of the same origin together, so that they can be at once seen, which is often very useful and important, and is a great help in learning a language, as it assists the memory. But a word not belonging to the small number of roots cannot be found unless its root is known; other wise it must be looked for in the index, or if there is none, .sought for by guess-work in many places. And as ety mologies will vary according to fancy or knowledge, no word, as Grimm says, will be sure of its place, and no arrangement is more destructive of the object and use of a dictionary. All its advantages may be secured by giving under each root a list of derivatives. Another, system, more rarely adopted, though perhaps more useful, is that of arranging all words under their leading ideas, so that all those relating to a subject are seen together, and the proper word to express an idea may be found almost as easily as the idea expressed by a word may be found in an ordinary dictionary. It is, in fact, a classed vocabulary of till the words of the language, with the sections arranged alphabetically, and resembles in its purpose the classified index of a bibliographical dictionary, while it is quite as useful and necessary. Boissiere has chosen about 2000 common words, under each of which he gives all the French words evidently attached to it by community of ideas, or by relations of habitual use, cause, means, effect, or any analogy whatever. This part, he says, shows how to call things by their right names, and, as he remarks, great care is taken to teach children grammar, but none to teach them words. In the upper part of each page he gives all the words in alphabetical order, with a reference to the group in which each will be found. Roget, in his Thesaurus, gives under each head (1000 in number) not only the words belonging to the idea, but their opposites, and adds at the end of the book an index of all the words. This system, on account of its very great use and value, might well be made a subsidiary part of a standard dictionary, the groups being placed iu the general alphabet, and a reference to each group being added to each word. The arrangement by terminations is of use grammatically and stenogra- phically, and for making out words of which the beginning is illegible or wanting. A dictionary of rhymes is similar, but not exactly the same, and is of little use except for making verses, and, when the rhymes are perfect, for showing the pronunciation. In the Semitic languages words are commonly placed under their roots, and in MS. lexicons the roots are often arranged alphabetically, accord ing to the last radical. When Lane was making his great Arabic lexicon, he generally had before him eight or ten native lexicons, containing three different arrangements of roots. In Chinese dictionaries the characters are usually arranged under the 214 radicals, which now serve as an alphabet. In former times the number varied, and was much greater. The characters under each radical are further subdivided according to the number of strokes used in making each character, in addition to its radical, or the abbreviation of its radical which each character contains. But no arrangement is attempted of the charac ters having the same number of strokes. Other systems are sometimes used, arranged by tones and endings, and by the characters (about 1040) called phonetics. In the separate articles of a dictionary the arrangement must vary very much with the language, as well as with the word itself. When necessary, the orthography, pro nunciation, and grammatical inflexions of the word should, be given, and any variations of these at different times and places carefully pointed out, as well as the character of the word, such as obsolete, provincial, &c.; and forms be ginning with a different spelling should be placed iu separate articles, with references to the main article. The etymology should be given, referring derivatives to their respective roots; and under each root giving, if not the derivation as far back as it can be traced, at least what Littre calls the secondary etymology that is, deriving it from a word not belonging to the language, as when a French word is traced to a Latin or German word with out proceeding farther; and cognate words should gene rally be enumerated, often with their principal meanings. This gives a primary meaning, but care must be taken that the derivation is a real one, not a mere fancy or guess. The times when the word was introduced or became obsolete should be noted, and the mean ing it bore at first, as well as those which prevailed at various periods. The meanings may be arranged in a series, not merely as they may be imagined to have been logically developed from each other, but as their connec tion may be traced, and can be shown to have existed in actual use; and where this connection cannot be traced, the defect should be pointed out. Sometimes, too, the meanings are, as Johnson says, collateral. In some kinds of dictionaries the explanations may be merely sufficient to identify the word, as in Bilderdijk s Voordenboek voor de Nederduitsche Spelling, or, as in most small dictionaries,
they may merely give the sense. They may also be full