THE LIBRARY JOURNAL OFFICIAL ORGAN OF THE AMERICAN LIBRARY ASSOCIATION VOL 7 JAN,-DEC.1882

and as Mr. Dui's new numbering-base consists of the same characters, I think it is proper for me to state that I am not a convert to his system, and that there is nothing in common between our two methods of applying these characters. I use the characters for a specific purpose, as explained in this section, for individual works alone, whereas Mr. Dui's plan is, if I understand it correctly, to use the numbers and letters interchangeably in place of the ordinary decimal system of ten characters. I first used the 35 title characters in precisely the same way as explained in this section, in 1878, whereas Mr. Dui's system was not published until 1879. Without expressing an opinion on the merits of his peculiar manner of using numbers and letters, I would say that I have not found it necessary to avail myself of his method of economizing characters, nor have I found it desirable to amplify them by means of the decimal system as applied in the plan of my friend Mr. Cutter, in which III stands before 2, and 299 before 3. The body of my system of notation consists of Arabic numerals only, used in the usual way, where 11 is placed after 110, and 299 before 300. I have found the ordinary decimal system of Arabic figures fully adequate to meet all the requirements, and the proof is that I do not need as many characters as any of the schemes devised since 1871. Mr. Dui, for example, requires seven characters to number 1,000 novels, and eight for a collection of 10,000. Mr. Edmands's fiction catalogue has seven in most cases without counting the class letter, which is understood. Mr. Massey generally uses seven, and Mr. Cutter, in a table he sent me, finds that he needs seven and eight characters in 40 per cent. of a class of 3,000 works, although, according to his calculation, his average for a library of 250,000 volumes would not exceed six or seven characters. In all these cases duplicates are excluded. While the number of characters would probably remain the same as at present if the libraries represented were twice or three times as large, the fact remains, that the largest number of characters are used in the most popular classes, such as Fiction and Biography.

To show the capacity of my system, it is

only necessary to consider that each ideal shelf admits of 36 works. A class with 300 ideal shelves has, therefore, room for 10,800 works, and a department for (10,000 X 36) 360,000. The whole scheme of 24 departments consequently provides for (360 X 23) 8,280,000 works, or 16,560,000 volumes, if we may assume that each work will average two volumes. Allowing 90 per cent. for waste, we shall still have room for nearly 2,000,000 volumes, without counting duplicates, or, in other words, for more books than are now in any library in the world.

D. Duplicates and Editions. Duplicates, that is, other copies of the same work, should have a letter in addition to the title-numbers. Another edition of a work, that merely differs in form, should be treated as a duplicate; but, if it contains additions or changes in the matter, it should receive a separate titlenumber. As a rule, it will be found that editions in fiction must be treated as duplicates, and, in other classes, as separate works. As there may be several copies of editions treated as duplicates, and as it is desirable to keep them together, the first edition should be numbered a, the second i, the third o, and the fourth u. This allows eight copies of a, and six each of i, o, u. Should there be more than four editions to be treated as duplicates, the letters e, l, r, and x could be used for numbering the first copy of each. This would provide for six editions with three copies, and two with four copies each. Assuming that four editions of David Copperfield are in a library, they would be designated as follows: N.1722.d would be the first copy purchased; N.1722.da, an edition in two volumes, of which N.1722.db, N.1722.dc, and N.1722.dd would be other copies; N.1722.di would be an edition printed in Boston, and N.1722.do and N.1722.du, editions printed in Philadelphia and Chicago, of which N.1722.dp and N.1722.dv were other copies.

E. Size.

In my original plan I distinguished one size from another by using a different series of numbers for each of the four sizes, 12°, 8°, and

folio. For example, a work in 12°, by Smith, would be numbered 412; in 8°, 747; in 4o, 883; and, in folio, 983. It is better, however, to use the same number, or the same number increased by a uniform addition, for the same combination in every class, and to distinguish the unusual sizes by class letters, as I have done in the present plan. The 12° and 8° sizes, which I formerly arranged in two alphabetical series, I have thought it best to combine in

one, not only to facilitate reference, but because the difference between the ordinary 120 and 8° is too trifling to serve any practical purpose by the separation of these two sizes.

To make the foregoing explanations more intelligible, I have added a number of examples, showing the application of the system in various classes. They immediately follow the author tables.