House, about 1838, the cumbrous and antiquated, but I imagine then nearly universal system of press-notation by Roman letters was exchanged for one by Arabic numerals.[221:1] These numbers were nevertheless consecutive, and thus no space was left for insertions. Supposing, for example, that you have three presses standing together, numbered 1, 2, and 3, and respectively occupied by Botany, Horticulture, and Agriculture, it is clear that when your press of Botany is full, you must either duplicate your No. 1, or commence your subject afresh with No. 4. Mr. Watts, however, set his numbers loose, leaving a set of spare numbers after each, for future employment, proportioned to the probable extent of the subject. Thus, in the case supposed, while his Botany would still have been 1, his Horticulture might have been 10, and his Agriculture 15. When more room is wanted for Botany, the other two subjects are moved one press farther on, leaving the press formerly occupied by Horticulture vacant for the Botanical additions. The numbering of the presses is altered, but not the numbering of the books, and the catalogue is not interfered with. The respective subjects thus never get out of due numerical succession; and when, on the opening of the new library in 1857, the books thus numbered were brought from their former confined quarters, and spread over a far larger area, the removal was effected without the alteration of a single press-mark. As the books in any one press may thus come to occupy another, it is, as observed by Mr. Winter Jones, essential that all presses should be exactly of the same dimensions.