In order to attain these advantages, two conditions are fundamental. First, the shelves must, as a rule, be fixed; secondly, the cases, or a large part of them, should have their side against the wall, and thus, projecting into the room for a convenient distance, they should be of twice the depth needed for a single line of books, and should hold two lines, one facing each way. Twelve inches is a fair and liberal depth for two rows of octavos. The books are thus thrown into stalls, but stalls after the manner of a stable, or of an old-fashioned coffee-room; not after the manner of a bookstall, which, as times go, is no stall at all, but simply a flat space made by putting some scraps of boarding together, and covering them with books. This method of dividing the longitudinal space by projections at right angles to it, if not very frequently used, has long been known. A great example of it is to be found in the noble library of Trinity College, Cambridge, and is the work of Sir Christopher Wren. He has kept these cases down to very moderate height, for he doubtless took into account that great heights require long ladders, and that the fetching and use of these greatly add to the time consumed in getting or in replacing a book. On the other hand, the upper spaces of the walls are sacrificed, whereas in Dublin, All Souls, and many other libraries the bookcases ascend very high, and magnificent apartments walled with books may in this way be constructed. Access may be had to the upper portions by galleries; but we cannot have stairs all round the room, and even with one gallery of books a room should not be more than from sixteen to eighteen feet high if we are to act on the principle of bringing the largest possible number of volumes into the smallest possible space. I am afraid it must be admitted that we cannot have a noble and imposing spectacle, in a vast apartment, without sacrificing economy and accessibility; and vice versa. The projections should each have attached to them what I rudely term an endpiece (for want of a better name), that is, a shallow and extremely light adhering bookcase (light by reason of the shortness of the shelves), which both increases the accommodation, and makes one short side as well as the two long ones of the parallelopiped to present simply a face of books with the lines of shelf, like threads, running between the rows. The wall-spaces between the projections ought also to be turned to account for shallow bookcases, so far as they are not occupied by windows. If the width of the interval be two feet six, about sixteen inches of this may be given to shallow cases placed against the wall.