Through the logic of events forcing those considerations to the front more and more, I anticipate that the larger and rapidly growing libraries will increasingly shun all such systems as the "D. C." and the "E. C.," of which the paradox is certainly true, that the better they are made the worse they become. The scheme of numbering classes recently adopted by Princeton University Library points in this direction, while the reclassification of Harvard University Library, which has been slowly carried forward during the last 20 years or more, represents a complete departure from the idea of any correlation between classes, as indicated in the notation, the order of minor divisions being a numerical sequence easily changed or modified, while each main class bears a mark suggesting no relation to another. For example, the military and naval sciences have lately been reclassified and brought under the designation War, which may be called (to represent a certain harmony with other designations) W-a-r. The location of any main class in the library is subject to change at any time, and is known to the attendants by a chart, which may be somewhat altered to-day, and replaced by a new one with large differences to-morrow or next year. Not that such changes would be made except for real occasion, but under this system, when they are necessary they are not deferred or regarded as hopeless as they must be under any highly organized system. Another obstacle to expansion closely related to elaborate methods of notation is found in the common practice of inserting the call-numbers in catalogs of all kinds, written or printed. When the Boston Public Library was moved into the new building it was naturally supposed that it would be completely rearranged to suit its ampler and entirely different shelf-room, particularly as much fault had already been found with its existing classification, which seemed quite outgrown. But when it is observed how the library was tied to its old numbering by an endless variety of catalogs, printed as well as written, it ceases to seem strange that it was thought best to transfer the old arrangement to the new building, with all its infelicities heightened by its new location and surroundings. And in this respect that library should serve as a warning to others to avoid, by any available means, such an entanglement. If it be asked what means of avoiding it are available, I would say that I am inclined to think that if I were starting with a new library I would try the experiment of putting no shelf-numbers or call-marks in any catalog, but rather have a key by which they could be found by means of the accession numbers which alone would be given in the author-catalog.