History of the General Library


This paper is a revision and expansion of a paper written in July 1986, when I was Desk Supervisor of the General Library, also known as Butler Circulation.

The original purpose of the research was to work out the patterns of the call numbers we had in the library, in order to help us proofread call numbers written by staff as input to our computer circulation system (Marilou) and call numbers written by readers on search request forms. I hoped to compile a ready reference to what call numbers were possible, so that we could identify some, just by format, as either impossible or as unusual and worth checking. One of the outcomes of the work was a Wylbur program, glproof, that tested input to the circulation system and printed warnings of impossible or questionable call numbers for humans to follow up. The other outcome was this paper.

Columbia Classification is the proper name of the variant of Dewey Decimal Classification used at Columbia. I was surprised to see how poorly documented it was. Apparently much of the details had been passed on by word of mouth, or by memoranda no longer in existence. It may be that had someone done this present work before 1967, when Columbia Classification was replaced by Library of Congress Classification in most of the libraries, more documentation would have been found.

The only surviving source is a microfilm, an incomplete and garbled document comprising a mixture of typewritten pages with pages of an early Dewey Decimal Classification (the 4th edition?) with changes written in. In some instances the shelflist itself was the documentation: the cataloger was required to study it, deduce the system, and add new items accordingly.

One thing that helped hold things together was undoubtedly that the same person would catalog in a subject area for many years and would "know" the system. On the other hand this also meant that inexplicable peculiarities (errors really) would go on for years, uncorrected by any authority-- for example, the single-letter "Cutter numbers" someone created for years in 838-839, and the mess of second lines someone else made of 222-229. I have come to believe that English, French and German literatures were due for new call numbers for the 20th Century (namely 826, 835, 845), and that personnel turnover wiped that plan from collective memory.

Much of the information about shelf order was passed on to me by Mr William E Carr, who worked in the Butler stacks from 1950 to 1989, and as Stack Supervisor for about the latter half of that time. Bill's comments on order always matched the shelflist and the microfilm document that I only found while researching this paper, and his uncompromising training of a generation of shelvers was the principal force against the chaos that the peculiarities of Columbia Classification could easily have created.

I hope that this information will be of assistance to the staff who will be dealing with the million books in Columbia Classification for years to come.

Joseph Brennan (MLS), now of Academic Information Systems, Columbia University, 1992.

Name and Locations.

The main library collection at Columbia University has never had a proper name of its own. It is commonly known today as Butler, which is properly the name of the the building in which it is located, or else as Circulation or the Stacks. The term General Library, used in some library documents, is probably the best name to use in an historical account, since the library has been in other buildings, has not always circulated books, and has not always been shelved in bookstacks. The General Library is the collection that goes back to the founding of the university after the American Revolution, and was originally the only library.

The name of the corporation itself was Columbia College in the City of New York until the trustees authorized the change in 1896 to Columbia University in the City of New York. The library was therefore called the Columbia College library before the name change. The reader must understand that references to the "college" before 1900 usually refer to the whole university including the graduate schools.

There are now two types of call numbers in the General Library. The older books are cataloged in Columbia Classification, more commonly called Dewey, though it is really a modification of the Dewey Decimal Classification. The Library of Congress Classification has been used for material cataloged since 1967. They both serve the same purposes:

  • To arrange the library by subject.
  • To give each book a unique number by which it can be filed and located.


The General Library is now in its fifth location:

Dates Campus Building
1784-1857 Park Place one room?
1857-1883 Madison Avenue "room with ten alcoves"
1883-1897 Madison Avenue Library Building: reading rooms.
1897-1934 Morningside Heights Low Library: most of the building.
1934- Morningside Heights Butler Library: central stacks.
The buildings on the two older campuses were torn down long ago. The opening of the Library Building in 1883 coincided with the introduction of the Dewey Decimal classification, almost complete by 1889, when it started to be gradually replaced by Columbia Classification. Books in the Library Building were probably shelved in call number order from 010 to 999 during Dewey's time at Columbia, but different arrangements preferred by the faculty probably took effect starting in 1889, involving both implementing the new Columbia Classification scheme and simply shelving books in a broken order, i.e. with jumps from one group of call numbers to another.

Low Library.

Low Library has been altered into the central administration building of the University, but the floor plan is basically the same as it was. The main (second) floor has a central rotunda, extending up to the dome, which once held the catalog and the reference and circulation desks. The shelving was located all over: in the rooms and hallways around the rotunda (on several levels) and in a stack area below.

The call number sequence from 010 to 999 was extensively broken up because the rooms were assigned to specific academic topics that did not correspond to the classification's numeric order. As time went on, a great amount of effort went into maintaining the subject integrity of the rooms-- the Librarian's reports mention that whole rooms full of books were shifted into other rooms of the appropriate size or location, extra shelves were added in rooms and nearby hallways, and, reluctantly, some less voluminous subjects were made to share their rooms with overflow from next door. Some of the rooms were used for seminars in their topic, and the books were unavailable during the class. Call numbers starting with B prefix and other items now in the Rare Book Library were kept in the stack area or locked rooms and were available by paging.

Two large sizes of books were still shelved separately for a time, folio and quarto. The quartos were finally interfiled with regular-sized books in 1911-12, in an early Low Library shift, and with that the Q size mark, still seen on some volumes, became obsolete. The F size mark is still used today for very large volumes.

Much of the original effect can be seen in Room 210 Low, the Columbiana Collection: it includes a balcony shelving level in the reading room and a small stack area in an adjoining room. Some of the upper level rooms in Low are smaller. Today's visitor, standing in the main rotunda of Low, can still see wooden shelf supports in the highest balcony. A visit to the fourth floor will reveal the use of that balcony as rather strange open-air offices looking out on the rotunda.

Butler Library.

When the General Library moved to South Hall (Butler) in 1934, the grouping by topic, rather than straight 010-999 call number order, was continued in the central bookstacks. The entrance level, stack level 6, had for example English literature (820-829) and history (941-942) and general linguistics (400-419), besides some caged areas. Level 6 is now used for staff workspace, and all the levels have been shifted many times. The notion of grouping by topics rather than using the call number order has continued, so that the library is still not in straight call number sequence.

When shifting was done to make room for Library of Congress Classification in 1966-67, it was planned that books on the same topic in both classifications should be on the same level, as much as possible (for example, D-DX was originally broken into eight segments shelved on four levels). Eventually, space problems forced more arbitrary shifting, so that the library was not quite arranged by either call number or academic subject.

The renaming of South Hall to Butler Library, around 1940, has led to many confusions. Calling a building a "library" at all is awkward because the term is also used for each of the several units inside. The name Butler Hall was already, and still is, in use for a University-owned residential building. The Philosophy library had already been named the Butler Library of Philosophy for a time, and there are still cards in the catalog that use the designation BUTLER for it. Lastly, of course, many people have taken Butler Library to mean the General Library, particularly since it has no other well-known name. The same collection was referred to as Low Library earlier, for similar reasons, but it's not clear what it was called during the "South Hall" period. (South Hall obviously was a temporary name; an even worse one was "New Hall", used briefly for what is now Carman.)

The Annex.

In 1987, some 300,000 volumes of Columbia Classification books were removed from the Butler stacks and placed in a new library off campus known as the Annex. The books were chosen by classes based on the ratio of use to holdings. This was the first time the General Library was split into two locations. The largest chunk was most of 310-389. Using the space that was left, the LC books were shifted to make space for growth. A shift in 1988 took H-J-K-L off level 10, into opened space from the 300's, and expanded the P's on 10 (including P, PE, PN, PR-PS, PZ) into nearly half the floor. Also in that year, the Paterno Library collection was brought into the Butler stacks, but not interfiled-- certain modifications had been made in the Paterno LC sections! In 1989 the D-DX group was consolidated into fewer separate runs, but only by separating them from the floors with the corresponding Columbia Classification call numbers. Future plans may possibly reunite the Columbia Classification into some site out of Butler. The 800-899 group, the most difficult part of the classification, is probably the part that will survive in Butler the longest, since the material, literature, has the heaviest use of older books.

Fixed Location (1784-1883)

Before Dewey, the Columbia College Library was shelved by "fixed location", a finding system in which books are assigned to specific shelves or bookcases, in contrast to the relative order of modern classifications. In the library room used up to 1883, there were ten alcoves that were designated for broad subjects. One account gives the ten divisions, although not the order:

  • Theology.
  • Mathematics, also Fine Arts, Architecture, Medicine, Natural History, Printing and Music.
  • Greek and Roman Classics.
  • Philology and Bibliography.
  • Moral and Intellectual Philosophy, Logic, Rhetoric and General.
  • Literature.
  • History.
  • Biography.
  • Jurisprudence.
  • Astronomy, also Transactions of Societies, Physics, Chemistry.
  • Polygraphy and Miscellany, including Encyclopedias, Journals, Atlases.

The shelves were set up for books of certain sizes: folio, quarto (4to), octavo (8vo), duodecimo (12mo), 18mo, 24mo. A section might have shelves set up for books of more than one size.

How were books found and reshelved? The last catalog, still preserved in the Columbiana collection, only suggests the system. It is a handwritten folio volume arranged alphabetically by author only, showing the size of each book and also a number pencilled in the right margin that apparently indicates the section. That much would lead the reader to the right alcove, and limit the search to shelves of the right size; but unless there was some further undocumented system, like arrangement by author's name, it was then necessary to scan all books of the right size. Books in the collection dating from before 1883 show no sign of old markings, inside or out, and it is not known how they were reshelved or inventoried.

That the section numbers in the catalog are in pencil, while the rest is in ink, suggests that some amount of shifting may have been done to accommodate growth. For each book shifted to a different section, the pencilled entry would have to be erased and rewritten. How "fixed" it really was is unknown; logically, as long as section 2 followed section 1, there seems no reason that the logical sections have to correspond to the alcoves. However, that leap may never have been taken.

Dewey Decimal Classification (1883-1889)

In 1883 Melvil Dewey accepted the post of Chief Librarian at Columbia. This event can be considered the beginning of the Columbia University Libraries as we know it today. Columbia College was starting to build its graduate schools, and Prof John Burgess of the Faculty of Political Science had convinced President Barnard that the library needed better organization and management to meet the needs of graduate education.

It is hard to realize today that undergraduates at that time were taught much like high-school students today: they worked from lecture notes and textbooks and were not required to write research papers or otherwise use the library. No students lived on campus, and classes were held continuously from morning prayer until the afternoon, when the students went home. The Columbia College Library was open only a few hours a day and acquired most of its books by donation. Until the graduate programs began, its importance in the life of the College was minimal. Only a few decades earlier, it had been run by a combination librarian and janitor, and at other times it was a part-time assignment for some unlucky professor.

In the last decades of the 19th Century, many graduate and professional divisions were established at Columbia, and many new subjects, including science, social science and modern literature, were admitted to the liberal arts curriculum, many as electives. Most of Columbia's "schools" date from this period, and in 1896 the name of the corporation was changed to Columbia University in the City of New York. (The former undergraduate School of Arts, that is liberal arts, was then renamed Columbia College, causing a confusion that has plagued Columbians ever since.)

Melvil Dewey had formulated the decimal classification scheme for which he is most famous while working in the Amherst College Library during his junior year of college, 1872-73. He convinced the college to adopt it and stayed on as Assistant Librarian until 1876, when he published the first small edition of the system and reported on it at the first American librarians' conference in Philadelphia. After leaving Amherst, Dewey established a company in Boston called the Library Bureau to manufacture and distribute library supplies, and wrote and lectured on the subject of library organization.

The library problem Dewey found at Columbia in 1883 was the same thing he had seen at Amherst, though on a larger scale: a library growing sharply in size and use but with an inadequate way to locate and arrange the books. A few years earlier, the library had become so hard to use that Prof Burgess, while setting up the first graduate programs, got the Trustees' approval to start a separate library for history and political science. Burgess also convinced President Barnard to put up a new library building (on the Madison Avenue campus at 49th St) and to hire Dewey to reorganize. There are several surviving photographs of the interior of the new Library Building, with its balcony level and high peaked wooden ceiling. The books were shelved around the walls of two or more reading rooms. In some photographs, single letters can be seen marked on the walls; possibly these oriented readers to bookcase locations, much as range numbers are used in bookstacks.

The main library had about 30,000 volumes in 1883, plus some 20,000 more in departmental libraries, some of which were to be merged into the General Library. Of the resulting collection, all but a few hundred were cataloged by Dewey's departure in 1889, and about 40,000 more were added new during the same time. Most of that collection is still in the General Library stacks today. Those with the same binding still have the "Madison Avenue bookplate" in the inside front cover. (To find some most easily, look for books with old bindings in the Classics (87-889) and Religion (200-299) sections.)

The principal advantage of Dewey Decimal and later modern classifications is that it is an indirect reference to the location of the book. This might at first seem a disadvantage: the older "fixed location" system told you what bookcase to go to, while what was called a "relative location" system does not; rather it gives you a location relative to other books, and you need to consult a second reference, the chart of where call numbers are shelved, to locate the book. But the two levels of reference provide a nice separation that makes everything work more easily: as the library changes, and books are shifted to different shelves, only the chart of locations needs to change, while the many many records in the catalog remain correct.

The Dewey classification, by using decimal numbers, provided a simple way to "expand" the system, whether to introduce new subjects or finer subdivisions. It also provided a way to make the system easier to learn-- and Dewey did hope readers as well as librarians could learn the system-- since the same standard expansions could be used in many places; for example the call number for a periodical is the base subject number plus "05". Making it simple to insert new call numbers went hand in hand with making it simple to insert new books among older ones. The overall arrangement could be maintained indefinitely.

The card catalog was also introduced by Dewey and shares the advantages that entries for new books can be interfiled in proper order among the old. Two card catalogs were started in 1883, one for authors and titles and one for subjects. Many of the small handwritten cards for the author-title catalog made under Dewey's tenure are still in the catalog today. Full-size (7.5 x 12.5 cm) cards were not used in the catalog until 1911.

The subject catalog was originally arranged by Decimal Classification (with multiple entries for each book) but was changed to a "dictionary" alphabetical arrangement after Dewey left, about 1890. The classed subject catalog was once again a question of indirect reference; readers would use the printed Relativ [sic] Index to look up their subject words and find what number to look under in the classed catalog. Once again, while a single level of reference-- looking up the subject words right in the catalog-- seems simpler, it is not. Because natural languages contain many ways of referring to the same thing (movies, film, cinema, motion pictures), the dictionary subject catalog has to use a set of standard subject terms that is almost as difficult as call numbers, and well-informed readers still start out using the printed L.C. Subject Headings to figure out what to look under in the catalog, instead of a Relativ Index.

Dewey also started a shelflist, but in a looseleaf book. A shelflist is a catalog in shelf order-- that is, call number order-- used for inventory and to prevent duplication of book numbers during cataloging. The shelflist was converted to a card catalog in 1904-05, when all the entries were transferred to typewritten "halfcards" (5 x 12.5 cm). Many of those original cards have very brief information. Finally, in 1939, five years after the move to Butler, full-size cards became standard in the shelflist.

Another Dewey innovation was using simple numbers for volumes of book sets (simply 14 for example), rather than Roman numerals (like XIV) or numbers with "v." or other prefixes (like v.14, Bd.14, or t.14). Yet another is to use boxes to hold pamphlets or journal issues until they can be bound. Spine marking of call numbers was popularized, if not originated, by Dewey.

Despite his reputation, his post at Columbia was Dewey's first opportunity to run a library, and he moved quickly to put his theories into practice. All the following familiar practices were started at Columbia under Dewey:

  • long hours, 8am to 10pm Monday to Saturday (it had been 11 to 4 or less).
  • circulation records using slips and arranged by call number.
  • a reference department.
  • a repair department.
  • a cataloging department.
  • an author-title card catalog and a subject card catalog.
  • shelf arrangement by subject using call numbers.
  • overdue fines for all users including officers.
  • a recall system to get books back from loan.
  • a training program for librarians which became the country's first Library School.

Unfortunately, Dewey's inexperience as a manager led him into trouble. In the course of introducing so much change, Dewey antagonized an increasing number of professors, and he was forced to leave Columbia after only six years. The faculty did not understand what he was doing and, it appears, he made little effort to explain his activities or involve the faculty in decisions. Some of the faculty were no doubt too conservative, complaining for example that the shelves looked neater with books arranged by size. Other complaints about the budget, loan regulations, admission to library courses, and the Decimal Classification seem to have been born from misunderstandings as much as substance.

After leaving in January 1889, Dewey became the head of the New York State Library in Albany. The odd thing is that, although he himself was removed, most of the changes Dewey introduced at Columbia were maintained by his successors.

Columbia Classification (1889-1967)

Almost the whole collection was classified into Dewey Decimal Classification between 1883 and 1889. About 2,000 books remained out of about 90,000. The new catalog was nearly complete and the books were shelved by Dewey Decimal Classification, probably in straight order from 010 to 999. The call numbers had all-numeric DDC class numbers on the first line, and at this early date most class numbers were just 3 digits. Each book was assigned a unique call number by the use of a second line derived from Charles Ammi Cutter's author table, which assigns a "Cutter number" composed of one or two letters followed by one or two numbers, based usually on the author's name.

Immediately after Dewey's departure in January 1889, the classification was modified, the beginning of a decades-long process. The annual Librarian's report does not mention until 1894, and then only briefly, that the classification was being rearranged to group the books more according to "academic study". Examination of the books still in the General Library and of the printed acquisitions lists shows that nothing was cataloged in Literature from January to March, and then starting in April 1889 the new classification scheme for the series 810-899, including the Goethe G, was in effect.

This first reclassing seems to be the outcome of a battle between Dewey and the faculty over the proper way to arrange books in the study of Literature. In Literature, Dewey Decimal Classification sorts works in each language first by "form" (poetry, drama, fiction...) and then by historical period. Works written by one author are split up if they are of different "forms". A book of Shakespeare's sonnets would be shelved with other English poetry, but apart from his plays, and apart from books about the man himself.

Instead, the new Columbia Classification of 810-899 sorts works in each language first into historical periods and then by author. This way, all books by and about Shakespeare form a continuous block on the shelf; but on the other hand, a reader searching for, say, poetry will have to choose a period and then find it among other types of literature. The new scheme even includes some related material that Dewey places outside 800's. Classes 420-499 for language and linguistics were cancelled and instead books about each language fall in 810-899 next to the literature of that language. Classes 870-889 for Latin and Greek were changed to include classical history and antiquities, because the academic study of the Classics typically includes more than just the language. Due to a total absence of documentation, the peculiar use of two-digit call numbers (85,86,87,88,92) and that Goethe G class cannot now be explained.

The Dewey Decimal Classification call numbers in 810-899 were not fully converted until 1920; the Librarian's reports say some 37,000 spines were re-marked in 1918-20 due to reclassing.

Close examination of "Madison Avenue bookplates" on the inside front cover of books in the 810-899 section will reveal the change from Dewey Decimal to Columbia Classification. Typically the first two digits were left intact, and the rest shows signs of erasure (ink erasing, which roughens the surface) with the new call number written over. Some thrifty person evidently couldn't see replacing the old bookplates, so the evidence survives. The oldest bookplates include a space for "other topics" of which the book "treats": any classification numbers appearing there are true Dewey class numbers, and they show how the book was listed in the classed subject catalog. It is harder to see the old number showing through on the spine, but it is there if the binding is old enough. Books that have been rebound since 1890 have lost all trace of the old call number. Around 1910 the "official" spot for the call number was moved from the bookplate on the inside front cover to the title-page verso precisely to ensure that it survives rebinding.

The new scheme for 810-899 not only changed the numeric classes but also introduced a set of alphabetic schemes for subdivisions that was applied in certain other call numbers. Notably, some Philosophy call numbers in 180-199 are like 800-899, and the Columbiana collection (about the University itself) has a C class that, like the Goethe G, is a truncated call number (G is 833G, and C is 378.7C). A few alphabetic conventions appear throughout, like suffix A for bibliographies, while others appear only in specific classes here and there.

Columbia Classification is much more difficult than DDC. The Librarian's report as early as the 1910's complained that the boys then used as pages were unable to shelve books in the proper place, or find them when shelved properly. As previously mentioned, all call numbers under Dewey were in the simple form of numerals over Cutter number (like 123 over B2, or perhaps 123.45 over B2). A number carelessly written down as "123B2" could be easily resolved into "123 over B2" and found on the shelf. The new scheme introduces letters in the first line, and both the division into lines and the distinction between capital and small letters became significant. Thus for example one finds that "823 over Sa", "823SA over something", and "823Sa over something" all may exist, are different, and are shelved some distance apart. To this day, users have difficulty locating books, and staff require careful training to learn the system.

Because of the many changes, mostly done by 1925, the system was no longer Dewey Decimal but what became known as the Columbia Classification or "modified Dewey", though still called "Dewey" in common usage. The modifications even involve special variations for the individual libraries-- for example, Philosophy works in the Philosophy Library have a scheme different from that in the General and Burgess Libraries. The updated editions of DDC issued after 1900 could not be used at Columbia because of the modifications, putting the burden of maintaining the classification entirely on the staff at Columbia. The difficulties of CC, caused by its many peculiarities and poor documentation, evidently put it beyond the comprehension of later staff, and it was not much changed during its last 40 years.

The Columbia Classification modifications of 810-899 had been done to make call number order reflect the desired shelving order, but no other similarly ambitious project was undertaken again, despite the evident desire to shelve books in an order different from the existing classifications. Other than certain smaller sections that were reclassed, the simpler solution taken was simply to shelve books in a broken order, requiring somewhat lengthy finding guides. For example, history and literature of a country is often shelved in the same general area, which fights against the classification scheme.

Perhaps the amount of labor involved over that one project had been more than anticipated, or at least more than could be done again. The continued growth of the collection meant plenty of work had to be done just to catalog new material, and of course any reclass would involve more and more books as the collection got larger. During the long period from 1934 to 1967, Columbia Classification stagnated, and then it was abandoned for a completely different classification scheme, that of the Library of Congress.

Library of Congress Classification (1967-present)

Neither the peculiarities of Columbia Classification nor the evident desire to arrange books in a different order from the classification accounts for the radical step taken in 1967, to abandon further Columbia Classification cataloging and use the Library of Congress's system instead.

The attraction of LC was the savings in cataloging costs. Libraries using LC do not merely use the Library of Congress's classification but their actual cataloging-- that is, a particular book has the same call number in all libraries using LC. What makes this possible is computer networking. New books arriving at Columbia today are checked first at a terminal against databases maintained by RLIN (Research Libraries Group) and OCLC, listing books cataloged by libraries across the country. If the book is already cataloged, the call number and entries for the catalog are usually just copied, the process called "cataloging with copy". Only if the book is not found will Columbia do "original cataloging", the more laborious process of analyzing the book so that it can be assigned a call number and catalog entries. Up to 1967, under Columbia Classification, all books had to go through original cataloging.

It is sometimes rationalized that LC has finer subject divisions and is therefore more suitable for a university library. Actually that is not the case. The reason the LC classification tables fill a small bookcase, compared to Dewey Decimal Classification's three volumes, is that LC tends to be an "enumerative" system, meaning that nearly all possible call numbers must be listed one by one, with their meaning. Dewey, by contrast, makes extensive use of standard tables. The fineness of divisions is about equal in both. Most DDC libraries restrict use of the tables to keep call numbers short (most are of course small public libraries where that is entirely appropriate), but at a length similar to LC call numbers, the subject divisions would be about as fine. Both classifications are a bit poor on any topic not popular in the USA a hundred years ago-- consider the length of call numbers in either classification for Russian Literature or for television.

The Library of Congress itself was established to provide information needed by members of Congress in their legislative work. Under the copyright law, copies of all books registered for copyright in the US are submitted to the Library of Congress, and while not all are kept the Library has become despite itself the "national library" and by far the largest in the country.

The Library of Congress originally used a 44-category fixed-location system devised by Thomas Jefferson (!), but work began in 1897 on a much-needed new system. The outline of the Library of Congress Classification (LC) is based on the "Expansive Classification" done for the Boston Athenaeum about 1876-79 by Charles Ammi Cutter. Cutter's classification was never widely used but his system of letter-and-number classes drew some attention at the time and was even considered for the Columbia College Library in 1882. Cutter's author number tables did find wide acceptance and were used at Columbia to form second lines of DDC and CC call numbers.

The first LC classes developed were Z Bibliography and Library Science in 1898, and E-F American History in 1901. Both use only one letter in the first line. Beginning with class D in 1902, two-letter combinations were used. An outline of the whole system was published in 1904 and the classes were more or less developed by the 1930's. Class K Law lagged far behind, the first part KF finally being published in 1969; the many books with incomplete K call numbers predate the completion of K. Each of the LC classes was developed separately by different people and the system therefore lacks the unity of Dewey. The classes share only the general style of call number and a few basic principles (like general works before specific).

The original scheme of an LC call number was (one or two letters) + (number from 1 to 9999) + (.) + (author number), something like "A 123 .B2", which can also be written unambiguously as "A123.B2". The subject was defined by the elements before the period.

Subjects were individually assigned numbers based on "literary warrant", the principle that each subject should be subdivided only to the extent appropriate to the number of books in that subject. In other words, plenty of numbers (subdivisions) were allotted where LC expected to have plenty of books, and topics expected to have fewer books got fewer numbers. Over time, the plan broke down as the popularity of topics changed. The limitation of an enumerative system is that space is sometimes not left for new topics, or for subdivision of old ones, although there are numerous gaps in the numbers to allow for new subjects. Literary warrant also results in a bias based on LC's collection policy, which is broad, but a little light in science, non-Western European literatures and some other topics that might be found in university libraries.

Deviations have been made from the standard call number. Some that occur in more than one class are: three letters (like KFN); numbers with decimal subdivisions (like E 183.5); placing a date between the subject and the author number to arrange books by time; omitting the author number for certain sets of works; and placing dates at the end of the call number to differentiate editions.

One very common deviation creates a logical problem in shelf arrangement, the so-called "double-Cutter" call number. In this type, the first letter-number code after the period is part of the subject classification, and the second letter-number is the usual Cutter-like author number. To make up an example, in FA 123 .R5, the subject is FA 123, while in a double-Cutter call number, FA 123 .A7 R62, the subject is FA 123 .A7. Logically, subject FA 123 should come before any of the more narrow subject FA 123 .A7, but libraries that attempt to shelve that way run into great difficulty; unless one knows the logic of the classification it is hard to see why FA 123 .A1 A1 would be put after FA 123 .Z99.

Because each classification has further peculiarities of its own, it is not possible to make very many useful generalizations about LC as a whole. The standard guide to LC is Immroth's Guide, which was written and published independently of the Library of Congress. The Library of Congress have not published an official guidebook of their system. Editions of the tables are so infrequent that the lists of additions and changes have become as long as the table itself, and books of cumulated revisions are put out by a private company rather than LC themselves. Contrast Dewey Decimal Classification, with its regular editions and associated publications.

Computer Circulation Systems (1968-present)

It may seem absurd to speak of computer databases and Columbia Classification in the same sentence, but since the books do circulate, they have had to be entered into computer systems.

Before 1968 or 1969 circulation records were kept on cards filed by hand in a very large array of open-topped wooden boxes. For some time toward the end the McBee Keysort system was used. McBee cards had holes around the edges that could be clipped according to a code to indicate due dates and the like. To extract cards for which action was needed, such as sending overdue notices, a rod was inserted into a chosen hole in a stack of cards, and those clipped at that hole would fall out when the rod is lifted, thus selecting those cards. Other than this, filing was done by hand. The cards were kept in call number order so that the file could be used to confirm whether a book was on loan, but this meant no other data sort existed. Staff worked on the circulation file during most of the hours the library was open.

The original home-made circulation system at Columbia, Marilou, was developed in 1968-69. A version using keypunch data entry was put into use in 1970, and the more familiar version that went through two generations of special terminals was used from 1974 to 1991. Marilou used a limited character set that did not include lower-case letters, so some of the fine points of Columbia classification order were lost.

Marilou did observe the line division that is so important, however, by storing the numbers in special format. Columbia Classification numbers had a required period in fourth position, so for example 945, 945B26 and 945.1 were collated by normal rules as 945., 945.B26 and 945.1, where blank, A-Z and 0-9 occur in that order. The only flaw was that upper and lower case letters could not be distinguished.

Marilou sorted LC numbers by a similarly ingenious method: the format for what was considered the "first line" was AAANNNN.NN, where A is any alphabetic or blank, and N any number or blank, and the position of the decimal point was held constant by using blanks, which collate ahead of printed characters. Thus, using a + here to indicate where the blanks occur, D+++++2+++, DA+8825+++ and DJK1234.56 would collate correctly.

The NOTIS systems are no better for Columbia Classification. While upper and lower case are finally recorded in the system, they sort the same, so there is no improvement with that. As a step back, line division is not used at all in arranging, so NOTIS's ordering diverges significantly from shelf order and from the logic of the system. Possibly a special line-division character could have been inserted that would have forced correct order; possibly not, since NOTIS is a vendor's system not designed for the unique problems of Columbia (both normal Dewey Decimal and Library of Congress do not need to observe line division). It looks now like the special problems of Columbia Classification will never be addressed.


Columbia University Libraries history:

  • Columbia University in the City of New York. Report of the Librarian. New York: Columbia University, 1909/10-1952/53. [Reference R027.1C72 over B].
  • Columbia University in the City of New York. Annual Report of the President. includes the Librarian's report. New York: Columbia University, 1882/83-1947/48. [L378.7CC over D].
The "official reports". Mostly pretty dry stuff, especially so many years later. Dewey's first report, for 1883/84, is well-written and worth a look.

  • Columbia College in the City of New York, Library. Catalog.... New York: manuscript, 1877. [Columbiana].
  • Columbia College in the City of New York, Library. Catalog of the books and pamphlets in the library of Columbia College, New York. New York: Printed for the College, 1874. [L378.7CPG over L].
  • Columbia College in the City of New York, Library. List of additions to the library, 1888-1891. New York: Printed for the College, 1888-1891. [L378.7CPG over L2].
The handwritten catalog of 1877 was the basis for the printed 1874 catalog; the 1877 date refers to the last of the additions lists appended in the back. This was the last catalog before the card catalog. The 1877 manuscript is the copy used in the library and has the shelf locations pencilled in. The list of additions 1888-1891, like the catalogs, can be used to locate General Library books in the collection at that time; note in April 1889 the first Goethe G books and the earliest 810-899 books classed in Columbia Classification.

  • Linderman, Winifred G. History of the Columbia University Library, 1876-1926. Ph.D. dissertation. New York: typewritten, 194?. [Reference R027.1C72 over E].
  • Mudge, Isadora Gilbert. Columbia University Libraries, Reference Dept, Thirty Year Report, 1911-1941. New York: typewritten, 1951. [Reference R027.1C72 over B2].
  • Straka, Mildred. An historical review of the Cataloging Dept of the Columbia University Libraries, 1883-1950. M.S. thesis. New York: typewritten, 1951. [Library Service D027.7 over C729].
Linderman is the best general book. Read especially the account of Dewey's career at Columbia, and that of the librarians just before and after him. "Miss Mudge" sums up her experiences in the Reference Dept with some inside stories and manages to explain a considerable amount about the General Library as well, such as the origin of the maltese cross stamp for 1-week loan. Straka is of less interest for the General Library although some information about the cataloging systems is recorded only here.

Columbia Classification:

  • Columbia University Libraries. Columbia Classification Scheme. New York: mostly typewritten, 1942. [Library Service microfilm F74]
  • Dewey, Melvil. Decimal classification and relativ index.... 4th edition. Boston: Library Bureau, 1891. [Library Service D025.4 over D5731]
See the introduction to the Columbia Classification tables later in this paper for a description of the microfilm. The Dewey 4th edition may be used to contrast contemporary "true Dewey".

Library of Congress Classification:

  • U.S. Library of Congress, Subject Cataloging Division. Classification... [volumes for each class or subclass]. Washington DC: Library of Congress, [various dates]. [Cataloging Dept; also Library Service]
  • Library of Congress Classification Schedules: a cumulation of additions and changes... [volumes for each class or subclass]. Detroit: Gale Research, [various dates]. [Cataloging Dept]
  • Chan, Lois Mai. Immroth's guide to the Library of Congress Classification, 3d edition. Littleton CO: Libraries Unlimited, 1980. [Z 696 .U4 I5 1980; also Library Service, Cataloging Dept]
Chan is an excellent introduction to LC. Since Columbia uses standard LC, there is no need to duplicate Immroth's guide here; refer to it for all the basics. Notice how long it is! The first two items are the actual classification tables themselves, a huge set of books.