Denton, William. “Book Numbers.” Feb. 2003. http://www.miskatonic.org/library/book-numbers.html.
I wrote this essay in February 2003 for a course in the theory of classification at the Faculty of Information Studies at the University of Toronto. The introductory cataloguing course had, of course, covered Dewey and Library of Congress, but the details of book numbers were dismissed with a quick "and you can add on something to indicate what volume or copy it is." I wasn't clear on exactly what a book mark was, or title mark or collection mark, and who said you could use them, and who made the rules. I wanted to know, so when I got the chance I wrote this ten-pager.
The professor gave it a good mark. There is one factual mistake: Dublin Core metadata doesn't have to be detailed, in fact, it can be very loose and vague, depending on who makes it. The professor pointed out where I wasn't clear about Kate Sanborn Jones's last name and where I quoted a chronological book number table and then used an example that wasn't actually shown in the quote.
There is very little written about book numbers (see References at the bottom for a list of all the major books on the topic). The Internet has almost nothing, so I thought I'd put this up in case anyone is curious about the bits that make up a call number and where Cutter numbers come from.
(Note: I had three footnotes in the first paragraph. They're at the bottom of that paragraph, not the document. Also, there is a reference to an attachment, which isn't part of this web version, which is too bad, because it would show you what some Cutter tables look like.)
Abstract: Book numbers (also called item numbers) combine with collection numbers and class numbers to form call numbers. Book numbers are a way of organizing and ordering books about the same subject that share the same class number. They collocate books on the shelf in a helpful manner and provide unique call numbers for every item in the collection. Book numbers are a minor but important part of classification and cataloguing. Brief surveys are made of their history, Cutter and Cutter-Sanborn tables, alphabetical and chronological orderings, Ranganathan's faceted book numbers, and Library of Congress call numbers. The future of book numbers is surveyed.
For decades, shelflisting has been the stepchild in the family of actions involved in the marking of books for shelf arrangement. The notation that represents the terms of a classification (and therefore the subject of the books) has received far more attention. That is as it should be, for the classification and its notation arrange an entire library. Though used throughout the library, shelflisting arranges only one class at a time. Nevertheless, to pay scant attention to shelflisting is to leave the frosting off the cake. The cake can be eaten, to be sure, but with less ease and appreciation.
-- John P. Comaromi (1981)
1. What is a book number?
Book numbers are parts of call numbers, together with collection numbers and class numbers. Book numbers come at the end, and arrange books about the same subject so that they can be given useful order on the shelf and a unique location in the collection. The collection number, if used, indicates a major grouping within a library or library system, e.g. REF for reference or J for the juvenile collection. The class number of a book tells what it is about, but many books can be about the same thing and share the same class number. Book numbers are different for each book having the same class number and will make the full call number completely individual.1 Just as different classification schemes lead to different class numbers, so do the different book number systems lead to incompatible book numbers. The library of the Faculty of Information Studies at the University of Toronto (at which copies of all the books in the bibliography can be found) uses the Dewey Decimal Classification (DDC).2 Their policy for making book numbers is to make an author number from the main entry,3 then add a title mark equal to the first letter of the title (personal conversation with Joseph Cox, 1 February 2003). For example, Satija and Comaromi (1992) has the call number 025.428 S2523B MC.
MC is the collection number, indicating in which of the many campus libraries it can be found. (The name has since changed, which is confusing.)
025.428 is the class number, indicating the subject is shelflisting. There are many books in the library about shelflisting. How to tell them apart? With a book number: for this book,
S2523 is from Satija, and
B is from Book Numbers. A Library of Congress (LC) example is a copy of David Copperfield by Charles Dickens, found in the main University of Toronto library at PR 4558.A1 1947 ROBA 1.
PR 4558 is class number, and stands for for David Copperfield, while
A1 indicates that copies are arranged in chronological order and
1947 is the year of this particular edition. The library has appended its own collection mark (
ROBA) and a copy number. S.R. Ranganathan devised his own, very detailed, faceted book number system, and in Colon Classification (1964), he helpfully includes a complete call number on the copyright page: 2:51N3 qN60.
2:51N3 is the class number (
2 for Library Science,
51 for Generalia Bibliography/Technical Treatment/Classification,
N3 for Colon Classification), and
qN60 is the book number (
q to show the form is "Code,"
N for 1900-1999,
60 to make the year 1960).
(Footnote 1: "Numbers" in this sense are also called "marks:" book marks, class marks, collection marks, etc. Book numbers are sometimes called item numbers, to cover other forms of material, but here the focus will be on books and shelves.)
(Footnote 2: The University of Toronto's library catalogues can be searched at http://www.library.utoronto.ca/. All but two use Library of Congress classification.)
(Footnote 3: "Author" will be used in the rest of the essay for "main entry," because that is how users think, and because the idea of "main entry" did not exist when book number schemes were first created.)
"Book number" means slightly different things to different people. Comaromi (1981) says a book number is a "combination of author numbers, Cutter numbers, author letters, and any other shelflisting device." Chan (1994) defines item number as "[t]hat part of a call number which designates a specific individual item within its class." Sartap and Comaromi (1992) say, "Class numbers alone produce groupings whose size depends upon the depth of the library classification and the closeness with which the classification used is applied. To organize or provide order within a class grouping, documents are given a further notation called a book number." Ranganathan (1964) said the book number "of a book is a symbol used to fix position relatively to the other books having the same Ultimate Class.... The Book Number of a book individualises it among the books sharing the same class number."
Book numbers do not usually reflect the subject of a book, but instead are based on external attributes such as author name or year of publication. Satija and Comaromi (1992) say that book numbers "may be based on one or a combination of some of the attributes of the document, such as author, title, language, year or place of publication, physical size, and physical make-up." (Book numbers may sometimes reflect a subject-related aspect of a book, such as when it is a volume of criticism. Ranganathan used a
g at end of the call number for that, and that the Library of Congress system uses its own indicators in some cases. This brings together on the shelf books and their criticism, a very helpful collocation.) Ranganathan (1964) said the book number "may consist of one or more the following successive Facets: Language Number; Form Number; Year Number; Accession Part of Book Number; Volume Number; Supplement Number; Copy Number; Criticism Number; and Accession Part of Criticism Number."
In general, book number = author number + title (or work) mark + edition mark + date of publication + volume number + copy number + anything else library policy dictates. Call number = class number + book number, with the collection number at the start or end.
2. Why are they needed?
Book numbers give a unique shelf location to each book in a collection. They bring a defined and consistent order to all books on a given topic, an order that may apply more generally to all subject groupings in the library. Depending on the size of the collection and the depth of classification, it may happen that very rarely do two books collide and share a class number, so book numbers are not thought necessary. Satija and Agriwal (1990) forcefully object to such imprecision:
For a rigorously fine arrangement of books, book numbers are indispensable. Yet their value is debated if not totally doubted. A sizeable number of librarians do not value them highly in shelf arrangement, no wonder then if these are meted out a step-motherly treatment in some libraries. Literature on them is thin and rare. Even those who use book numbers think of them as merely and adjunct--a tool of the perfectionist only. Yet their value in impeccable shelf classification cannot be underestimated. In close access libraries these have comparatively more value in pinpointing the location of books. And for collocation of host and associated books, and to bring together a book and its sequels, the book numbers are quite indispensable. Their utility and importance becomes more pronounced libraries using broad classification such as Rider's International Classification or even the DDC. It is not to suggest that in use with depth classification systems these are less desired. Whatever be the size of the library and the kind of classification used, book numbers add [the] last touch to the ultimate shelf arrangement.
In 1937 Bertha Barden listed six important reasons for using book numbers (though bar codes have made numbers four and six less relevant now):
[B]ook numbers in addition to class numbers are needed:
- To arrange books in order on the shelves.
- To provide a brief and accurate call number for each book.
- To locate a particular book on the shelf.
- To provide a symbol for charging books to borrowers.
- To facilitate the return of books to the shelves.
- To assist in quick identification of a book when inventories are taken.
Broad classifications will make many books share the same class number. Deep classifications will mean fewer do. Some libraries, such as in elementary schools, may use very broad classifications, perhaps Dewey to the tens. What happens when books collide? Ordering the books alphabetically by author's last name (the most basic of book numbers, though perhaps an invisible one if the name is only on the book cover and not on the spine label) seems obvious. For some collections this is enough. It will not be enough in large or specialized libraries or any library where precision and detail are valued, in order that the needs of both the librarian who organizes material, and the user who searches for it, are best served.
3. Early History
Before the DDC, books had fixed locations on the shelf. Call numbers identified where on which particular shelf a book could be found. Books might be grouped by subject, but new books were added at the end of a shelf, not mixed in. Melvil Dewey introduced relative location: his call numbers tell what a book is about, and organize the books "in terms of their relationship to one another without regard to the shelves or rooms where they are placed" (Chan 1994). When a new book comes in, it can be placed between two existing books. One small effect of Dewey's work was the need for book numbers, because two books could now share the same class number. Dewey first tried using the author's full name, but then switched to accession numbers, ordering books by their arrival in the library: "3428.3 and 3428.28 were respectively the third and twenty-eighth books in the child care class" (Comaromi 1981). This might bring a rough chronological ordering, but it was not good enough.
4. Cutter and Sanborn
Making rules for book numbers was a busy field in the late nineteenth century (Lehnus 1980, Comaromi 1981). Several schemes were devised, some using author names and some publication date. One man's work has survived: Charles Cutter's. The two- and three-figure Cutter tables and the Cutter-Sanborn table are the basis for modern author numbers, and "to cutter" is a common expression when cataloguing.
Cutter began turning author names into letter-number combinations in 1879, first published about it in 1880, and by the mid-1880s, after several editions and with Dewey's imprimatur, his system had become common (Satija and Comaromi 1992). A copy of the two-figure author table is attached (it is long out of copyright, but updated versions are still in print). The rules are neatly specified: "Use one letter for words beginning with consonants (except S), two for words beginning with vowels or with S, three for words beginning with Sc" (Cutter n.d.). For example, to turn Denton into an author number, go to the fragment on the table that precedes it alphabetically: "Deno 43." If there are no other names in the Deno-Dent range already, then Denton would become D43. If Denovich had already claimed D43, Denton could become D435, leaving room for 431-434 to be used for other nearby names. The numbers are to be treated as decimal fractions, as in the DDC. Bruce Sterling would become St4, Albert Einstein Ei6, and Joseph Schumpeter Sch8. The three-figure table, which published in 1901 after Kate Sanborn's work, is an extension of compatible with the two-figure table, and is suitable for larger libraries (Comaromi 1981).
Kate Sanborn (later Jones) was asked by Cutter to revise the two-figure table but ended up making a new table that was incompatible. Author numbers would be one letter plus one to three letters. The distribution of numbers through the letters of the alphabet was different from Cutter's original work. Her table of name fragments and numbers was large, with more than 12,000 numbers, but Cutter's later three-figure table was larger still, with more than 20,000 (Lehnus 1980).
Cutter also included instructions for complete book numbers: one could add title marks based on the first one or two letters of the title to distinguish different books by the same writer; copy numbers could be indicated with a 2 or 3; translations could be marked with the initial letter of the language of origin; biographies would be filed by subject but with an additional letter to show the writer; commentaries would add a -Y; dictionaries and concordances would add a -Z (Cutter and Jones n.d.)
The Cutter and Cutter-Sanborn tables are still popular, but though the detail may help with the distribution as names are turned into numbers, constantly referring to lookup tables is tiresome. Various other marks are still used with DDC much as Cutter set out: title marks being the first letter of the title; biographies indicated by inserting a "z" between the Cutter number and work mark; commentaries done the same way but with a capital "Z;"edition marks shown with edition numbers or year of publication; copy and volume numbers shown with a "c" or "v" (Chan 1994).
5. Library of Congress
The LC classification removes the worry about how to specify book numbers. The rules are laid out in the system because it specifies almost complete call numbers. For book numbers, they use simplified Cutter numbers and a variety of indicators that show when books are collected works, commentaries, being ordered chronologically, etc. Their numberings are chosen to fit their shelving needs. "The Library of Congress classification is a very close classification and if numbers from either the Cutter-Sanborn table or the Cutter three-figure table were used many call numbers would be unnecessarily long. Therefore the Library of Congress developed its own author number table to be used with is classification schedules" (Lehnus 1980). The table has five short rules and can be found in any book that covers LC cataloguing (e.g. Chan 1994).
(1) After initial vowels for the second letter: b d l-m n p r s-t u-y use number: 2 3 4 5 6 7 8 9 (2) After initial letter S for the second letter: a ch e h-i m-p t u w-z use number: 2 3 4 5 6 7 8 9 (3) After initial letters Qu for the second letter: a e i o r t y use number: 3 4 5 6 7 8 9 For initial letters Qa-Qt, use: 2-29 (4) After other initial consonants for the second letter: a e i o r u y use number: 3 4 5 6 7 8 9 (5) For expansion for the letter: a-d e-h i-l m-o p-s t-v w-z use number: 3 4 5 6 7 8 9
D 644.M32 2002 is the LC call number for Paris 1919: Six Months That Changed the World, by Margaret MacMillan.
D 644 indicates History (General)/World War I/Peace/General works,
M32 is MacMillan (a short Cutter number), and
2002 is the year of publication. This is a fairly simple call number. Some classes include "A" and "Z" Cutter numbers, where "a span of Cutter numbers at the beginning or end of the alphabetical sequence is assigned special meanings" (Chan 1994). Works about a person have this: A2 is appended for collected works by date, A4 for letters by date, A5 for speeches and lectures by date, etc. This will collocate such collections in chronological order after books written by the person in question. That a book is a translation, or other such details, is shown by a second Cutter number.
Work marks can be added to the year of publication ("a" for the second edition that year, "b" for the third), but that is the limit of detail in LC call numbers. QA 76.76.T49 G64 1997, for Writing GNU Emacs Extensions by Bob Glickstein, is about as complicated as they get. It has a decimal in the class number and two Cutter numbers (albeit no work letter for the year of publication). Libraries can add collection numbers, and copy and volume numbers, as needed.
6. Chronological Ordering
All of the book number systems seen so far (except for Dewey's first attempt, using accession numbers) arrange books within a topic by author's last name. This is an obvious and sensible method of ordering. It will bring together all the books by one writer on the same topic, a helpful arrangement. However, the chronological ordering will be completely disrupted. Books on evolution by Charles Darwin will be followed by those by Richard Dawkins (metaphorically apt, since he is an important Darwinist). What of the 150 years between them? There is little room for names between Darwin and Dawkins, but they span the entire history of thought on evolution. A chronological ordering would put Darwin first (or near it), and moving across the shelf would show how the science progressed up to the latest work in the field.
Satija and Comaromi outline the arguments for and against chronological book numbers. In their favour, they are simple; there is no confusion when the same writer gets different Cutter numbers in different classes; when a book has multiple authors, one is not favoured over the others; the development of a subject can be they are an aid to weeding out of date books; there are no problems making Cutter numbers for non-European names. Against them, they note there is no well-developed system for using chronological numbers; that the arrangement is more helpful for organization than retrieval; that it separates different editions of the same book; and most importantly, that people remember names, not years.
Chronological book numbers are as old as author numbers. W.S. Biscoe, a disciple of Dewey, devised a system that the great man admired. James Duff Brown and Fremont A. Rider created their own systems (Satija and Comaromi 1992). All have rules for how to turn year numbers into shorter combinations of letters and numbers, which is their great fault. Biscoe's rules included (Satija and Comaromi 1992):
B up to 999 (B33 = A.D. 33, B 685 = 685)
C 1000-1499 (C236 = 1236, C423 = 1432)
D 1500-1599 (D20 = 1520, D 85 = 1585)
Y 1980-1989 (Y5 = 1985, Y6 = 1986)
"Biscoe was not concerned that his table would reach its limit with the letter Z in the year 1999, and commented that before this limit was reached someone would have devised a better scheme" (Lehnus 1980). Brown's Extended Date Table covered 1450-2125 using pairs of lower case letters (Satija and Comaromi 1992):
1970-1995 ua-uz (up = 1985, uq = 1986, ur = 1987)
All these systems have an enormous failing: they are needlessly and overwhelmingly complicated. The Indo-Arabic numerals are the best system for counting anything in western civilization, and they are recognized all around the world by speakers of many other languages. To save one or two characters at the cost of making library users memorize an obscure conversion mechanism is no bargain. Proper names are made up of letters, and trimming them down to author numbers may not be as informative as using the full name, but they are still easy to read and ordered alphabetically. There are no such advantages to turning 1985 into
One particular advantage that chronological orderings have is that they do away with problems about making author numbers for non-European names. Cutter's tables try to spread numbers around so that the distribution over the range of names is proportional, but when Chinese, Indian, Arabic, and other names are in the collection, the tables do not work as well. Librarians in those cultures have made their own tables. Indeed, "[i]n view of the multiplicity of languages almost every Indian script and language has its own author number table for book numbers" (Satija and Agriwal 1990). There are different calendars used around the world, but far fewer than the number of languages, and the Gregorian calendar is widely known. Arranging chronologically leads to fewer problems when managing a varied and multicultural, multilingual collection.
7. Colon Classification Book Numbers
S.R. Ranganathan's faceted Colon Classification (CC) is extremely detailed, precise, and informative, as is the book number system he made to go with it. In his Colon Classification (6th ed.) (Ranganathan 1964) he defined these terms:
03 The Book Number of a book is a symbol used to fix position relatively to the other books having the same Ultimate Class.
030 The Book Number of a book individualises it among the books sharing the same class number.
031 The Book Number of a book is the translation of the names of certain of its specified features into the artificial language of ordinal numbers, specified and elaborated in the rest of this chapter.
03012 The Book Number consists of an intelligible concatenation of one or more of the following symbols: the twenty-four Roman Capitals got by omitting I and O; the twenty-three Roman smalls got by omitting i, l, and o; the punctuation marks dot, hyphen, semicolon and colon; and the ten Indo-Arabic numerals.
0302 The Book Number may consist of one or more of the following successive Facets: Language Number; Form Number; Year Number; Accession Part of Book Number; Volume Number; Supplement Number; Copy Number; Criticism Number; and Accession Part of Criticism Number.
He also defined this formula for book numbers:
[L][F][Y][A].[V]-[S];[C]:Cr[Cr#] [L] = language number (can be left out) [F] = form number (can be left out) [Y] = year (the most important; comes from a table: e.g. N=1900-1999, P=2000-2099) [A] = accession part of year number (if more than one book from the same year) [V] = volume number [S] = supplement number [C] = copy number Cr = g indicates a volume of criticism [Cr #] = accession part of criticism indicator
This is the most orderly book number system discussed. The language and form facets can be left out if they are the default (e.g., English books), leaving the year of publication, which Ranganathan felt was the most important facet, first. He believed strongly in chronological ordering.
In many of the schemes of Book Numbers the name of the author is used to individualise a book. In the Colon Classification the Year of Publication and some other characteristics also in some cases, are used for the purpose. For except in Literature and in the case of the classics in any subject, where the author is made into a class in the Colon Classification, it is felt that the Year of Publication will be a more relevant and helpful characteristic than the name of the author for individualising a book. If we remember that the library is a growing organism, it is more often the year of publication that determines the value of a book in all cases except the ones excluded above. The majority of readers are interested in the latest books in an ultimate class, while antiquarians may be interested in the oldest books. Most pedestrian works cease to have value to an ordinary reader at the expiry of ten to twenty years after publication. Any work with long-persisting value is likely to come out again in a new edition (Ranganathan 1964).
For example, a 1987 book written in Urdu about Indian history would have call number V,44 168 M7 where
V,44 is the class number and
168 M7 is the book number (
168 showing it is written in Urdu and
M7 standing for 1987) (Satija and Agriwal 1990). A more complicated book number is
15w1K8.1g which indicates a book of criticism of volume 1 of a 1968 book of Sanskrit verse:
15 for Sanskrit (language),
w1 for form (verse),
K8 for year (1968),
.1 for volume 1, and
g for criticism.
As with all chronological systems, one writer's books on a topic will not be collocated. Further, different editions of the same book will be separated. A book published in 1968 with a second edition in 1975 would have copies filed at
L5, with many books in between. Ranganathan said that if editions were to be kept together, it could be done by treating later editions as copies of the first. In this example, the first edition would be
K8 and the second
K8;L5. This brings together different editions of the same book, but at the same time destroys the chronological ordering that shows the progression of knowledge. Ranganathan knew that not everyone would like complete chronological ordering, but he felt its advantages outweighed its faults. He liked the ordering, and ways of turning numbers into shorter coded forms, so much so that in Colon Classification (1964) he includes two tables for doing this. One is especially for book numbers and turns four-number calendar years into three-character letter and number combinations.
Book numbers are not as interesting as class numbers. The methods for making them are straightforward and can be defined in a couple of pages, while classification systems can take many volume to define class numbers. Nevertheless, book numbers are important. They are the final step in classification, the one that gives a book a unique place in a collection. Books about the same subject need to be ordered in a useful and helpful manner so that users can find the books arranged in an informative way that will save their time in choosing a title.
The most important part of choosing a book number system is whether to arrange books alphabetically by author or chronologically by year of publication. Most people are familiar with alphabetical arrangements, and this is what the Cutter and Cutter-Sanborn tables and LC do. Chronological ordering has its partisans, but its unfamiliarity would make it uncomfortable to most users. Both systems attempt to save space (especially on the spine label) by turning names and years into coded short forms. That is useful for alphabetical arrangements, but confusing and unhelpful for chronological orderings, where using the full year is much clearer. Chronological book numbers may be useful when dealing with a collection that includes many non-European names.
Little has been written about book numbers. A search in the Library and Information Science Abstracts database on "book number*" with a filter to exclude International Standard Book Numbers showed only seven articles between 1990-2003. "Item number*" showed four articles. Though mostly ignored, book numbers should be appreciated by all librarians who value precision. It will be interesting to see how book numbers fare in the future, when more and more documents are available electronically, not on a shelf. Even shelved books can be virtually reordered by doing different sorts on fields from the catalogue database. Metadata systems like the Dublin Core will mean that everything is catalogued and classified with so much detail that there is no more need for book numbers. Book numbers, which bring so much detail and accuracy to DDC, LC, or CC, become blunt and unrefined instruments of arrangement when compared to very thorough description and classification systems.
Barden, Bertha. Book Numbers: A Manual for Students with a Basic Code of Rules. (Chicago: American Library Association, 1937.)
Chan, Lois Mai. Cataloging and Classification: An Introduction. 2nd> ed. (New York: McGraw-Hill, 1994.)
Comaromi, John P. Book Numbers: A Historical Study and Practical Guide to Their Use. (Littleton, Colorado: Libraries Unlimited, 1981.)
Cutter, Charles A. Two-Figure Author Table. (Chicopee Falls, MA: H.R. Huntting Co., n.d.)
Cutter, Charles A. and Kate Emery Jones. Explanation of the Cutter-Sanborn Author Marks: Three Figure-Tables. (Springfield, MA: H.R. Huntting Co., n.d.)
Lehnus, Donald J. Book Numbers: History, Principles, and Application. (Chicago: American Library Association, 1980.)
Ranganathan, S.R. Colon Classification. 6th ed., with amendments. (Bombay, India: Asia Publishing House, 1964.)
Satija, Mohinder Partap and S.P. Agriwal. Book Numbers (Some Indian Methods). (New Delhi, India: Ashok Kumar Mittal, 1990.)
Satija, Mohinder Partap and John P. Comaromi. Beyond Classification: Book Numbers, with Special Reference to Chronological Book Numbers. (New Delhi, India: Ess Ess Publications, 1992.)