Applying Chain Indexing in Dictionary Catalogues: A Modified Approach

Most students learning chain indexing first meet it as a tool for the classified catalogue, where words and class numbers live in two separate parts. But the technique was flexible enough to serve a very different design too: the dictionary catalogue, where everything sits in a single alphabetical sequence. Adapting chain indexing for this format required Ranganathan to rethink some of his own rules. The result is a modified approach that produces consistent, machine-like subject headings while still fitting the one-part structure that dictionary catalogue users expect. This post walks through how the adaptation works, why it was needed, and how it plays out in real examples.

How dictionary catalogues differ 🔗

A dictionary catalogue is unipartite. Author entries, title entries, subject entries, and cross-references are all interfiled in one continuous alphabetical sequence, much like the words in a dictionary. This is the structure the Library of Congress adopted for its card catalogue in 1898, leading eventually to the Library of Congress Subject Headings.

The classified catalogue is built differently. It is bipartite: a main classified part arranged by class number, plus an alphabetical index part that guides users from words to those numbers. In that design, chain indexing produces class index entries that point the reader from a term to a class number.

The dictionary catalogue has no separate classified part to point to. Its subject entries must stand on their own and lead directly to documents. A subject heading here is attached to a specific book rather than to a class of subjects. So when Ranganathan set out to use chain indexing for this format, the headings could no longer simply refer a term to a number. They had to behave like full subject headings in their own right.

Why Ranganathan adapted chain indexing 🔗

Chain indexing, also called chain procedure, is a method of deriving alphabetical subject headings from the class number of a document. Ranganathan first described it in his 1938 work Theory of Library Catalogue, and it became central to his Colon Classification and Classified Catalogue Code system. The cataloguer starts where the classifier finished, interpreting the class number digit by digit in reverse order to generate verbal headings. Because the work flows directly from the class number, there is no duplication of effort.

Ranganathan was unhappy with conventional dictionary catalogue practice. The enumerated subject heading lists of the time, such as the Sears List and the Library of Congress list, only offered a choice among synonyms and often left newer concepts unrepresented. They gave no consistent rule for selecting the most specific heading, and identical material could end up filed under different headings by different cataloguers.

In 1953, librarians associated with the British National Bibliography pointed out that dictionary catalogue users would value headings derived mechanically and consistently from class numbers, while still keeping the familiar feel of a subject heading list. Ranganathan found chain indexing versatile enough to deliver exactly this, provided he modified the rules. He set these out as an alternative procedure in the fifth edition of the Classified Catalogue Code, published in 1964, under the rules now cited as KSD.

Adapting chain indexing for single-entry systems 🔗

The core challenge is structural. In a classified catalogue, a class index entry can be brief because it simply directs the reader to a number. In a dictionary catalogue, the heading is the destination, so it must carry enough meaning to identify the subject clearly and file sensibly in the alphabetical sequence.

Ranganathan’s modification rests on a simple but powerful idea: each common isolate is treated as a part of its own chain. Instead of one long, unbroken chain producing a single string of terms, the chain is broken into conceptual parts, and a heading is built around each part. Common isolates include geographical divisions, time periods, and form divisions such as dictionaries, periodicals, or bibliographies. In the classified catalogue these often appear as steps within a single chain, but in the dictionary catalogue each one becomes the basis of a separate, self-contained heading.

Sought, unsought, false, and missing links 🔗

The underlying mechanics of chain indexing still apply. The cataloguer represents the class number as a chain where each link has two parts: the notation and its verbal translation. The links are then sorted into types. A sought link is one a user might reasonably search under. An unsought link is unlikely to be searched on its own. A false link carries no clear meaning by itself, often a connecting digit. A missing link is a step in the conceptual hierarchy that the notation fails to express, and the cataloguer supplies its verbal equivalent so the heading reads completely.

The specific subject heading is derived from the last sought link and built upward by taking the necessary and sufficient sought links in reverse rendering. Where space, time, or form isolates appear, the chain is split at those points, and each part generates its own heading. The handling of missing links matters especially in the Dewey Decimal Classification, where broad hierarchy levels are sometimes not reflected in the digits.

Breaking chains into conceptual parts 🔗

A worked example shows the logic. Consider a book on the history of India during the Mughal period, classified in DDC as 954.025 (the exact figure depends on edition; the principle is what matters). The chain runs from the broad class for history, through Asia, to India, and then to the Mughal period.

In the dictionary catalogue, the cataloguer derives a specific subject heading by rendering the sought links in reverse, producing something like:

MUGHAL PERIOD, INDIA, HISTORY.

Because each common isolate is a part of its own chain, the geographical and broader levels each support their own reference headings, such as:

INDIA, HISTORY. For documents in the class and its subdivisions, see the alphabetical sequence under more specific headings.

ASIA, HISTORY. For documents in the class and its subdivisions, see the more specific headings.

Where the notation skips a level, the cataloguer supplies the missing link verbally so that no conceptual step is lost. The heading derived from the last sought link is written prominently, traditionally in capitals or red ink, to mark it as the sought entry.

Examples of application 🔗

Religious texts 🔗

Religious literature is a strong test of the method because the DDC 200 class for religion has deep, uneven hierarchies. A book on the rituals of a particular Hindu tradition might pass through religion, then a specific religion, then a branch, then a practice. The cataloguer identifies which of these levels users actually search under, drops the false and unsought links, and renders the sought links in reverse. The form isolate, if the book is, say, a dictionary of religious terms, becomes its own conceptual part and generates a separate heading rather than being buried inside the main string. This keeps a reference work findable both by topic and by form.

Social sciences 🔗

The social sciences (DDC 300) frequently combine a subject with a place and a period, which is exactly where breaking chains into parts pays off. Take a study of rural unemployment in West Bengal. The subject, the geographical isolate, and any time period each anchor a part of the chain. The specific heading captures the full compound subject, while the geographical part supports a reference heading under the place name. A user searching by region and a user searching by topic both reach the right material, because the single alphabetical sequence carries entries for each conceptual part.

Medicine 🔗

Medical material benefits from the same treatment because users search by organ, by disease, and by procedure. A title on the surgical treatment of a heart condition would yield a specific heading from the last sought link, built upward through the relevant anatomical and disease links. The structured vocabulary of medicine, reflected in tools like the National Library of Medicine’s Medical Subject Headings, illustrates why specificity and consistency matter so much in this field. Chain indexing’s mechanical derivation helps a single cataloguer, or many cataloguers over time, assign the same heading to the same subject every time.

See and see also references 🔗

Cross-references are what hold a dictionary catalogue together, and the modified procedure manages them carefully. There are two kinds.

A see also reference connects related headings. In the modified procedure, subject reference entries, the “see also” references, run from each upper sought link to the specific subject heading. The cataloguer exhausts the terms from the upper sought links, so that a user who enters the catalogue at a broad term is guided onward to the more specific headings where the actual documents are entered. This recreates, within a single alphabetical file, some of the hierarchical guidance that the classified catalogue provides through its arrangement.

A see reference redirects a user from a term that is not used as a heading to the one that is. These handle two situations. First, synonyms and variant terms: a user searching under an alternative name is sent to the chosen heading. Second, and importantly in this method, when a heading would begin with a space, time, or form isolate, a “see” reference is used rather than a “see also” reference, because such a term should not stand as an independent sought heading on its own.

Finally, the cataloguer prepares cross-references for any alternative or synonymous terms found in both the specific and the subject reference headings. The structured use of see and see also references is a defining feature of the cataloguing standards tradition that Ranganathan’s Classified Catalogue Code helped shape.

Strengths and honest limitations 🔗

The modified approach gives dictionary catalogues something their enumerated-list rivals lacked: headings derived consistently and specifically from a classification scheme, with cross-references generated by rule rather than by guesswork. The same subject reliably produces the same heading.

It is not effortless, though. Ranganathan himself noted there are no rigid, mechanical rules comparable to a descriptive cataloguing code like AACR2. Deriving good headings is a work of skill and judgement, and many cataloguers find subject cataloguing demanding for that reason. The method also inherits a known weakness of all chain indexing: it depends entirely on the classification scheme. If the class number is built on a structurally weak scheme, the headings suffer too, and steps can vanish through the “disappearing chain” problem. These limitations later motivated systems such as POPSI, developed to reduce that dependence on the class number.

What do you think? If you were cataloguing a book that combined a subject, a place, and a time period, which conceptual part would you expect most of your users to search under first? And do you think mechanically derived headings serve readers better than a curated list of preferred terms, or is some human selection always necessary?

References

1. https://www.loc.gov/aba/cataloging/subject/
2. https://en.wikipedia.org/wiki/Colon_classification
3. https://www.oclc.org/en/dewey.html
4. https://www.nlm.nih.gov/mesh/meshhome.html
5. https://www.iflastandards.info/