How Alphabetical Indexes Are Organized

An alphabetical index is not merely a list sorted by a computer. It is a map from the words a reader is likely to seek to the places where a subject is treated. Alphabetization makes that map predictable, while headings, subentries, locators, and cross-references make it useful.

Letter by letter and word by word

English-language indexes commonly use one of two ordering systems. In letter-by-letter order, spaces and most punctuation are ignored until a difference in letters appears. The entries are compared as though their words ran together. In word-by-word order, a space ends a unit, and a shorter complete word sorts before the same letters continued.

The exact display can vary when a style guide applies additional rules, but the central distinction remains: letter by letter compares a continuous character string, while word by word gives the word boundary priority. A publication should choose one method and apply it consistently. Readers rarely need the method announced unless the index is large or specialized, but indexers and software do.

Spaces, punctuation, and capitalization

Capitalization usually does not determine alphabetical position. “Apple” and “apple” sort together, with another rule used to resolve a tie if both must appear. Initial articles such as “a,” “an,” and “the” may be ignored for titles, but personal names and established phrases need context-sensitive treatment.

Spaces, hyphens, apostrophes, quotation marks, and other punctuation are often ignored or treated specially for sorting. Thus a hyphenated form may be alphabetized as if closed up. This is a convention, not a reason to remove the punctuation from the displayed heading. The reader should see the correctly written term even if a normalized form controls its position.

Diacritics may be ignored at an early sorting level in English, then used to break a tie. Other languages have their own alphabetic orders: a letter with a diacritic may be distinct, a digraph may have special status, or an entirely different script may govern. Unicode code-point order is not a substitute for a language-aware collation rule.

Numerals and initials

Indexes handle numeral-leading headings in several defensible ways. They may sort numbers before A, spell them out for placement, or create a separate numbers-and-symbols group. A title beginning “1984” might therefore appear before A or under N as though read “Nineteen Eighty-Four.” The visible title should remain accurate, and a cross-reference can support the alternative a reader may try.

Numbers also create a choice between numeric and character sorting. Numeric order places 2 before 10; character order places “10” before “2” because 1 precedes 2. In most reader-facing indexes, numeric order is more natural for comparable quantities.

Initials are commonly alphabetized as the letters they represent, disregarding periods and spaces: “T. S. Eliot” under E for Eliot in a name index, or an acronym under its written letters in a subject index. Names require a consistent rule for surname, particles, compound surnames, and organizational names. Authority records or a house name file can keep repeated names stable.

Prefixes and similar headings

Prefixes such as Mac, Mc, de, van, and O’ should be handled according to the chosen style, language, and the person’s established name. Some older systems grouped Mac and Mc as if both were “Mac”; many modern systems alphabetize the letters as written. Software defaults should not silently override identity or an editorial policy.

When headings repeat the same opening word, subentries expose structure more clearly than a long run of inverted phrases. Instead of separate main headings “Paper, coated,” “Paper, recycled,” and “Paper, sizes,” an index might use:

The main locator points to a general discussion; subentries identify distinct aspects. A string of undifferentiated page numbers—sometimes called locator “dumping”—makes readers inspect each page. Subentries should add meaningful distinctions without fragmenting a simple topic unnecessarily.

Cross-references

A see reference redirects from a term not used as a heading to the preferred term: “colour. See color.” It should not send the reader somewhere empty or merely repeat the same wording. A see also reference connects a valid heading to a related one: “paper sizes. See also printing.” The original heading retains useful locators.

Cross-references support synonyms, acronyms, variant spellings, broader and narrower terms, and likely reader language. They also need maintenance. Circular references, references to deleted headings, and long chains make an index feel broken even when its alphabetization is correct.

A worked miniature

Suppose a short manual discusses page dimensions, paper weight, recycled stock, and printing. A useful index might read:

This example redirects a familiar phrase to the heading where the subject is actually developed, groups repeated concepts, and distinguishes general treatment from specific aspects. Alphabetical order helps retrieval, but the intellectual work lies in choosing what the entries mean.

What makes an index useful

A strong index uses the reader’s likely vocabulary, distinguishes passing mentions from substantive treatment, and supplies enough substructure to answer questions quickly. Its locators use a consistent notation and point to the smallest useful span. It is tested for missing targets, duplicate variants, circular references, and headings stranded without locators.

Conventions differ among style guides, subjects, software, and languages. The best rule is therefore not “alphabetize everything the same way,” but “declare a suitable collation policy, apply it consistently, and build reader-oriented access around it.” Sorting arranges the entries; indexing designs the route to information.

Sources and further reading

• NISO, ANSI/NISO Z39.4-2021 Criteria for Indexes
• Unicode Technical Standard #10, Unicode Collation Algorithm
• American Society for Indexing, Best Practices

Related reading

• A Practical Markdown Punctuation Reference
• A Concise Guide to Roman Numerals