Reading for fine indexing


There are many modes of reading books: acquiring information; studying to memorize exam subjects; taking pleasure in craftily contrived literature; escaping in lighter fiction; skim-reading for particular passages or references. Proofreaders read word by word, watching for errors or cross-checking to ensure copies are identical; reviewers critically assess strengths and weaknesses; actors memorize and interpret; young children painstakingly plod. People about to index books read them at least three times, in three different ways unlike any others.

I speak of the indexing of narrative texts, not technical ones - histories, biographies, and other printed accounts of human life and works - a species of indexing coming to be known as 'narrative indexing' (or perhaps 'fine indexing,' as opposed to applied or technological?).


The first time indexers read the text in hand, it is straight through, with unbroken concentration, to get to know the whole of the material on which they will work - as a potter does with clay. They need a clear, overall comprehension of the whole work. For this reason, indexers really should be supplied with all the parts of the volume that will help to give this full insight - those prelims, illustrations, maps, notes, and other appendages that are often withheld from us as 'they aren't available yet, and won't need to be indexed.' Included in the index, perhaps not - but included in and contributing to the indexer's understanding, they should be.

Two further advantages derive from indexers' first reading the pages through untouched by human hand, pencil, or highlighter. We see how fully each character and theme will be developed, so know which will need many subheads to divide them, with each reference specified; and know that however minor and negligible the early appearances of some characters and themes may seem, they will later loom large and the first glimpses need to be retrievable. The second advantage of full first reading is inoculatory, so that we are not in danger of being distracted from our work later by sheer interest in the development of the text.


Once we have seen the text steadily and whole, we proceed to the entry-making reading. This is a more disjointed journey through the text, reducing it to its component parts and strands - not merely name or capital-letter spotting. In deciding which elements to make index entries for, we are not extracting selected terms and leaving an unindexed mass, but reducing the entire text to denser units, with only the vaguest, most general passages not included under some broader heading.

We may underline or highlight our selected terms, or write in the margins those we have devised, or subheads we have added. This helps us to find the references again at later stages, but breaks the rhythm and makes the work even more fragmented, as we read a bit, mark the text, and make the index entry on our cards or slips or computer screen. The simplest texts, or most technical, or most closely subheaded, may have all the terms needed in the index already there, so we need only spot and copy, for instance, in a handbook on primary education: classroom, maths, textbooks, curriculum.

Indexing a biographical account of a war-torn childhood, though, would be more complex. We would list the names of all places and people mentioned - the central character, members of his or her family, neighbours -- and know from our preliminary reading which of these would eventually amass too many references to be left undifferentiated, so we add subheads, 'childhood,' 'in Silesia,' 'during World War II,' as seems appropriate in each case. Some broader topics, too, will need to be listed - 'Silesia,' 'Children,' 'Evacuation,' 'Food rationing,' perhaps; even abstract concepts not named on the page, such as 'Courage,' 'Loyalty,' 'Fear,' 'Patriotism.' We watch and list those characters and events in the foreground, while noting also the minor characters who remain constantly or recurrently in the background, as well as broad, overall, possibly implied but unstated topics and themes. We focus on one paragraph at a time to see what features in it, while keeping in mind the unit of the chapter and the relation of all its elements to the whole book. And meanwhile ..., we must keep noting, and meanwhile ... We are analysing and documenting all human life and relationships at several levels - assuredly, a complex business.

As we work through a book in this second reading our lists grow ever longer: both the overall, single list that will form the ultimate index, and the subsidiary ones for each major character and theme, to be integrated into the larger. As characters and topics recur after their first entry, we pick up their cards or recall their entries in our computer file to add to them. We may add a new page number to others, or add a new subhead, or scan the whole entry so far to see whether the new mention should be subsumed under an existing subhead. My lists for major characters come to look more and more like rough jottings for an essay.

This entry-making reading is disjointed in several ways. Not only is our manner of reading and working at this stage fragmented; we are consciously unravelling the carefully composed text. We are trying to understand the author's mind in order to undo his or her work of synthesis.


Having reduced the text to discrete elements, we must reunite and assemble them in index order, editing and arranging in our third reading or readings. The overall sorting process is simple, alphabetical, whether performed by card shuffling or computer program. There are niceties as to word-by-word or letter-by-letter arrangement, how to treat hyphens, symbols, numbers, Mac, Mc, and St; as for these, in the United Kingdom the British Standards Institution's Recommendations will take care of them for us.(1) The crucial questions in fine indexing at this point are in what terms to express - encapsulate - the events of the book, and how to arrange the subheads in the long entries.

For the terms used in main headings - usually proper nouns - indexers can simply copy the text. Subheads, though, are usually our own, supplied by us as précis of the passages. Recording/presenting/interpreting indicates attitude, and indexers must ensure that the attitudes implied in the index accord with those of the text. We must choose to use 'Terrorists' or 'Freedom fighters,' 'Crowd' or 'Mob,' 'Street riot' or 'Protest,' 'Refugees' or 'Illegal immigrants.' We must try to find terms indicating passionate relationships outside marriage that are not censorious (not easy, this). The wording of subheads must convey attitude - as shown by a nicely contrasting pair in the index to Elizabeth Longford's 1976 biography, Byron:

Byron, George Gordon, 6th Lord: ... his courtship and marriage, 60-79

Byron, Annabella, née Milbanke, wife of B. ... vicissitudes of her marriage, 71-7

The term vicissitudes does not occur in the text.

The order in which to arrange subheads in narrative indexing is a highly vexed point. The development of the text is usually chronological. As we work through the book, adding entry after entry, the result, untampered with, will be in order of occurrence in the narrative - fairly chronological sequence, perhaps with glitches when the first chapter looks forward to the whole, or when there are flashbacks, or when developments in the life of one person are told only at the point at which they affect a more important one. But order of entry is indisputably the easiest order in which to leave each block of subheads, and has a certain justifying logic. Its results have sometimes been scarifyingly castigated, as when Bernard Levin devoted an entire article to 'the full, almost heroic awfulness' of the index to Ian Ker's biography of Cardinal Newman (Clarendon Press, 1988), complaining particulary that 'the hundreds of references (under the entry for the main charcter) are not in alphabetical order at all, but only in the order in which they appear in the book.(2)

The course advocated by Levin, alphabetical arrangement of subheads, has certain apparent advantages. As this will almost certainly be the arrangement of the main headings, there is an elegant consistency in carrying the same principle through to the subheads. Also, computers can manage the entire operation this way - in fact, this is the only way in which they can arrange subheads automatically, except for page-number order (by first page number in the group only). How tempting to leave it to the computer, arguing this the preferred method anyway!

While alphabetical arrangement of subheads is highly suitable for many types of index, for narrative indexing it has two major drawbacks. First, it can result in absurd variance from both chronology and logic: Brown, William: death; divorce; education, marriage; youth. Second, though it may appear that alphabetical order makes it easy to find what you seek because you know exactly where to look, this is true only if you know the term under which the sought item will be entered. Under what letter would you seek reference to someone's cheerful disposition? kindness to friends? encouragement of the young? interest in collecting gemstones and porcelain? recurrent attacks of gout? G? Yes, if you knew that gout was the cause of the character's frequent suffering. If not, you might try S for suffering, or I for ill health, or H for health, or give up.


I choose a fourth way to arrange subheads in narrative indexes, requiring more work than alphabetical, chronological, or page-number order, and not possible to delegate to a computer. For long entries, I prefer logical grouping of subheads, perhaps into paragraphs headed 'Family,' 'Character,' 'Career,' 'Relationships,' 'Letters,' 'Works,' as appropriate. Within these sections, entries appear as sub-subheads, suitably arranged: chronologically within 'Career'; alphabetically for relationships, recipients of letters, books written by the character. But each long entry represents a single strand through the book that must be traced, when that entry is edited, back through the whole work, in the search for the overall pattern of the references to the one person.

Reading the book thus repeatedly and selectively in the third group of readings, one is working in the way described by A. S. Byatt as that of A-level literature study: 'One was required to discuss the function of characters in the plot, and ... what extra individuality they had, what intrinsic nature ... The other thing ... was trace recurrent images.(3)

Tracing one major character and theme after another singly through a book to finalize its index entry certainly does remind me vividly of student days, preparing for exams in literature. The same process is described by Thomas Hardy in his diary entry for June 1882: 'As in looking at a carpet, by following one colour a certain pattern is suggested, by following another colour, another; so in life the seer should watch that pattern amid general things which his idiosyncracy moves him to observe, and describe that alone.'

This concentrated collocation of references by the indexer is likely to discover inconsistencies and omissions in the text that escape notice through the production process until this stage. Only the indexer will see that a character is Ann or Miss Phillips on page 26, Anne or Miss Philips on 206; ask whether Paul in chapter 1, Mr Anstey in chapter 10, and Paul Anstey later are the same man; ask for certain missing forenames to complete name entries, drawing attention to the fact that the full names should really have been given to start with. This third group of indexer's readings functions also as a supplementary proofreading.

If a book is good, I think its index should pay it the compliment of reflecting its attitude, its ideas, and their associations. Each long entry should form a coherent whole faithfully conveying the tenor of the text. Two famous examples of narrative index entries that do this come from R.C. Latham's 1983 index to Pepys's diary, and F.A. Pottle's of 1950 to Boswell's London Journal.The terms chosen for the subheads are most felicitous; the arrangement is perfect. Even Levin would hardly call to have them rearranged alphabetically.

All the criteria of good indexes that seem most important to me -- faithfully maintaining the attitudes of the text in the language used; correspondence of importance in the text with space allotted in the index; the fittest arrangement of subheads to form a coherent whole under each long entry -- must derive from a close, sensitive reading of the text. Authors, know that your indexers, usually unacknowledged and disregarded, are often your closest readers.


1 Recommendations for Preparing Indexes to Books, Periodicals, and Other Documents (British Standards Institution BS3700, 1988)
2 Bernard Levin, 'Don't come to me for a reference,' The Times, 10 Nov. 1989
3 A.S. Byatt, The Virgin in theGarden (Chatto and Windus, 1978)
4 R. C. Latham and W. Matthews eds, The Diary of Samuel Pepys, vol. ii, Index (Bell and Hyman, 1983)
5 James Boswell, London Journal, ed. F.A. Pottle (Heinemann; McGraw-Hill, 1950)

-- by Hazel K. Bell in Scholarly Publishing Vol. 23 No.2, January 1992, 115-121