photograph of the indexer

My view of indexing in 2010

I see indexing as an integral part of non-fiction publishing because, without an index, any reference book becomes as ephemeral as a thriller. On the other hand, I sympathise with publishers wanting to cut the delay indexing traditionally introduces. Embedding can offer considerable advantages here, with index completion achievable within a few days of delivery of the final chapter of a book. Here are some more observations about indexing, expressed as answers to nine frequently asked questions.

1 -- Do all books need indexes?

All except those that are either very short (say, less than 30 pages) or intended only to be read once and discarded (like detective fiction). Supplying an index to a reference work (whether it's an IT manual or a cookbook) is an essential courtesy to the customer. But it is much more than that...

2 -- What are the benefits to authors and publishers?

A few studies and much anecdotal evidence suggest that, when comparing different publishers’ offerings on a chosen subject, purchasers look in the index for topics they’d expect to see covered and base their choice on what they find on the indicated pages. So a good index should actually increase sales.

This was always the case in bricks-and-mortar bookstores but’s 'look inside' feature extended the facility to online purchases. Perhaps publishers shouldn’t see an index as a piece of altruism, offering those who have purchased the book a complimentary after-sales service, but rather as a subtle and cost-effective way of giving their product an extra marketing boost?

And, if that's not self-interested enough, reviewers depend heavily on indexes: provide them with an index and they are more likely to give your sales a head start:

'There should be a special corner of the inferno reserved for a publisher who puts out a scientific book minus an index' – Patrick Moore, reviewing Dava Sobel's the Planets in The Guardian, 3rd September 2005

Customers are very unlikely to bother to return a book that turns out to be unusable because of a deficient or missing index, so sales of that title may still be quite good. Indeed, your readers may not even realise why it is unusable, but they may well resolve to find something from a different publisher next time. Routinely providing a serviceable index will ensure your imprint remains associated with the higest standards in the eyes of purchasers, reviewers, compilers of reading lists and authors who trust you to find an indexer to do full justice to their creations.

Finally, though of course your book is going to be the first ever released without a single error or ambiguity, an indexer provides a valuable last minute check, often revealing problems that are hard to detect by conventional proofreading.

3 -- So why are indexes seldom praised or commented on?

The perfect index would be transparent to the reader and, like other efficient tools that we tend to take for granted (latchkeys, telephones), would probably never be noticed. Using it would just reinforce the reader's conviction that they'd bought a good book. It’s only when locating something takes several attempts that, for all the wrong reasons, a poor index gets the reader’s attention.

Indexers can always impose their personalities, just by inserting capricious cross-references like ‘Blair Tony, See under Bush, George W’ (perpetrated, I hasten to add, by an author!) but that’s really just showing off. The readers' needs should be paramount.

That said, they are sometimes praised

4 -- How can I guarantee a good index?

It’s been said of many a professional service that you can have it done cheap, you can have it done fast, or you can have it done well. You can even choose any two of these requirements, but that choice makes the third unachievable. In fact, fast indexes are seldom good. Experience suggests that indexing more than about 80 pages a day leads to a detectable dip in quality.

Choose a qualfied indexer with a recent record of satisfied customers, sufficent subject knowledge and adequate communications skills. Give them as much notice as possible, make your requirements plain, allow enough time for the indexer to do a good job, enough space to accommodate a respectable index and a sufficient budget to pay for it. Easy!

5 -- How long should an index be?

There's a natural length for an adequate index, which allows optimum coverage of the topics dealt with and is dictated by the density of the material presented. Highly technical books will need more index entries: a coffee table book should need fewer (especially where the illustrations are mere eye-candy, in contrast to some textbooks where illustrations, graphs and tables may actually deserve deeper indexing than the text they displace). Overall, about 3.5-5% of the number of indexable pages is a good starting point. Indexers are skilled at compression but too tight a size limit will eventually lessen the index's value to your readers.

6 -- How much time will it take?

Most indexers have stories of working flat out for four days to finish a 400 page book for a favourite customer who'd been let down by someone else. It can be done, but if you make a habit of expecting it, you're not likely to become that favourite customer. After all, finished books don't arrive on your desk without warning. Also, indexing at 100 pages a day seldom results in top quality work. In any case, indexers do have lives, families, holidays and probably other commissions as well. After the first year or so, we're not likely to be sitting about waiting for your call. Generally, I'd expect customers to allow three weeks for a short book (under 400 pages).

If waiting a matter of weeks is likely to be a serious problem, you might like to think (in advance, please!) about using embedded indexing, in which I have considerable experience. With this technique, index terms are added as hidden text prior to pagination so that the index can usually be generated within a couple of days of the release of the final section of text. It's even possible to index drafts, provided subsequent amendements aren't significant enough to require a complete re-index.

7 -- Wouldn’t the author produce a better index anyway?

Almost never. First, a few authors do have the skills needed to index well, but no more than are adept at illustrating their own works, or designing a dust jacket, say. Most don’t. Crucially, authors tend to see subjects in terms of their own personal vocabularies, so their indexes risk being little better than computer-generated concordances. An indexer will approach the work from a reader’s viewpoint and tend to provide those - sometimes quite different - entry terms that the reader will in fact expect to see, not just those the author used. Using a professional indexer avoids forcing your readers to think exactly like the author.

Finally, even the most skilful writer with a talent for indexing will lack the specialist training, depth of experience, professional backup and dedicated software that together keep professional indexers productive enough to earn a living. They'll therefore take far longer to do a decent job.

For multi-authored works, any policy other than using a professional indexer is a recipe for chaos and a near guarantee of reader dissatisfaction

It’s really no more sensible to have authors index their own work than to let them proofread it. In both cases, a fresh insight will always pay dividends. It's just a huge tragedy for the reading public that UK publishing conventions often make authors responsible for the cost of index provision, encouraging them to undertake for themselves a specialised professional task in which few of them are really skilled.

Incidentally, lists of keywords provided by the author are not always helpful and certainly don't make the indexer's task any easier or faster. An index gives access to concepts, not keywords; uses the likely reader's vocabulary as well as the author's and provides analytical entries indentifying significant mentions, not a list of occurrences. Even with the list, we're still going to have to read the book! That said, advice from an author on specific points will usually be greatly appreciated if any interpretative problems arise in the course of indexing.

Finally, I can't recall when I was last given a book to index in which I didn't find errors, undetected by the author, editor and sometimes even the proofreader. This is partly because the indexer is often the first truly independent reader, and so more alert to inconsistencies and simple slips, but also because the act of alphabetisation juxtaposes terms widely-separated in the main text and highlights any mismatches.

8 -- Can’t we just use a computer?

Certainly not! Computers are great at finding and sorting words but, (as I've just said!) twenty-first century indexes aren't lists of keywords but present alternative ways to access key concepts. Computers can't recognise concepts, and can't deal with allusion, class relationships, synonym, backward and forward references or negative mentions or resolve homographs, nor can they distinguish significant material about a topic from mere repetition or summary. An index isn't something pre-existing, that just needs extracting from the text: it's a separate and complementary creative work, reflecting the insight, subject knowledge and professional skills its compiler has brought to bear on your text.

Of course, all the indexers I know use computers heavily: they're ideal for rapid transfer of documents and indexes, copying and pasting, sorting and merging, searching and formatting and they make home working a doddle... but they can't index.

9 -- What about fully-searchable text?

Sorry, no. This has the same shortcomings as computer-produced concordances. It's also totally intolerant of any spelling errors, doesn't collocate alternative terminology that refers to the same thing and gives no helpful links to related material. It's fine on Google, where you just want some information; quite inadequate for a reference book, where the reader will want to find everything included about a given subject.

In practice, fully-searchable text is even more user-unfriendly than it first seems. Because it won't break down large numbers of occurrences, it will especially frustrate the user seeking the main information content. This is likely to be discussed frequently, so each search term will simply return huge lists of undifferentiated locators, driving the hapless reader back to guessing and scanning (though probably not before cursing the publisher).

To take an example from a short book, the 123 pages of Chemistry in the Garden contain the word 'soil' 102 times; the index has ten references under 'soils', each modified by an informative sub-entry so the reader should never need to look in more than two places (the maximum number under any subheading). No reader would ever be prepared to look in 102 places but, even if they did, they'd miss any alternative terminology ('earth', 'humus', 'surrounding medium'...)!

Full-text searching is quite simply the most overrated phenomenon since the Millennium Bug!

- - - - - - - - - - - - - - -