Indexing in the Dark

Microsoft Word uses what’s known among professional indexers as “embedded indexing.” That means the index entries are placed as codes in the text of the document being indexed. Then, later, the codes are used to generate the index automatically. (You can learn more about using Word’s indexing features by searching for “Index” in Word’s Help file.)

Embedded indexing offers one big advantage over traditional indexing: if your pagination changes (for whatever reason), you can easily regenerate the index with fresh, new page numbers for all the entries.

But embedded indexing also has a big *disadvantage* over traditional indexing: there’s no way to see your entries in alphabetical order or even in one place, so it’s like working in the dark. In books with many pages (the kind I tend to get), this is a real problem. For example, I may make an entry for “Gandhi, Mohandas” on page 300, not remembering my earlier entry for “Gandhi, Mahatma” on page 30. That means my index will need lots of editing after it’s been generated.

Until I release my super-duper indexing program (patience, patience), you can alleviate the problem somewhat by opening your document in two windows at once, scrolling to the bottom of the second window, generating your index, and using the index for reference as you create more entries in the first window. Here’s the procedure:

1. Open the document you want to index.

2. Place your cursor in some text where you want to insert an index entry.

3. Click Insert > Index and Tables > Index > Mark Entry. (In Word 2002, click Insert > Reference > Index and Tables > Index > Mark Entry.)

4. Type in your main entry, a subentry, and any other information you want to include.

5. Click the Mark button. If you like, you can enter more index entries for the same text selection, clicking the Mark button for each one. When you’re finished, click the Close button.

6. Repeat steps 2 through 5 to create a few additional entries.

7. Open your document in a new window by clicking Window > New Window.

8. Click Window > Arrange All to put one Window at the top of your screen and the other at the bottom. If you have our Editor’s ToolKit program, click Windows > Arrange Documents to place the windows side by side–or arrange them that way by hand.

9. Place your cursor in the second window and press CTRL + END to go to the end of the document.

10. Click Insert > Index and Tables > Index > OK to generate the (unfinished) index. (In Word 2002, click Insert > Reference > Index and Tables > Index > OK.)

11. Place your cursor in the first window and insert some more index entries.

12. Go back to the second window and update the index (so you can see your new entries in place) by placing your cursor in the index, clicking the right mouse button, and clicking “Update Field.” On a big book with lots of entries, this may take several seconds. (On my not-so-fast computer, a 500-page document with 2,400 entries took 45 seconds to update.)

13. Repeat steps 11 and 12 as needed.

This is far from being the perfect solution to the problems of embedded indexing, but at least it will keep you from having to work completely in the dark.

If you like the idea of automatically arranging windows side by side, you can learn more about Editor’s ToolKit here:



Word guru Steve Hudson sent some useful tips for indexing with a concordance. Thanks, Steve!

Ya know Jack over at The Editorium, right? Well he and I have two completely different approaches to indexing. Yet some of the fringe bits are compatible. However, we both get the job done.

He is making tools for helping hand-build an index. I am making tools for helping clean up a concordance approach. Neither does the job properly without a skilled hand guiding them.

That having been said, naturally I have a heretical stance on the whole thing. This is an abridged and appended version of a longer yack I had with a writer up in the mountains last weekend. She is working on cleaning up my Word Spellbook. I’ve barely started indexing because there is more dump left. This is also the exact same issue I face doing up development documentation: there is always more to add, and that added stuff needs to be indexed like the old stuff.

Quicker, Easier Indexing by Subtraction


How to Use a Concordance File and Still Produce a Decent Index


The Heretic’s Hack ‘n’ Slash Method of Indexing

Note key terms on the way by indexing them. It’s just as easy to mark all as to mark one. Keep on developing away. Time for a minor, internal release. Update yer dynamic index. Copy it to a new doc, flatten it with ctrl+shift+f9 and be clever with find and replace wildcards to blow away numbering, leaving terms ready for use in a concordance file. This then re-performs “mark all” on all your entries.

This works great for getting a good start together. First you review for addition. Get all new terms in there. Either index them all or add them to the concordance. Do this until you are satisfied all key terms have been identified. Search out used synonyms and either kill them or add them to the index. Etc. Hunter-gatherer mode.

Then you review by subtraction, accountancy-management style. If it ain’t important, slash it from your budget. During your passes, you marked separate instances of your word stems:







Time to rationalise, quickly. Use find and replace to do the dirty work for you el pronto! I am planning to help this part with a macro to do stem matching and an interface for hand-matching synonyms and keeping that information in select peer-shared databases. Technically speaking, you can insert HERETIC-NOT-nnnnn bookmarks with the same range as spurious concordance artefacts for future proofing, and auto-expand multiple similar references based on a sliding log scale of the distance of the inference–but that’s a while off yet.

This leaves you with a poor index. Now you do the stuff that good “hand” indexers do as part of their addition process that you’ve missed, which is pretty simple by now. Simply scan through the text looking at your index field placements. Forget the words themselves; we’re beyond words now, we’re being artistic.

Let’s imagine that every major subject in the index is a colour. If it’s a small range, it’s a saturated strong colour; if it’s a large range, it’s pale. Synonyms are varying shades of that colour. This is badly implemented by a simple macro I wrote ages ago to highlight index entries. (Highlight doesn’t have a custom color range.)

If I look at a document from a chapter perspective, I see a rainbow of the base colours with colour boundaries being clearly defined. I zoom in to section level. I see the base colour for that chapter and some interesting hues from cross-over colours where index entries straddle colour boundaries for their multiple relationships. Thickens the spectrum right out for that colour. Some sparkling of other colours is also starting to show.

I zoom in to topic level. Surprisingly at this level, from what we’ve seen above, the base hue is quite pale. A kaleidoscope of colours of all shades is present. Well, at least it SHOULD be, but it probably isn’t if we’ve just finished the sluggo approach I outlined.

What you will have is lots of strong shades and no pale ones. So, we look at the patterns in front of us. Seas of white are either bad and need rectifying or they are long references or graphical content.

Pale shades should feature regularly and will generally be of the hue of the section. However, there should be patches of pale contrasting colour as well, otherwise our index is just a TOC and is useless. A tint of every colour should be represented, somehow, everywhere in a section.

If you see clusters of the strong colours, you need to smudge them and make them paler. Don’t let areas of the same shade sit beside each other; make them paler and covering the whole area.

On a real-world level this means looking behind the words still for meta-concepts that flow from areas as well as ensuring your master : slave pairings are suitable and a good whack of ’em represented richly.

Indeed, it is possible even to try a network theory approach. The words themselves are scale free, but we don’t index them all. Major word hubs are trivials. We try and deal with any minor hubs by clever document structures (TOC) rather than the index, yet still have power terms with many sub-entries. The index picks up the lesser nodes of interest. Log(references) x log(incidences) wouldn’t be a straight line. References x incidences would be closer to a flat line. I’m sure there’s an existing work that’s been done on it somewhere 🙂



After reading today’s article, you may want to know more about where to get help with embedded indexing in Word. If so, check out the WordIndexers discussion group. The group description reads, “Indexers who use Word for embedded indexing will find support, tips, tricks, and a safe place to scream in frustration.”

This entry was posted in Indexing. Bookmark the permalink. Post a comment or leave a trackback: Trackback URL.

Post a Comment

You must be logged in to post a comment.

  • The Fine Print

    Thanks for reading Editorium Update (ISSN 1534-1283), published by:


    Articles © on date of publication by the Editorium. All rights reserved. Editorium Update and Editorium are trademarks of the Editorium.

    You may forward copies of Editorium Update to others (but not charge for it) and print or store it for your personal use. Any other broadcast, publication, retransmission, copying, or storage, without written permission from the Editorium, is strictly prohibited. If you’re interested in reprinting one of our articles, please send an email message to

    Editorium Update is provided for informational purposes only and without a warranty of any kind, either express or implied, including but not limited to implied warranties of merchantability, fitness for a particular purpose, and freedom from infringement. The user (you) assumes the entire risk as to the accuracy and use of this document.

    The Editorium is not affiliated with Microsoft Corporation or any other entity.

    We do not sell, rent, or give our subscriber list to anyone.