An American Editor

December 24, 2018

Indexes: Part 7 — Lessons Learned in Using DEXembed for the First Time

Editor’s note: This version of the post incorporates corrections made by the author to the Options and Advice sections.

Ælfwine Mischler

I recently created an embedded index in Word for a book that will be published as an ebook and in print. I chose to use DEXembed because colleagues advised that its syntax — a space between the curly brackets and the enclosed text — will work better when the text is converted to an ebook.

A quick explanation of an embedded index: For a print book, the index is written after the book has been designed, using a PDF file of the final pages and page numbers as locators. This is changing, and many publishers are now asking for embedded indexes. For an embedded index, the indexer uses something else as locators. Depending on the program used, this could be paragraph numbers, word numbers, or temporary bookmarks. After indexing, the program embeds the entries by inserting field codes that look like this: { XE “main entry:subentry” }. The index is then generated from the field codes so the pages numbers are displayed. In an ebook, they may also be linked to the location in the text. If the book is designed as hardcover and paperback with different pagination, the embedded index entries will give the correct page numbers for each edition.

Embedded indexes are more work for the indexers, so most of us will charge more for an embedded index.

Options in DEXembed

DEXembed (available from the Editorium) is a Word add-on that allows the indexer to use dedicated indexing software rather than Word’s clunky built-in indexing function. DEXembed can use paragraphs, words, or numbers as locators — but only one type in a given document. Paragraph number was the best choice for this project, but the author had sometimes used auto-spacing and other times had used Enter twice between paragraphs. I told him repeatedly that he had to remove the extra Enters and make the spacing between paragraphs consistent (which he did) and that he could not change the paragraphing after I had started indexing. (More on that in the second part of this article in February 2019.)

Experienced colleagues in the Digital Publications Indexing Special Interest Group (DPI SIG) say that Word does not handle ranges of locators well. It is therefore better to mark only the beginnings of entries that are less than two pages long. DEXembed offers three options for ranges: Mark them with bookmarks, mark them with beginning and end codes, or do not mark them. The documentation for DEXembed says that publishers usually prefer begin and end codes.

Before starting my index, I sent two small sample indexes to my author’s publisher — one using bookmarks and one using begin and end codes — and asked which worked better for them. They got better results with the bookmarks, which also meant one less step for me in the end. Hurray!

I Won’t Talk to You

DPI SIG members also advised me that Word and InDesign use different syntax for some things, and I had to take this into consideration while indexing. I also found that my Sky indexing software and Word do not always communicate well.

This index required a separate scripture index of Qur’an verses. In Word, you can use an f-switch that is coded with \f followed by a name to make two indexes at once { “heading1” \f “subject” } and { “heading1” \f “quran” } (See Seth A. Maislin’s blog for more.) However, my colleagues advised that InDesign will reject XE fields with a backslash.

A suggested solution that I followed was to use two levels of subentries, with the main entries for the two indexes. That is, I had only two main entries, for which I used bold text, and my first level of subentry was the real main entry I wanted. The sub-subentry was the real subentry I wanted. The designers can adjust the indentation and spacing to make these appear as two separate indexes:

The chapter and verse numbers presented two other problems of their own. How to write something like 2:10? First, Word signals heading levels with a colon, so I had to use a backslash before the colon to tell Word that this was a literal colon, not a subheading signal. I admit that at that point, I had forgotten the warnings of my colleagues that InDesign would reject these entries.

As of this writing, I am waiting for the author’s comments and corrections, and the results of a small test index for the publisher: three entries using a backslash and colon, and three using a plus sign to be replaced by a colon in the generated index. If I do indeed have to remove \: from the index, I want to be sure that + is not a signal for something else in InDesign.

A second problem in writing chapter and verse numbers was the sorting. I knew that in Sky, I had to enter one- and two-digit chapter numbers with preceding zeros so they would sort properly. Thus, Chapter 2 was entered as 002 and Chapter 16 as 016. The verse numbers following the colons, however, sorted properly in Sky without additional zeros.

Word was not happy with that, but I could only learn that at the end. I finished my index, embedded the entries, generated the index, and then found that Word had mis-sorted the verses so that, for example, 18:70 came before 18:7. I had to open Sky, add the zeros to the verses, re-embed the entries, generate the new index, and remove the extra zeros from the generated index.

Maybe I’ll Talk a Little Bit

Another difference between Sky and Word is how they handle text to be ignored in sorting. Sky’s sorting automatically ignores prepositions at the beginning of subentries, but  Word’s does not. Sky also allows the indexer to code other things to be ignored in sorting. I commonly do this with the al- that begins many Arabic names.

For the embedded index, I had to enclose items to be ignored in angle brackets, but then in Sky, they all sorted to the top because they started with symbols. I was not sure that Word would put <al->Bukhari, <al->Ghazali, <al->Tabari, etc., in the proper places in the generated index. On this, I did have success, but I had to go back to the few subentries that begin with prepositions and enclose the prepositions in angle brackets.

DEXembed uses a text file to embed the entries, and all the bold and italics are lost in the process, although their coding remains. Once the entries were embedded, I had to edit the XE fields to get the bold and italic formatting back. (See Sue Klefstad’s blog post for details.) This was not difficult with a Find and Replace using wildcards (but be sure to turn off Tracked Changes!), but it was an extra step to perform.

Advice for Embedded Indexing

It is important to communicate with the author and publisher before beginning an embedded index. Learn how the Word manuscript will be handled after indexing and how it will be published. (There is more information on the resources page of the DPI SIG website.)

Once you have written your index in your dedicated indexing software, always embed in a copy of the document. Always keep the original “clean” and do not embed in it. Sometimes Word does not embed the entries properly and you might have to try again. DEXembed does have a function to remove embedded entries, but if Word gives you run-time errors as it did to me (see the second part of my February 2019 column), you will want to try again in a clean copy so there is no chance of stray coding in the file.

My thanks to colleagues Sue Klefstad and Seth A. Maislin for their invaluable blog posts, and to other colleagues in the DPI SIG for their advice in e-mail messages.

Ælfwine Mischler is an American copyeditor and indexer in Cairo, Egypt, who has been the head copyeditor at a large Islamic website and a senior editor for an EFL textbook publisher. She often edits and indexes books on Islamic studies, Middle East studies, and Egyptology.

July 16, 2018

Book Indexes — Part 3: The ABCs of Alphabetizing

Ælfwine Mischler

The alphabetizing I learned in school so many years ago — all before PCs and the Internet, of course — was easy. Go by the first letters — Bincoln, Fincoln, Lincoln, Mincoln — and if they’re all the same, look at the second, then the third, etc. — Lankin, Lanky, Lenkin, Lincoln, Linkin. I rarely had to alphabetize anything outside of school assignments (I did not organize my spices alphabetically), but I had to understand alphabetization to find a word in a dictionary, a name in a phone book, a card in a library catalog, or a folder in a file cabinet. Hunting for an organization or business whose name was just initials or began with initials was sometimes tricky, but I soon learned that if I did not find something interspersed with other entries, I could look at the beginning of that letter.

As an indexer, I have to know the conventions of alphabetizing so I can enter terms in the software program, and like so many other things in editorial work, there are different standards to follow. There are two main systems of alphabetizing — word-by-word and letter-by-letter — with some variations within each system. If you are writing an index or hiring an indexer, you have to know which system the publisher uses. Occasionally an indexer might find, in the midst of a project, that switching to the other system would be better, but this must be cleared with the publisher.

Word by Word

In the word-by-word system, generally used in indexes in Great Britain, alphabetizing proceeds up to the first space and then starts over. According to New Hart’s Rules, 2nd ed., hyphens are treated as spaces except where the first element is a prefix, not a word on its own (p. 384). However, the Chicago Manual of Style, 17th ed., treats hyphenated compounds as one word (sec. 16.60).

Letter by Letter

Most US publishers prefer the letter-by-letter system, in which alphabetizing continues up to the first parenthesis or comma, ignoring spaces, hyphens, and other punctuation.

If you are writing your own index in a word processing program, it will use word-by-word sorting. Dedicated indexing software can use either system along with variations. The following table comparing these systems uses Microsoft Word and SKY Indexing Software with various settings. (The items in the table were chosen to demonstrate how the different systems handle spaces, hyphens, commas, and ampersands. Not all of them would appear in an index. The variations on Erie-Lackawanna, for example, would normally have another word, such as “Rail Road,” following them.)


Entries with Same First Word

In the first edition of New Hart’s Rules, names and terms beginning with the same word were ordered according to a hierarchy: people; places; subjects, concepts, and objects; titles of works. You may see this in older books, and it occasionally comes up in indexers’ discussions. However, the second edition of New Hart’s Rules recognizes that most people do not understand this hierarchy and that alphabetizing this way is more work for the indexer. The second edition (p. 385) recommends retaining the strict alphabetical order created by indexing software.

Numbers Following Names

Names and terms followed by numbers are not ordered strictly alphabetically. These could be rulers or popes, or numbered articles or laws, etc. An indexer with dedicated software can insert coding to force these to sort correctly. If you are writing your own index in a word processor, you will have to sort these manually.

When people of different statuses — saints, popes, rulers (perhaps of more than one country), nobles, commoners — share a name, these have to be sorted hierarchically. See New Hart’s Rules, 2nd ed., section 19.3.2, and Kate Mertes, “Classical and Medieval Names” in Indexing Names, edited by Noeline Bridge.

Numerals and Symbols at the Beginning of Entries

Entries that begin with numerals or symbols may be sorted at the top of the index, before the alphabetical sequence. This is preferred by the International and British Standard, and when there are many such entries in a work. Alternatively, they may be interspersed in alphabetical order as if the numeral or symbol were spelled out, and they may be also be double-posted if they appear at the top of the index.

However, in chemical compounds beginning with a prefix, Greek letter, or numeral, the prefix, Greek letter, or numeral is ignored in the sorting.

Greek letters prefixing chemical terms, star names, etc., are customarily spelled out, without a hyphen (New Hart’s Rules, 2nd ed., p. 389).

If you are writing your own index in a word processing program, you will have to manually sort entries with Greek letters or prefixes to be ignored, and entries beginning with numerals if you do not want them sorted at the top. Dedicated indexing programs can be coded to print but ignore items in sorting, or to sort numerals as if they were spelled out.

That’s Not All, Folks

This is just the beginning of alphabetizing issues that indexers face. While most of the actual alphabetizing is done by the software, indexers have to know many conventions regarding whether names are inverted; how particles in names are handled; how Saint, St., Ste. and Mc, Mac, Mc in surnames are alphabetized (styles vary on those); how to enter names of organizations, places, and geographical features. In addition to checking the books mentioned above, you can learn more about indexing best practices and indexing standards on the American Society for Indexing website and from the National Information Standards Organization.

Ælfwine Mischler is an American copyeditor and indexer in Cairo, Egypt, who has been the head copyeditor at a large Islamic website and a senior editor for an EFL textbook publisher. She often edits and indexes books on Islamic studies, Middle East studies, and Egyptology.

Blog at

%d bloggers like this: