How long should a book be?

What word count is normal in this genre?

Named categories have published boundaries, and the markets that buy them draw their own lines inside those boundaries. The Nebula Award rules put a novella at 17,500 words or more and fewer than 40,000.¹ The novella program at Tordotcom asks for submissions between 20,000 and 40,000 words, which starts 2,500 words above the award definition.² Clarkesworld, published by Wyrm Publishing, takes fiction of 1,000 to 22,000 words and states that there are no exceptions.³

Three sources, three different sets of numbers, and all three are correct for their own purpose. The award rules describe what the field calls a novella. The imprint and the magazine describe what they will buy.

An author submitting to a particular market is bound by that market's number and not by the general one. The definition is useful for naming the thing in a query letter, and useless as a defense of a manuscript that misses the limit.

Is my book too short to sell or too long to be read?

Those are two failures, and only one of them is measurable. A manuscript outside a market's stated range is refused on the count alone, before anyone reads a line, and Clarkesworld's phrase for that is no exceptions.³ Nothing about the writing enters into it.

Whether a book is too long to be read has no published threshold. No association, publisher, or style manual we could find states a length past which readers stop, and any figure offered for it is somebody's impression rather than a measurement.

What an editor can report is more local and more useful. A structural pass says which chapters restate work already done, where the same information arrives twice, and where a reader is likely to set the book down. That is a reading judgment, offered as a reading judgment, and it does not come with a number attached.

Is the length a problem, or is the pacing the problem?

Cutting and reordering are different operations, and a manuscript can need one, the other, or both. Editors Canada defines structural editing as assessing and shaping draft material to improve its organization and content, and lists revising, reordering, cutting, and expanding among the tasks it covers.⁴ A book 30,000 words over a stated limit has an arithmetic problem. A book of the right length that feels slow has an organization problem, and cutting it will not fix the order.

The two get confused because both are reported as length. An author told the middle drags will often cut the middle, when the finding was that two chapters do the same work and the second arrives late.

We report which passages repeat, where the sequence doubles back, and what a chapter is contributing. Which words leave the book is the author's decision, and it is not one an editor should be making alone.

Does genre word count still constrain what can be published?

Evidence for the constraint is easy to find and easy to check. It sits in submission guidelines that state hard limits and enforce them before reading, as Clarkesworld and Tordotcom both do.³² Those are constraints in the plainest sense: a manuscript outside the range is not considered.

The broader claim, that a debut over some particular length will not sell, is repeated constantly and we could not trace it to a statement by any publisher. It is left out of this answer for that reason rather than because it is false.

Publishing independently removes the submission gate and leaves everything else in place. There is no limit to break, no form to fail, and the same reader deciding whether to finish. An author in that position is choosing a length rather than meeting one.

Is the chapter the right unit for this material?

Chapters are a convention with house rules attached rather than a natural division. The Chicago Manual of Style devotes part of its chapter on manuscript preparation to the format of chapter titles and titles of other parts, and to the format of subheads.⁵ Books that use parts above chapters, numbered sections instead of chapters, or unnumbered breaks are all provided for.

Princeton University Press asks for a list of planned chapters with a paragraph describing the contents of each.⁶ That request works as a test outside a proposal. A chapter whose paragraph keeps turning into two paragraphs is usually doing two jobs.

Where the material is not naturally sequential, a chapter can be the wrong container, and the alternative is normally shorter numbered sections. That is a design decision, and an editor can describe the consequences of each without making it.

Are the chapters roughly even, and do they need to be?

Evenness is not a requirement anywhere we looked. We found no style manual or publisher guideline that sets a chapter length or a permitted variation between chapters. What the proposal guidelines ask for is a description of what each chapter contributes to the argument, which is a statement about function and not about size.⁷

Evenness matters for a different reason, which is that readers use chapter breaks to decide when to stop. A book of twelve-page chapters with one chapter of sixty pages has changed its own promise partway through.

Where the unevenness is deliberate, it usually needs a signal. A part division, a differently styled heading, or a section break tells a reader that the shape has changed on purpose. Without one, a long chapter reads as a chapter that got away from its author.

An editor checks the pattern rather than the individual number. What gets reported is the shape of the whole list of chapters, and which entries in it sit outside the pattern the book has set for itself.

What happens when one chapter is four times the length of the others?

It gets queried, and the query is not automatically a request to cut. Structural editing covers cutting and expanding alike, and both are on the table.⁴ Three findings come up often enough to name.

The first is that the long chapter is two chapters that were never separated. The second is that the long chapter is the center of the book and the short ones around it are underwritten, in which case the answer is to expand rather than to cut. The third is that it holds material a reader does not need in sequence, which belongs in an appendix or in notes.

An editor marks the imbalance and sets out which of those the manuscript looks like. Deciding among them requires knowing what the book is for, and the author knows that better than we do.

What is the current word count, and why does an editor ask first?

Every quote begins with this number. Princeton asks for the expected length of the manuscript in total words including notes, or in double-spaced pages.⁶ The University of California Press asks for an estimate of the whole book including main text, footnotes, bibliography, and any appendices.⁷ Editing is priced by the word or by the hour, so the count sets both the fee and the time the work will take.

The number an editor needs is the whole file rather than the prose alone. Notes, captions, tables, and appendices are read, checked, and made consistent along with everything else, and a count that omits them produces a quote that will move later.

Word processor counts differ slightly between programs, which rarely matters at this scale. Sending the file settles it. Every document a client sends us is treated in confidence.