Which Style Manual to Use, and What Changes If You Pick the Other One

The destination names the manual

Journals say so outright. The International Committee of Medical Journal Editors says in section I.B of its recommendations that authors should use them along with individual journals' instructions to authors.³ PLOS ONE names its own choice, requiring references in Vancouver style, numbered in the order they appear in the text and cited in square brackets.⁴

Presses do it by convention, and most American trade and scholarly publishing runs on Chicago, whose eighteenth edition appeared in 2024 and treated fiction for the first time.⁵ Scientific publishing has the CSE Manual, whose ninth edition arrived the same year with overhauled reference recommendations and advice on American rather than British conventions.⁶ Federal agencies run on the GPO Style Manual.

An author choosing a manual before knowing the destination is answering a question nobody has asked. An author who knows the destination usually finds the answer already published. In the federal cases the manual is also free to read.

Some destinations publish rules no manual would predict

The Office of the Federal Register issues its own Document Drafting Handbook, whose August 2018 edition carries revision 2.2, dated June 2025.⁷ Two of its instructions show what a house handbook does that a style manual does not. A part heading is usually a phrase rather than a statement or a question, and each section heading ends with a period or a question mark.⁷

No general manual produces that second rule, and care with Chicago would not satisfy it. Where a recipient publishes its own handbook, the handbook governs, and the manual settles only what the handbook leaves open. An author who reads the manual and not the handbook has read the less authoritative document.

What actually changes when the manual changes

Six things change, and the rest of the language stays where it was.

Numbers are the first, and the disagreement described above is the whole of that category. Punctuation in a series is the second. GPO rule 8.46 puts a comma after each member within a series of three or more words, phrases, letters, or figures joined by and, or, or nor, and its example is red, white, and blue.⁸ Reference format is the third, and a numbered Vancouver list and an author-date list look nothing alike on the page.⁴

Capitalization of headings is the fourth. GPO rule 3.50 sets centerheads in capitals and sideheads in sentence case, and rule 3.53 capitalizes all principal words in matter set in caps and small caps.⁹ Treatment of abbreviations is the fifth, and ICMJE asks for standard abbreviations only and advises against using them in a title.¹⁰ Units are the sixth, and ICMJE asks for metric units of length, height, weight, and volume, with temperatures in degrees Celsius.¹⁰ Almost nothing else moves, because the manuals do not disagree about grammar.

Where the record of the decision lives

A manuscript that has named its manual has converted six open questions into six answered ones, each answer checkable against a published source. A manuscript that has not named one answers them differently in chapter two and chapter nineteen, and no later reader can tell which answer was intended. The record of those answers has a name and a place in the profession. Editors Canada puts developing a style sheet, or following one that is provided, inside copy editing, alongside checking the consistency and continuity of mechanics and facts.¹¹ The Editorial Freelancers Association describes copyediting in the same terms, down to preparing the style sheets that guide consistency across a manuscript.¹² Asking for a style sheet at the end of any copyediting or proofreading job is worth doing, because the sheet records the manual chosen and every decision the manual did not cover.

What no manual settles

No manual rules on whether a sentence is clear, whether a paragraph belongs where it sits, or whether a figure is the right figure. Those questions belong to an author and their reviewers, and a manuscript can be perfectly conformant and still hard to read. Where a client has chosen an authority, EditFast works to it and returns the style sheet with the manuscript. Where a client has not chosen one, we ask where the document is going, because that usually settles it. Everything a client sends is treated in confidence, including unpublished manuscripts, submissions, and correspondence with a publisher.

Three manuals are named above, and the two federal documents cited here cost nothing to read. The choice that cannot be repaired afterward is the one nobody made. A document holding two answers to the same question carries no record of which one was meant.