Q. Can you tell me the origin of (and reasoning behind) the rule that requires at least two B-level headings beneath every A-level heading in a book? Is this rule still commonly followed?
A. I do not know the origin of such a rule, but I suspect logic has something to do with it (a logic supported by CMOS; see paragraph 1.56). Say you want to divide something—a book, an article, or a chapter—into chapters, parts, or sections. Note that this is already a matter of plurality. One cannot have a book that contains a part 1 but no part 2 or a chapter 1 but no chapter 2. Subheads are often unnumbered, unlike parts or chapters, but why “divide” a chapter into just one subsection? Similarly, why introduce just one second-level subhead? If you do, you might want to think about inventing a second, complementary subhead. But if this principle of division is a rule, there are exceptions to it. For example, a unique subsection like “Notes” or “References” might be necessary for chapter-ending notes or references. And, especially at the B level or lower, a single subhead in a given section might turn out to make sense—especially if it corresponds to subheads of the same level in other sections within the same work. So, in general, follow the rule for the sake of logic; break it only when you can cite a superior logic.
[This answer relies on the 17th edition of CMOS (2017) unless otherwise noted.]