Marginalia
← blog

How to finish the books you start, and when to quit one

Books stall in the middle third, and they stall because you stop mid-chapter with nothing pulling you back. The fix is a stopping rule: end each session a paragraph or two into the next section rather than at the end of the last one. The quitting rule is separate, and it is a real rule, not a failure.

Where exactly do books stall?

Between about 30% and 60%. The opening did its job, the ending is not close enough to pull, and the argument or the plot is in its widest, least resolved state. This is true of both novels and non-fiction, and it is why a shelf of half-read books is almost never a shelf of books abandoned at page 12.

Knowing the location helps, because it means the stall is structural rather than a verdict on the book or on you.

What is the stopping rule?

Stop two paragraphs after a break, not at it.

Finishing a chapter and closing the book feels tidy, and tidy is the problem: you have completed something, so there is nothing unresolved to return to. Read three sentences into the next chapter and stop there instead. Tomorrow you rejoin a sentence in progress rather than facing a fresh chapter heading, and the difference in how hard it is to start is much larger than it has any right to be.

Writers have used this on themselves for a century — Hemingway's version was to stop when he still knew what happened next.

What else moves a stalled book?

Three things, in descending order of usefulness.

Reduce the unit. If twenty pages a night has become nothing a night, make it five. Five is not a compromise, it is what keeps the habit alive through the middle third.

Change the slot, not the book. A book that dies at 11pm is often fine at 8am. Before concluding the book is boring, try it awake.

Read the last chapter's first page. Mildly heretical for fiction, extremely effective for non-fiction: seeing where the argument lands makes the middle legible, and the middle is where you are stuck.

When should you quit a book?

When the honest answer to both of these is no:

  1. Will finishing this change what I think or what I do?
  2. Am I enjoying the sentences?

Non-fiction that fails the first and fiction that fails the second are finished, in the only sense that matters. Put it down.

The test that does not work is "have I read enough of it to be allowed to stop", because that is a sunk-cost question wearing a moral hat. The forty pages you have read are spent whether you continue or not.

Should quitting be recorded?

Yes, and this is the part people skip. An abandoned book that is quietly deleted teaches you nothing; an abandoned book with a page number and one line saying why teaches you what kind of book you do not finish. After a year of that, your shelf tells you something real about your reading — that you never finish 500-page business books, say, or that you abandon novels in translation at exactly the same point.

That is worth more than the tidy 100% completion rate you would have had by lying about it.

Does a shared shelf help or hurt?

It helps, as long as nobody is scoring it. Seeing that two colleagues also stalled at page 140 of the same book converts a private failure into a fact about the book. Seeing that one person finished it and thought it was worth it is often enough to get you through the middle third.

That social pull is the actual mechanism behind book clubs that work, and it is why the meeting should happen whether or not everyone finished.

The short version