A document revised four times has been proofread none

How revision makes errors

A first draft has the errors of writing: typos, slips, sentences that got away. Those are visible and they get caught. What revision produces is a different class of error, and it is invisible because each individual change was correct.

• Half-finished sentences. A sentence is rewritten from the middle and the old ending is left attached. It scans, because the author reads what they meant.
• Orphaned cross-references. A section is moved and “as described above” now points forward. A figure is deleted and the numbering after it is wrong everywhere.
• Two versions of a number. A value is updated in the results and not in the abstract. This is the one reviewers notice, and it is the one that costs most.
• Terms that drifted. A variable renamed in one section and not in another, so the paper describes two things where it has one.
• Tense and voice shifts at the joins between old and new text.

Multiple authors make it worse

A manuscript circulated to five co-authors comes back with five sets of changes, made without knowledge of each other. Merging them produces contradictions that nobody wrote: two authors soften the same claim, so it is now hedged twice; one adds a limitation another has removed; the methods and the discussion describe the analysis differently because two people fixed the same sentence in two places.

The fix is procedural rather than editorial. One person holds the master file. Changes are merged in one direction. And the merged version is read end to end by someone who did not make any of the changes, before it goes anywhere.

The most expensive moment

Responding to reviewers is when this does the most damage, because the changes are surgical and scattered. An analysis is rerun and three numbers change — but the same numbers appear in the abstract, a table, a figure legend and the discussion, and the author updates the ones they remember.

An editor who finds a number in the abstract that contradicts the revised results does not conclude that you forgot. They conclude that the revision was not done carefully, which affects how the rest of the response is read.

The last pass after the last change

The rule is simple and almost universally broken: proofread after the final change, not before 1. A document proofread and then edited has not been proofread. This is why a proofread booked in advance for a fixed date so often lands in the wrong place — the deadline moves, the changes keep coming, and the pass happens in the middle.

Practically: freeze the content, then proof. Where a late change is unavoidable, the passage that changed and everything that refers to it are reread, not just the sentence that was edited.

Five checks that catch most of it

• Read every number in the abstract against the results. Every one, against its source. This is fifteen minutes and it catches the most damaging error.
• Search for every cross-reference — “above”, “below”, “Figure”, “Table”, “Section” — and confirm each still points where it should.
• Check that every citation in the text is in the list and every entry in the list is cited, after the last reference was added or removed.
• Read the methods and the results consecutively and confirm they describe the same analysis in the same words.
• Read the whole document aloud, or have it read to you. Half-finished sentences are almost impossible to see and very hard to miss when heard.

Common questions

@misc{revision_damage_2026,
  author       = {{Pubrica Academy}},
  title        = {A document revised four times has been proofread none},
  howpublished = {Pubrica},
  year         = {2026},
  month        = {apr},
  url          = {https://pubrica.com/insights/revision-damage/},
  note         = {Accessed: <span data-cite-acc="iso">2026-10-15</span>}
}