Tracked Changes or Clean Copy: Which Files Journals Want Back.

For most resubmissions the answer is both. Journals that ask for tracked changes still want a clean copy, and the two files do different jobs: the clean copy becomes the typeset manuscript, while the marked version lets an editor confirm each edit without rereading the paper. Some journals want only one file, so the decision letter or the Guide for Authors, not habit, decides which you upload.

What do journals mean by marked copy, tracked changes, and clean copy?

Three names, two documents. "Tracked changes" and "marked copy" both mean the manuscript with your edits visibly flagged; "clean copy" means the same manuscript with those edits accepted and hidden.

Track Changes is the feature in Word and similar editors that records every insertion and deletion so a reader can see what moved. Publishers label the flagged file differently, which is where the confusion starts. PLOS ONE calls it a "Revised Article with Changes Highlighted." Nature portfolio journals ask for a "marked-up version." Elsevier journals often just say the "tracked changes" version. All three describe one thing: the paper with its revisions on display. The clean copy is that same paper with Track Changes turned off and every change accepted, so it reads as a finished manuscript with no markup left.

Which version does your journal actually want, and where do you check?

Check the decision letter first and the Guide for Authors second. Requirements differ by publisher, and sometimes between journals under the same publisher, so there is no rule you can carry from one submission to the next.

The major publishers show how far apart these requirements sit. PLOS ONE asks for three files on a revision: a clean manuscript uploaded as the main file, the marked-up "Revised Article with Changes Highlighted," and a response to reviewers. Nature portfolio journals ask for a marked-up version that highlights the changes plus a clean, non-highlighted version, with the point-by-point response kept as a separate document. Elsevier states that some publications require a second copy with tracked changes and some do not; where both are required, the clean copy is uploaded as the primary manuscript file and the tracked version as a separate file type, and usually only the clean version is built into the review PDF while the tracked file is linked for the editor to download. Reading the decision letter closely is faster than guessing and re-uploading.

How do you produce both versions without version chaos?

Keep one working file with Track Changes on, and generate the clean copy from it by accepting all changes on a duplicate. Do not maintain two manuscripts by hand.

The reliable way to lose a revision is to edit a clean draft and a marked draft as separate files. They drift within a day: a sentence fixed in one but not the other, a figure caption that exists in only one copy, line numbers in your response that no longer match either file. One source file avoids all of it. Turn Track Changes on before you touch the manuscript, and use comments to leave notes for yourself where a reviewer asked for something you are still weighing. When the revision is done, duplicate the file, run Accept All Changes and stop tracking on the duplicate, and that duplicate is your clean copy. Name the two files so you cannot mix them up, something like paper_R1_marked and paper_R1_clean, and switch on continuous line numbers so both carry the same coordinates.

How should tracked changes map to your point-by-point response?

Every place your response says "we changed this" should point to a specific, findable edit in the marked copy. The response document and the marked manuscript are read together, not in isolation.

An editor rarely rereads the whole paper. They scan your response, and for the claims that matter they open the marked copy and look for the edit you promised. When your reply says a paragraph was rewritten on page 6 and there is a visible tracked change on page 6, the claim verifies itself in seconds. When the reply points somewhere with no markup, the whole response starts to read as unverifiable, and that suspicion spreads to the replies the editor did not check. So give a page and a line for each change, and keep the wording of the reply close to the wording of the edit. The point-by-point response format exists to make this mapping easy for the editor, and the structure of the response letter sets the tone that carries it. This coupling is also why a marked copy that annotates each edit with the journal rule it satisfies saves the editor a step: Waraq produces tracked changes cited to the specific style or submission rule behind each one, which is the same link an editor would otherwise reconstruct by hand.

Should line numbers point to the clean copy or the marked copy?

Number the clean copy and cite that. Line numbers in a tracked-changes file are not stable across readers, because the number a line lands on depends on whether the reader views the markup inline, in balloons, or with changes hidden.

If you cite "line 240" from your own tracked view, the editor who reads with markup collapsed may see line 240 fall on a different sentence, and every location in your response drifts by the same amount. The clean copy has no such ambiguity: accepted text sits where it sits for everyone. Add continuous line numbers to the clean file, cite those, and the marked copy still does its separate job of showing what moved. Whether the revision counts as a major or minor revision does not change this part of the mechanics, only the volume of edits you are pointing to.

One workflow covers all of it: build one marked file, export a clean one from it, number the clean copy, and make every line you cite in the response land on the same text for whoever opens it.