What to Keep and What to Shelve: A Four-Pile Sorting Routine for Chapters Over Their Word Limit

There is a particular kind of fatigue that arrives when a chapter is finished in substance but far too long in form. The argument works, the evidence is there, and yet the word count sits several hundred — sometimes several thousand — words above the limit. At that point the standard advice to 'tighten your prose' begins to feel less like guidance and more like a reproach. It presumes you already know which sentences matter.

A more useful starting point is to accept that an overlong chapter is rarely the product of loose writing alone. It is usually the product of accumulation. Material that served an earlier version of the research question, background written before you understood what your reader needed, and paragraphs that report reading rather than argue with it all tend to settle in the draft and stay there. Sorting that material is a study skill in its own right, and it is one that rewards method rather than willpower.

Begin with a sentence the chapter has to defend

Before touching a single paragraph, write one sentence: This chapter argues that… Keep it visible on the desk. Every sorting decision below is made against that sentence. Without it, you are not editing, merely rearranging, and you will find yourself re-litigating the same paragraph three evenings running.

If the sentence will not form, that is information rather than failure. It usually means the chapter is carrying two arguments at once, and the word count is the symptom rather than the cause.

Number first, judge second

Print the chapter and number each paragraph in the margin. Read with a pencil and resist the urge to improve anything on this pass. The task is description, not correction: for each paragraph, note the job it performs. Most will do one of four things — advance the claim, supply evidence for it, anticipate a counter-position, or orient the reader. A number will do none of these, and those are the ones to watch.

The four piles

Retain

These are the paragraphs the chapter cannot function without: the statement of the analytical move, the evidence on which the claim rests, the signposting that keeps a reader oriented across sections. Mark them clearly and treat them as protected. Naming them first matters psychologically — it reframes the exercise as selection rather than loss.

Compress

Compression is rewriting, not deletion, and it is where most of the savings live. Typical candidates include three paragraphs of policy context that could become one paragraph and a footnote; a long block quotation that could become a short phrase with a page reference; a narrative of how data were collected that duplicates the methodology chapter; and the small courtesies at the start of paragraphs — 'It is important to note that', 'As has already been mentioned' — which can go without any loss of meaning.

A worked example: a chapter section of roughly 600 words introducing the policy background is reduced to 350 by keeping the two sentences that frame the analytical problem, replacing a list of statutory instruments with a single citation, and moving the chronological detail into a footnote. The reader loses nothing they need.

Remove

Removal is for material that does no work in this chapter. Paragraphs summarising a theorist's biography, alternative explanations the chapter never returns to, data collected but not analysed, and background the reader can reasonably be expected to hold already are all fair candidates. Keep a separate document for deleted text — an 'offcuts' file — so that nothing feels final and the decision is easier to make.

One caution: removal should never mean stripping citations from claims that remain. Reducing attribution to meet a word count is an integrity problem, not an editing technique, and examiners notice.

Relocate

Some material is valuable but belongs elsewhere: full research instruments, coding frames, extended tables, transcript extracts beyond those actually analysed, or supplementary case detail. Appendices exist for this purpose. Two checks are worth making before you rely on them. First, consult your programme regulations, since some set limits on appendix length and some specify what may not be placed there. Second, apply the dependence test: if the chapter's argument depends on the material, it belongs in the chapter. An appendix should support a claim, not carry it, because examiners are not obliged to assess argument placed outside the main text.

Testing whether the argument survived

When the sorting is done, read only the retained and compressed paragraphs, and read only their topic sentences in sequence. If the sequence still reads as an argument, the chapter has survived. If it reads as a list of observations, something load-bearing has been removed.

There are two honest repairs. The first is to write a bridging sentence that restores the connection the removed material used to supply. The second is to reinstate one paragraph and find the words elsewhere. What is not a repair is leaving a gap and hoping the reader fills it.

Keep a short decision log

A three-column table — paragraph number, pile, words saved — turns an anxious task into an auditable one. It lets you total your savings against the overage, shows at a glance whether you are on course, and gives you something concrete to discuss in supervision. Being able to say 'I removed 400 words of contextual material and moved the instrument to Appendix B' is a far stronger position than reporting that the chapter still feels long.

It also helps to set an explicit target. Calculate the overage, aim to finish a little under the limit rather than exactly on it, and expect to need two passes: one for the structural decisions above, one for sentence-level work.

Knowing when to stop

A chapter is finished when it answers its question as economically as the evidence allows — not when the counter lands precisely on the permitted maximum. Cutting, approached this way, is not a penalty imposed at the end of a project. It is the point at which you find out what your chapter was actually arguing, which is knowledge worth having well before your examiner forms a view of it.