| name | manuscript-writing-style |
| description | Style and structure for research manuscript text: IMRaD, hourglass narrative, formal tone, topic sentences, reproducible Methods, captions that tell the story. Use when drafting or editing titles, abstracts, sections, captions, or reports. |
Manuscript writing style
Guidance from Chamba, Knapen & Black (2022), Papers I–II
(arXiv:2207.12959 — optional; skills stand alone).
Applies to .tex body text, abstracts, and captions.
Narrative structure
- One message — every section advances the story from planning (Paper I).
Cut prose that does not serve it.
- IMRaD: Introduction → Methods/Data → Results → Discussion → Conclusion.
Do not mix roles (no new background in Discussion; no results in Methods).
- Hourglass (Paper II, Sect. 2.4): broad intro → narrow methods/results →
broad discussion/conclusion.
Section checklists (Paper II)
| Section | Job |
|---|
| Title | ~15 words; accurate; keywords; main result if possible; no vague "On…"/"Towards…"; no jokes |
| Abstract | Scene → problem → method → results → conclusion → optional outlook (~150 words) |
| Introduction | Big picture → sub-topic → literature gap → your aim → what you did → optional section map |
| Methods | What data/tools and why they meet your goals; enough to judge and reproduce; state caveats honestly; past active ("we used…"); no results here |
| Results | What you found; figures drive structure; foreshadow main points; compare (i) "Figure 3 shows…" vs (ii) "To show how x relates to y, Figure 3 shows…" — prefer (ii) |
| Discussion | See seven-point list below; limitations explicit |
| Conclusions | Separate section (or "Discussion and Conclusions"); many readers skip here — summarise why/how/what; refer to key figures; strong final sentence with implication |
Discussion — seven aspects (Paper II, Sect. 2.8)
Work through in order, widening the hourglass:
- Repeat why you did the study.
- Return to the main question from the abstract; answer it.
- Show how the answer is supported by data, models, figures.
- Compare to literature — agreement and disagreement.
- Explain disagreement or alternatives without losing the main message.
- Wider implications, applications, recommendations.
- What future work would clarify or advance the field.
Formal tone (Paper I, Sect. 2.5)
- Inform, do not entertain. Objective, neutral, professional.
- No contractions (write "do not", not "don't").
- No imperative to the reader ("note that…" → "the data are…").
- No colloquialisms, slang, or pop-culture references.
- Criticise prior work professionally, not personally.
- Inclusive language: avoid violent metaphors common in astrophysics when
alternatives exist (Paper I, Sect. 2.5.1); define non-standard terms on first use.
Paragraphs and sentences (Paper II, Sect. 3.1–3.3)
- Topic sentence first — main idea in sentence 1, then evidence (~3–7 sentences).
- One idea per paragraph; split half-page blocks.
- Link sentences — repeat a key noun at the start of the next sentence; new
information at the end of the sentence (English convention).
- Sign-post words: "for example", "first/next/finally", "however", "in contrast",
"consequently", "similarly".
- Length: average 18–32 words; limit long "which/that" chains (~1 in 3–4).
- Subject–verb–object; keep subject and verb close.
- Active voice when possible ("we measured…"; also "Figure 1 shows…",
"this paper presents…").
- Tenses: present for facts and for referring to figures in this paper; past
for what you or others did in the study; present perfect for recent work with
current relevance ("algorithms have been proposed…").
- Dangling modifiers — place phrases next to what they modify (Paper II, Sect. 3.4).
- Ambiguous "this" — follow with a noun ("this evolution", not bare "this").
- Phrasal verbs — avoid in formal prose ("investigate" not "look into";
"discuss" not "talk about"; Paper II, Sect. 3.10).
- Data is usually plural; amount for uncountable quantities, number for
countable objects (Paper II, Sect. 3.6, 3.9).
Before heavy notation: one plain sentence on what the quantity is and why it matters.
Captions and figures (text side, Paper II Sect. 2.3)
- Captions: full sentences; explain every panel element; units, sample size, method id.
- Figures should tell the story — many readers see only title, abstract, figures.
- End captions with a one-sentence takeaway when helpful.
- Consistent typeface, text size, and colour coding across figures in one paper.
- Text and captions must match current figure files and run ids.
- Abstract/conclusion numbers must trace to the same artifacts as figures.
Visual export (vector PDF, colour-blind-safe line styles + colour) → scientific-plotting.
Clarity and honesty
- State caveats and limitations; do not bury serious doubts only in an appendix.
- Prefer concrete ids (
method v2, N=512) over vague "the model".
- Cite claims; read cited papers — verify they support your statement.
- No hype without numbers.
No em dashes
Do not use em dashes (U+2014) in manuscript text, captions, or user-facing reports.
Use comma, parentheses, colon, separate sentences, or en dash for numeric ranges
(e.g. 10--20 samples). Hyphenate compound modifiers (cross-validated).
Self-check before submitting text
- Search for
— and replace every instance.
- Skimmer test: result clear from title + abstract + figure captions alone?
- Methods: could a colleague reproduce the analysis from this section?
- Discussion: all seven aspects addressed; limitations stated?
- Conclusions: strong final sentence, not only "more research is needed"?
- Formal tone: no contractions, phrasal verbs, or imperative-to-reader?
- Numbers in abstract/conclusion match manifest-backed figures?
Integration
- paper-writing-workflow: planning, literature review, compile/VLM loop.
- plotting/ skills (scientific-plotting, plot-check, vlm-figure-audit): figures.
- paper-layout-review: layout only — not wording.