Literature notes that preserve the source and feed the draft

Begin with a question, not an empty notebook

Write the question that brought you to the source before taking notes. ‘What makes readers trust an explanatory guide?’ will produce different notes from ‘What examples does this author use?’ A question gives attention a filter and makes it possible to stop reading when the source has answered what you need.

If you are exploring rather than researching a defined piece, use a provisional question such as ‘What is this author changing my mind about?’ The question can evolve. Its job is to prevent every interesting sentence from receiving equal weight.

Add a stopping rule to the reading question. You might stop when you can name three competing explanations, locate the method behind a quoted result, or decide whether the source belongs in a chapter. Without a stopping rule, note-taking expands to match the length of the source. With one, you can capture an unexpected insight without confusing comprehensive coverage with the evidence your project actually needs.

Create a source record once

At the top, record author, title, publication, date, stable link or identifier, and the edition or version. For a PDF, note page numbering quirks. For audio or video, record duration and timestamps. Save a copy when the material is load-bearing or likely to disappear.

Add two sentences in your own words: what the source argues and why it matters to your project. If you cannot write them after reading, the source may be background rather than evidence. Mark it accordingly instead of pretending every item in the library is equally understood.

The source record should hold the bibliographic details, stable link or identifier, edition, access date when relevant, and the file you actually read. Add one sentence on authority and one on limitations. A preprint, revised edition, translated passage, and publisher summary are not interchangeable. Recording the exact object prevents a later citation from pointing to a cleaner-looking page that does not contain the evidence you used.

Use a four-part note for each useful idea

Each note contains: the source claim, the evidence or reasoning offered, your response, and a locator. Keep direct quotations visibly inside quotation marks. Label paraphrases. Put your own interpretation in a separate field so it cannot quietly turn into something the author said.

Title the note as a claim - ‘Visible revision history changes how students explain authorship’ - Rather than ‘notes on chapter four’. A claim title makes scanning useful and allows one source to produce several independently retrievable ideas without losing their shared provenance.

Keep four labels visible inside each note: source claim, exact evidence, your interpretation, and uncertainty. For example, the author claims that delayed feedback changes revision behaviour; the evidence is table 2; your interpretation is that timing may matter more than comment volume; uncertainty is that the study involved one course. The labels stop a promising inference from slowly turning into something the source supposedly proved.

The useful unit of reading is not a highlight. It is a claim you can find, explain, and place.

Write the bridge to your project

End the note with one line beginning ‘This could help me…’. Name the draft section, question, counterargument, example, or decision it might support. This is not a commitment to use the source; it is the bridge that lets future-you understand why the note entered the system.

If no bridge exists, store the item as background or let it go. Collecting source summaries with no imagined use creates a second reading list, not a knowledge base.

Write more than 'use in chapter three'. Name the paragraph job: define the competing model, qualify the opening claim, provide a counterexample, or explain why the method does not transfer. Also state what must sit beside the note. A statistic may require the sample description; a quotation may need the author's opposing example. The bridge should tell your future self how to use the material responsibly, not merely where to paste it.

Connect agreement, disagreement, and dependence

Link notes when two sources answer the same question differently, use different definitions, or depend on the same dataset. Label the relationship. ‘Disagrees because the population differs’ is more useful than an unlabelled backlink.

These relationships become the structure of a synthesis later. A literature review is rarely a procession of summaries; it is an account of patterns, disagreements, methods, and gaps. You discover those shapes while connecting notes, not while copying highlights.

Relationships become useful when they are stated as sentences. Note A agrees with B about the mechanism but uses a different population. Note C disputes their measure rather than their conclusion. Note D depends on the definition introduced by A. Those distinctions produce a literature review with movement. A plain 'related' link only says that two notes share a neighbourhood and leaves the intellectual work for later.

Close the source with a short memo

When you finish, write what changed in your understanding, which claims are strong enough to use, what remains uncertain, and the next source or question the reading suggests. Five minutes is enough. This closure converts a pile of notes into a position.

Do not polish every note. Correct the ones that enter a draft, verify their quotations and locators, and leave the rest as working material. The system exists to support writing, not to produce a perfect museum of everything you have read.

Close with a five-sentence source memo: the question asked, the answer offered, the strongest evidence, the most important limitation, and the job this source may perform in your project. If you cannot write the memo, the source may need another pass or may not deserve a place. The memo also lets you reject a source without losing the reason, which prevents the same unproductive reading loop months later.