Twenty minutes after an interview ends, the best line of the week is already going soft at the edges. The page number has gone, the name of the person who said the sharp thing has gone, and the conversation has collapsed into a handful of scribbles that made sense at the time. The Cornell method will not rescue a bad interview, but it stops good material from being wasted.

Devised for students revising lectures, the layout suits writers at least as well, because it keeps apart the two things that usually get muddled: the raw material and your thinking about it.

The layout, adjusted for source work

An ordinary Cornell page has three zones. A narrow column down the left, roughly a third of the width, for cues. A wide notes area on the right for the material itself. A summary block across the bottom, two or three inches deep, written after the fact.

For research and interviews, add a fourth zone: a source strip along the top of every page. It takes fifteen seconds to complete and saves an afternoon of hunting later. On a page of interview notes it holds the speaker's name, the date, the recording file name. On a research page it holds author, title, publisher, year, plus page numbers if you are working from a book, or the web address and the date you accessed it if you are working online.

• Source strip — who or what this page is about, and how to find it again.
• Notes column — quotations, paraphrases, observations, all with locators.
• Cue column — short labels, cross-references, gaps, candidate uses.
• Summary block — three to five sentences in your own voice.

The discipline is in the timing. Cue column and summary are filled in after the reading or the interview, never during. Notes first, thinking second. That order is the whole method.

What goes in the notes column

The notes column is the evidence locker, so keep it literal. Quote exactly, in quotation marks, and always with a locator: a page number, a paragraph number, a timestamp. If a sentence is too long to copy in full, quote the part that matters and paraphrase the rest outside the quotation marks. Never paraphrase inside them.

Mark your own reactions differently: a symbol in the left margin, a different pen, square brackets. When you return in three weeks you need to know at a glance whether "this is the whole argument in one line" came from your source or from you at half past eleven at night.

For interviews, anything that might become a quote gets written in full and word for word, hesitations included if they carry meaning, with the timestamp beside it. Everything else is paraphrased as you go. Guessing at a half-heard word is how a misquote is born; put a wavy line under it instead and check the recording.

The cue column is where the thinking happens

Filled in badly, the cue column becomes a second set of notes. Filled in well, it becomes the index to your project. Leave it empty while you read or listen, then go back and add:

• theme labels in your own vocabulary ("cost of repair", "reluctance to complain")
• cross-references to other pages ("see BV-04; contradicts p.12")
• gaps the source leaves open ("no figures before 1998")
• candidate uses ("possible opening quote", "background only", "check this claim")

A cue entry should be short enough to scan in a second. If you find yourself writing whole sentences, they belong in the summary.

Interview notes: capture now, clarify soon

Before

Write your questions down the cue column and number them. You will not follow that order, and that is fine. The list exists to stop you leaving with a hole where the obvious question should have been.

During

Notes column only. Verbatim quotes in quotation marks with timestamps; everything else in your own words. Keep a running note of what surprised you, because surprise is usually a sign that a good question is nearby.

Afterwards

Within an hour if you can: expand your abbreviations, tidy the timestamps, fill in the cue column, and write the summary while the voice is still in your head. Then check the quotations against the recording before you file the page. If a line cannot be verified, paraphrase it or leave it out. A slightly weaker sentence you can stand behind beats a brilliant one you cannot.

Keeping sources straight over a long project

For a single article, one page per source is usually enough, with the cue column carrying the cross-references. For a book, or a subject you keep returning to, try one page per theme instead, with source codes in the margin. A code such as BV-04, initials plus a running number, is quicker to write than a full citation and easy to trace back through an index page at the front of the notebook. Number every page as you make it, and keep that index updated. It takes a minute and pays for itself the first time an editor asks where something came from.

A digital equivalent works the same way. A two-column table with a summary row underneath reproduces the layout closely enough, and search replaces the index, provided you stay consistent with your source codes. The medium matters far less than the four zones.

A routine that survives a busy week

1. Set up the page before you begin: source strip, cue column, notes column, summary space.
2. Take notes in the notes column only. Do not think in the notes column.
3. Within a day, fill in the cue column and write three to five sentences in the summary block: what this source says, what it is good for, where it might sit in the piece.
4. Verify every quotation against the recording or the page before it reaches a draft.
5. Before you start writing, read only the cue columns and the summaries. That is your outline, and it already has your voice in it.

Cornell notes earn their keep not because the layout is clever but because it forces two short acts of reflection, a minute on the cues and two on the summary, at exactly the point where most notes are abandoned. For a writer with six sources and a transcript, that is often the difference between a folder of material and a draft.

Photo: DS stories / Pexels