Six months into anything – a renovation, a thesis, a side business, a long repair – you can usually remember what you did. What you cannot remember is why you chose it over the other option, what the result actually was, and what you had decided to do next before the project went quiet.
There is a published specification for exactly this problem. The National Institutes of Health issued Guidelines for Scientific Record Keeping in the Intramural Research Program at the NIH through its Office of the Director, first edition December 2008, prepared by the intramural scientists on the NIH Committee on Scientific Conduct and Ethics. It is short, free, and considerably more useful to a person building a shed than most things written for people building sheds.
I have written about notebooks that were kept as working records on this site more times than I can defend without ever handing anyone the format. This is the format.
The Ten Fields
The guidelines say a record should describe or explain ten things. Reproduced as the document words them, with what each one means when the project is yours:
- Who did it (the person making the record). Trivial alone, essential the moment anyone else touches the project – a contractor, a co-founder, a supervisor.
- What you did. The action, not the intention.
- When you did it (clearly stating month, date and year). The year is the one people drop, and the year is the one that matters in three years.
- Why you did it. The single highest-value field and the one that evaporates fastest.
- What project the research was part of. Which strand this belongs to, if you have more than one running.
- How you did it (including the methodology). Enough that you could do it the same way again by accident.
- What materials were used. Brands, versions, sources, part numbers, the specific tin of paint.
- The findings. What actually happened, including when nothing did.
- Your interpretation. What you think it means.
- The next step. What you decided to do about it.
The document’s own summary of the standard: “Records should be legible, clear, timely, thorough, complete, secure, backed-up, and well-organized.” Of those eight adjectives, timely is the one doing the work. A record written a week later is a memory of a record.
And the sentence that justifies the whole exercise against the objection that a task list would do the job:
A laboratory notebook is a record of both physical and mental activity.
The Three Fields Nobody Writes
Fields one to seven are the cheap part. Most people who keep any kind of project log already record what they did, when, and with what. The three at the end – the findings, your interpretation, the next step – are the ones almost nobody writes down, and they are the entire reason to reread anything.
Here is why the gap matters. What you did is often recoverable from the world: the wall is painted, the chapter exists, the code is committed. What you concluded from it is not recoverable from anything. It lived in your head for about a day.
An illustrative entry, constructed for this article rather than copied from a real notebook:
12 May 2026 – kitchen, damp on the north wall.
What: lifted two floorboards at the skirting, left of the door. Moisture meter on the joist ends, six readings.
Why: damp patch has come back twice after repainting, so I wanted to rule out the floor before touching the wall again.
How: Brennenstuhl meter, pin mode, readings 40cm apart along the joist.
Materials: same meter as March, batteries replaced 2 May.
Findings: 18-22% along the joist ends nearest the external wall, 12-13% two metres in. Boards themselves dry.
Interpretation: it is coming in low and lateral, not down. The repainting was always going to fail. Probably the outside ground level.
Next step: measure the outside ground level against the damp course before ordering anything.
Delete the last two lines from that entry and you have a perfectly tidy record of a wasted afternoon. Keep them and you have the thing you will actually want in September, when you have forgotten every number and remember only that you did something with a meter once.
The reader’s test from the guidelines is worth stealing directly. Records should be kept “in sufficient detail to allow another scientist skilled in the art to repeat the work and obtain the same results.” Your version: could a competent stranger – or you, next year, which is nearly the same person – repeat what you did from what you wrote?
This is what Edison’s 3,500 notebooks and field notebooks kept for patient observation have in common, underneath the very different subjects. Neither is a diary of a working life. Both are records built so that the work could be picked back up.
The reason most project notebooks stop being useful is not that people abandon them. It is that they record the doing and not the concluding, so rereading them returns nothing. Small structural fixes like this one are most of what I write about. The newsletter is here if you want them as they turn up.
One Rule That Changes How the Notebook Reads
From the bound-notebook section of the guidelines:
Errors should be lined through with a single line, never erased or obliterated.
Corrections to earlier entries are to be initialled and dated. Data, the document says, “should be recorded in ways that cannot be altered.”
For an institution that is about audit and integrity. For a person it does something quieter and more interesting: it preserves the record of what you used to think.
A notebook you tidy as you go tells you that you were always right. A notebook where the wrong measurement is struck through once, still legible, with the correction dated underneath, tells you how long you held a bad idea and what finally shifted it. That is a genuinely useful thing to be able to see about yourself, and you can only see it if you resist cleaning up.
In practice, for a personal project: write in pen, strike through with one line, and never silently rewrite an old entry. If you use software, add a dated correction underneath rather than editing the original text.
Structure: Numbered Pages, Contents at the Front
The guidelines describe a bound volume with numbered pages, a table of contents listing the work at the front, and a list of abbreviations beside it. Data is to be kept “in a linear fashion without skipped pages” – the only pages left blank are the ones at the beginning reserved for the contents.
Three things fall out of that for a personal notebook:
- Number the pages before you start. Ten minutes, once. Nothing else makes a notebook retrievable.
- Leave the first two pages blank and fill in the contents as you go. One line per session: page number, date, what it was about.
- Do not skip forward to a fresh page because the current one looks messy. The gap is where the record breaks.
Anything that will not fit – a photo, a quote, a receipt, a printout – lives elsewhere and is indexed back to the notebook page, which is precisely what the guidelines prescribe for tissue sections and X-ray films.
Paper, Binder, or Software
The document is unusually even-handed here, and it is worth answering this question from its lists rather than from taste.
A bound notebook is portable and needs nothing but a pen. Against it: “Not easily searchable”, “May not have backup unless copied on routine basis”, and it requires legible handwriting.
A ring binder with loose sheets lets you use pre-printed sheets so the fields stay consistent, and lets you keep printouts and photographs next to the entry they belong to. Against it: pages “may ‘fall out'”, and it is still not searchable.
Electronic is searchable, legible, backed up if you set it up properly, and holds files and images natively. Against it: it needs a computer, it is risky around spills and theft, and data can become inaccessible as formats age.
That last disadvantage is the one people underweight, and it is why the honest answer for most personal projects is: whichever one you will actually open at the moment you finish working. The wider trade-off is covered in the wider trade-off between pen and screen.
The shape has survived the move to software, incidentally. The NIH’s current guidance for electronic lab notebooks still lists, for every entry: background, design, procedures, materials, observations and raw data, analysis, and “Interpretations, conclusion, next steps”. Sixteen years and a change of medium later, the last three fields are still the last three fields.
How This Differs From a Work Diary
They look similar and get read at completely different moments.
A five-minute log of the working day is about your employment. It is organised by date, it captures decisions and wins, and you read it when a review is coming. A personal-goals loop rather than a record of work is about intention: a goal, a weekly action, a daily check.
A project journal is about one body of work. It is organised by session rather than by day, it captures what you found and what you concluded, and you read it when you pick the project back up after a fortnight away. If you have a job and a project, you may well want both, and they should not live in the same notebook – the project one is consulted mid-work with dirty hands, and the work diary is not.
What It Is Not
It is not a project-management system. Nothing here schedules anything or tells you what to do next week.
It is not a legal instrument. The guidelines do discuss intellectual property, and the reason institutions care about unalterable records is patent disputes – but the safeguards that make a notebook count there involve witnessing, signatures and authentication that a personal notebook does not have. Do not assume yours proves anything to anyone.
And it is not evidence that the project will work. The argument for keeping one is that the record survives, which is a smaller and more reliable claim than the ones usually made for productivity systems.
Starting One Today
- Take any bound notebook and number the pages. Leave the first two blank.
- Write the ten field names on the inside cover. You will use six of them most sessions and you will forget the other four exist if they are not in front of you.
- Write at the end of each working session, not at the end of the day. Timely is the standard.
- Never skip the last three. Findings, interpretation, next step – even when the finding is “nothing worked and I do not know why”, which is itself worth dating.
- Strike errors through once. Date any correction to an earlier entry.
- Add one line to the contents page before you close the notebook.
Full specification, if you want it, is in the NIH’s guidelines for scientific record keeping – twelve pages, and about the same length as this article.
The rest of the formats on this site are in the rest of the template library. This one is the least emotional of them and quite possibly the most useful, because it is the only template here designed to be read by a stranger, and in six months you will be one.
If you want the version of this that runs across a whole journaling practice rather than one project, my newsletter goes out occasionally with whatever I have found in the source documents that week. And if you would rather work through it as a structured course, I made one – it is on Udemy, and 36,700 people have taken it, which still surprises me.
