The conventions of the paper notebook are old and remarkably stable. Kanare (1985) describes them in full: a permanently bound book with numbered pages, entries in ink made at the time of the work, each page dated and signed, unused space struck through, and errors corrected with a single line so that the original remains readable. A table of contents at the front makes the book searchable. The logic behind every rule is the same. A reader who was not present should be able to tell what happened, in what order, and that nothing was added or removed afterwards.

Regulated laboratories formalised these habits. The OECD Principles of Good Laboratory Practice require raw data to be recorded directly, promptly, accurately and legibly by the person entering them, signed or initialled and dated, and require that any change leaves the earlier entry visible and gives a reason (OECD, 1998). Later data integrity guidance condensed the expectation into the acronym ALCOA: records should be attributable, legible, contemporaneous, original and accurate. Guidance from the World Health Organization and from the United Kingdom medicines regulator extends the list with complete, consistent, enduring and available, and stresses that the same attributes apply to paper and electronic records alike (WHO, 2016; MHRA, 2018). For electronic records, the United States rule known as Part 11 adds requirements for audit trails and electronic signatures (FDA, 1997). A teaching laboratory is not bound by any of this, but the vocabulary is the one graduates will meet in industry.

The third strand comes from academic research, where the concern is less with proof and more with reuse. The FAIR principles ask that data be findable, accessible, interoperable and reusable (Wilkinson et al., 2016), and funders increasingly expect a data management plan that says how this will be achieved (Michener, 2015). Alongside these sits a practical literature on everyday habits: keeping raw data untouched, naming files consistently, organising one folder per project and backing up in more than one place (Noble, 2009; Borer et al., 2009; Wilson et al., 2017). Schnell (2015) applies the notebook conventions to computational work, which matters here because even a wet-laboratory course now ends at a spreadsheet or a sequence file.

Planning begins with a list. For each practical exercise the instructor writes down every record it will generate: notebook entries, printed protocols with annotations, gel and plate images, plate-reader and spectrophotometer exports, spreadsheets, sequence files, and the labels on tubes and plates, which are records too. Against each item go its creator, its format and whether it is raw or derived. The exercise is dull and worth doing, because it exposes the records nobody owns, such as the image left on the imager's hard disk.

The choice is between a bound paper notebook, an electronic laboratory notebook, and a hybrid in which the narrative is kept on paper and files are kept in a shared folder. Electronic notebooks offer search, time stamps and easy linking to data files, and published guidance on selecting and introducing them is available (Kwok, 2018; Higgins et al., 2022). They also need accounts, devices at the bench and someone to administer them. For a small teaching laboratory the hybrid is often the realistic option: paper survives spills and needs no login, and it teaches the conventions in their plainest form. Whatever is chosen, the plan should state the reason.

An entry standard tells a student what a complete entry looks like. A reasonable minimum is the date, a title, the aim in a sentence, the protocol followed with its version and every deviation from it, the reagents with lot numbers or stock identifiers, the instruments used, the observations as they occur, the names of the files produced, and a short interpretation written afterwards and marked as such. Entries are made during the session and not copied up neatly at home, since a fair copy is by definition neither original nor contemporaneous. Group work needs a rule of its own: each student keeps a notebook, and each entry names the partner who carried out any step the writer did not perform.

Digital records fail in predictable ways, and most of them can be designed out. A naming pattern that puts the date first in year-month-day order, followed by the exercise, the group and a brief description, sorts chronologically and can be read without opening the file. Spaces and special characters are avoided. Raw instrument exports are saved once, set to read-only and never edited; analysis is done on a copy (Wilson et al., 2017). Spreadsheets deserve particular care. Broman and Woo (2018) recommend a rectangular layout with one value per cell, consistent codes, no information carried by colour and a separate data dictionary. Their warning about automatic conversion is well founded, since gene symbols altered into dates by spreadsheet software have been found throughout the supplementary files of published papers (Ziemann et al., 2016). Open formats such as comma-separated text and FASTA remain readable when the software that wrote them is gone (Hart et al., 2016).

Sample identifiers tie the two halves of the record together. If the code written on a tube is the code in the notebook and in the file name, an image can be traced to a bench entry in seconds. A simple scheme of course, section, group and a running number is enough.

A record nobody reads soon deteriorates. The plan should say who reviews entries and when. In a teaching setting a teaching assistant can initial and date each notebook before the student leaves the bench, checking against a short list: dated, legible, deviations noted, files named, blank space struck through. This takes a minute per student and gives the countersignature of the bound-notebook tradition a teaching purpose. The correction rule is stated just as plainly. A wrong entry is crossed out with one line, the correction is written beside it with the date and initials, and nothing is erased, overwritten or torn out. In electronic files the equivalent is a dated note of what was changed and why. Figure 1 places review within the life of a record, from the planning of an experiment to the point at which its records are retained or destroyed.

The last step answers three questions. Where does each record live? A shared institutional drive with one folder per course, term and group is usually sufficient, with instrument computers emptied into it on a fixed day each week. How is it protected? The familiar rule of three copies, on two kinds of media, with one held elsewhere is a sensible target, and institutional storage often meets it without further effort (Hart et al., 2016). How long is it kept? Student work is normally held until marks are confirmed and any appeal period has passed, under the institution's own records policy, while protocols, strain lists and reagent logs belong to the laboratory and should outlive the cohort. The schedule names a person responsible for disposal, because records that are nobody's job to delete accumulate indefinitely.