Aliquot labelling and the laboratory record: date stamping and batch traceability

An aliquot label has to answer four questions without anyone opening a computer: which parent batch it came from, which aliquot of that batch it is, what it holds and at what nominal quantity, and when it was made — with the date written as YYYY-MM-DD, not in a form that reads one way in Athens and another in Chicago. The fifth element does not fit on the label and lives in the register: the entry tying the aliquot code to the supplier's lot number, to the certificate of analysis, and to whoever did the splitting. ISO/IEC 17025 states this as a requirement rather than a good habit: the laboratory must operate a system of unambiguous identification, retained while the item is under its responsibility, that ensures items cannot be confused physically or in the records ^[1].

The weak link is almost never the supplier's label. It is the second label, the one the laboratory writes when contents move into another container. The first is printed, checked and accompanied by a file; the second is often written in marker pen, on a surface never designed to be written on, by someone in a hurry who still remembers what they are doing. From the moment of splitting onward, traceability depends not on the supplier's quality but entirely on your own discipline in recording it.

What an aliquot label must state

No standard defines an "aliquot label" as a fixed form. There is a requirement of outcome, and it is strict: identification must survive transfer into a new container, so two items cannot be confused on the shelf or on the page ^[1]. Whatever achieves that is acceptable; whatever does not is decoration. The table below is the minimum set of fields that keeps the chain closed.

Where the label sits is part of the label. A cap is not a marking surface: caps get lost and swapped with the neighbouring container. The label does not cover the fill line, because visual inspection of the contents is part of the check. When the container is too small for every field, the answer is distribution rather than omission: identity, batch and aliquot codes and the date stay on the container, while the full information moves to the outer packaging or the rack position. That logic is not improvised — the CLP Regulation provides comparable derogations for very small packaging of substances placed on the market ^[9]. An internal aliquot does not fall under CLP, but the way the regulation divides information between inner and outer container is the right working model.

Date stamping: which date, and in what format

"04/03/2026" is not a date; it is two dates sharing one spelling. In Greece it reads as 4 March, in a US document as 3 April, and the difference surfaces exactly when a third party tries to reconstruct what happened. ISO 8601-1:2019 settles it with one rule: components run largest to smallest — year, month, day — in the form YYYY-MM-DD ^[6]. The side benefit is one no other format offers: alphanumeric sorting coincides with chronological order, so a folder of files sorts correctly with no conversion. Where a time is needed, write it on the 24-hour clock with an explicit offset from coordinated universal time.

Confusing an expiry date with a retest date is among the commonest transcription errors, and it is not a matter of terminology. ICH Q7 treats them as different instruments: material with an expiry date is not used beyond it, whereas material with a retest date may stay in use provided it is re-examined and found within specification; the first appears on the label and the certificate, the second on the label and/or the certificate ^[2]. Two consequences follow. An aliquot never acquires a longer life than its parent batch — splitting does not manufacture time. And a retest date is not copied across as though it were an expiry date: if nobody will perform the retest, the honest record is that beyond that point the material counts as uncharacterised.

The code that ties an aliquot to its parent batch

The most durable numbering scheme is a composite one: the parent batch code verbatim, plus a serial suffix for the aliquot. It has a property no internal sequence number has — it reads on its own. A code such as "B2471-A03" tells anyone, with no access to the system, that the container came from batch B2471 and is the third aliquot of the first splitting run. A bare "17" says nothing once the lookup table is lost, and the lookup table is always lost later than would have been convenient.

• The parent batch code is never shortened or "tidied" of hyphens and leading zeros; it is copied character by character from the supplier's label ^[1].
• A retired code is never reused for another container, even years later.
• Two batches of the same substance are never combined into one container, however identical their certificates — combining erases the traceability of both.
• A new label goes beside the old one, never over it; a label covering another is record deletion by another name.
• The aliquot is entered in the register when it is created, not at the end of the day: recording at the time of the action is an explicit documentation requirement ^[4].
• If material is reconstituted or diluted, the new container carries the parent batch code and a note that the contents are no longer in the form the certificate of analysis describes ^[2].

The register: fields, quantity reconciliation, retention

The label holds what has to be read by hand at the freezer; the register holds the rest. ISO/IEC 17025 requires technical records to contain enough information to identify the factors affecting a result, together with the date and the identity of the personnel who performed each activity ^[1]. On retention, the same standard leaves the laboratory to set the period, provided it sets it explicitly; it gives no number ^[1]. A useful yardstick comes from ICH Q7, which asks manufacturers to retain production and distribution records at least one year past the batch expiry date, and for materials carrying a retest date at least three years after the batch is completely distributed ^[2]. A research laboratory is not bound by those intervals, but it needs them as a starting point: a policy of "we keep what we can" is not a policy.

• Aliquot code and parent batch code, in exactly the characters shown on the label.
• Date of preparation and the identity of the operator, plus a second person where verification applies.
• Quantity placed in each aliquot and the quantity remaining in the parent container after splitting.
• Storage location, specific enough to find without searching: unit, shelf, rack, position.
• A pointer to the stored certificate of analysis and safety data sheet, with the lot number inside each filename.
• Every subsequent event for the container: opening, transfer, each freeze–thaw cycle, depletion, disposal.

One register field deserves particular attention, because it is the only one that detects errors by itself: quantity reconciliation. The quantities placed in the aliquots, plus what remains in the parent container, plus anything recorded as loss, must equal the original nominal quantity. When they do not, the gap is not rounding — it is an aliquot that was never entered, a label carrying the wrong number, or a transfer that did not happen as recorded. A register that does not close arithmetically is not incomplete; it conceals an error that someone will discover at a worse moment.

What makes a record acceptable: the ALCOA principle

Regulators have compressed what makes a record trustworthy into an acronym, and it is the most useful mnemonic in existence for a laboratory file. In the UK MHRA's data integrity guidance, a record must be attributable to a person, legible, contemporaneous with the action, original and accurate — and the extended form adds complete, consistent, enduring and available on request ^[3]. None of these properties is about technology; all of them are about the moment the record is written.

• Attributable: it is visible who performed the action and when; shared initials or a shared login destroy this property ^[3].
• Legible and enduring: written in indelible medium, on a material that does not vanish at the first wipe ^[4].
• Contemporaneous: written at the time of the action, not when the schedule clears ^[4].
• Original: the first record is kept; transcribing onto a neat form does not replace the original ^[3].
• Accurate and complete: it includes the inconvenient results too, with a note of what was done about them ^[3].

Correction is where a real record separates from a decorative one. A wrong entry is struck through with a single line that leaves the original readable, and the correct value goes beside it with initials, date, and the reason where it is not obvious. Anything that hides the original entry — correction fluid, a scribble, a new label over the old one — turns a correction into a deletion. Completing records retrospectively and back-dating them are not imperfections; data integrity guidance treats them as falsification ^[3], and the requirement that handwritten entries be clear, legible and indelible, made at the time each action is taken, is explicit ^[4].

Paper or software: the audit trail

The question "notebook or computer" is the wrong one. The real question is whether someone can change an entry and leave no trace. A bound notebook with numbered pages makes that hard; a spreadsheet makes it invisible, because a changed cell leaves nothing behind. European guidance on computerised systems asks, on the basis of a risk assessment, for a record of all relevant changes and deletions — an audit trail generated by the system itself — along with access control and backups ^[4]. A spreadsheet is not forbidden; it simply does not document. The minimum compensation is locking committed rows, named rather than shared accounts, and periodic copies in a non-editable format.

How often labelling fails — and what we do not know

The best available data on specimen identification errors comes not from research laboratories but from clinical ones, and it is useful only as an order of magnitude rather than a prediction. A multicentre Q-Probes study by the College of American Pathologists across 120 institutions recorded an overall rate of identification errors reaching released results of 55 errors per one million billable tests ^[7]. A longitudinal analysis at a US university hospital examined 16,632 specimen errors and found mislabelled specimens accounted for 1.0%, requisition mismatches for 6.3% and unlabelled specimens for 4.6% of the total ^[8]. Both date from 2006, in workflows with far higher volume and entirely different collection points from a laboratory splitting material at the bench.

For the one intervention that consistently appears to work there is data as well: in a paediatric oncology hospital, introducing bar-code identification accompanied a fall in mislabelled specimens from 0.03% to 0.005% within the twelve months following full implementation ^[5]. Honesty requires stating what the finding does not cover: it compares two time periods with no control group, in a setting where scanning happens at the point of collection. No published data series exists for labelling error rates on research-material aliquots — that literature has not been written. The mechanism transfers: hand-copying a code is the step that fails, and replacing it with machine reading removes one failure point. The numbers do not.

Labels that survive: freezing, condensation, solvents

There is no numbered "cryo-label" standard; there is only the requirement of outcome, that identification is retained while the item is under the laboratory's responsibility ^[1]. The failure modes are few and predictable: adhesive loses grip at low temperature and the edge lifts, condensation on the return to room temperature dissolves the ink, and a wipe with ethanol or isopropanol erases marker pen and inkjet printing alike. The test takes ten minutes per label type: stick one on an empty container, write on it with the medium you use, put it through one freeze–thaw cycle, then wipe it with ethanol. If the text smears or the label lifts, the problem is not the person writing. The second rule is redundancy: the same code written on the rack position makes a fallen label a nuisance rather than a loss of material.

The Greek framework: ESYD, ELOT EN ISO/IEC 17025, EOF

In Greece, laboratory accreditation is granted by ESYD against ELOT EN ISO/IEC 17025, the Greek adoption of the same standard that sets the identification and technical-record requirements above ^[1]. If you send analyses to a Greek laboratory, the question is not whether it is "certified" — a word that means nothing here — but whether that method falls inside its published scope of accreditation, and whether the certificate carries your lot number. EOF is the competent authority for human medicines and does not supervise research materials; it becomes competent the moment a material's labelling places it in the medicinal field, and that is exactly the risk a strictly descriptive label minimises. Where a material is classified as hazardous and placed on the Greek market, the hazard label is written in the official language of the member state ^[9]: internal aliquots may carry whatever language the laboratory works in, but their correspondence with the Greek-language safety data sheet must be obvious to someone who was not present when they were written.