Did you know that a beautifully completed science table can become misleading because of two small paper labels? A team weighs two pots, moves them to different locations and continues the experiment. Halfway through, someone notices that the labels on the pots may have been exchanged. The numbers are still neatly written. The problem is that the meaning of those numbers is now uncertain.
A sample-label swap occurs when the identifying label attached to one experimental sample is accidentally assigned to another. The first correct response is to pause affected comparisons, preserve original records and check whether independent identifiers or dated evidence can restore the mapping. If the team cannot tell which physical sample produced a result, it must mark that result as uncertain rather than guess a label. Repair the identity trail, not merely the spreadsheet.
There are two kinds of information in every experimental row
A school science table normally contains a measurement, such as 585 g, 14 cm or 22 °C. It also contains an identity: which pot, container or specimen produced that measurement and under what treatment. A measurement without a reliable identity cannot answer a treatment-comparison question, even if the instrument was accurate.
Suppose Pot A receives light from a window and Pot B is placed under a different condition. If the label A is moved onto the second pot, the team could correctly measure 585 g and still assign it to the wrong group. This is not an arithmetic mistake. It is a provenance mistake: the data have lost their reliable connection to the physical object.
A parallel concern exists in professional laboratory practice. The NIST Biorepository’s description of specimen records explains why controlled handling, associated metadata and a traceable sample history matter. A school pot experiment does not follow that specialised facility’s protocols, but it benefits from the same simple principle: preserve which object produced which observation.
Do not fix a label problem by changing numbers
The temptation is to open the spreadsheet and exchange the rows labelled A and B until the results seem sensible. That is not a reliable repair. Which treatment should yield a larger change may itself be the question the experiment is testing. Selecting the mapping that makes the expected hypothesis look correct destroys the fair test.
Instead, separate the question ‘What did we measure?’ from ‘Which sample was it?’. Original numbers may remain valid as unassigned measurements. Sometimes photos, fixed markings or teacher notes can establish identity. Other times the evidence is insufficient, and the valid outcome is to report an identification limit and repeat the affected comparison if feasible.
The five-part evidence check
- Freeze the current record: retain the original notebook, files, photographs and timestamps. Make a separate corrected copy or clearly marked revision.
- Identify the interval of uncertainty: find the last moment at which each pot’s identity was certain and the first point at which the labels were discovered swapped.
- Check independent identifiers: permanent ID marks, distinctive physical features already documented, dated position charts or supervised handling logs may help.
- Apply one mapping consistently: if evidence supports a repair, update the identity labels with a written explanation without overwriting the measured quantities.
- Mark what remains unknown: label ambiguous results as uncertain, and avoid using them as confirmed support for one treatment.
This process is not about blaming whoever moved the pots. In ordinary group science, labels can fall off or students can misunderstand how a tray is arranged. What matters educationally is whether the team can reconstruct evidence responsibly, not whether it can find a culprit.
Worked example: two pots, two labels and a hidden identifier
All numbers and events are fictional examples, not real pupils’ results. A class uses two sealed-down tracking tags that cannot easily be moved: P101 on one pot and P102 on another. Separate removable cards say ‘Treatment A’ and ‘Treatment B’. At the start, P101 is A and weighs 600 g; P102 is B and weighs 620 g. Assume the same balance and unit convention were used, with weighings taken at stated times.
At the next measurement the scale reads 585 g for P101 and 598 g for P102. A photograph, taken before the second weighing and showing both permanent IDs beside the containers, is available. Unfortunately the removable cards now show P101 as B and P102 as A. The team realises the treatment cards were accidentally interchanged while the pots were being rearranged.
We can calculate total mass changes using permanent IDs. For P101, 600 − 585 = 15 g decrease. For P102, 620 − 598 = 22 g decrease. The hand check is 585 + 15 = 600 and 598 + 22 = 620. These are total pot mass changes, not automatically measurements of plant transpiration: soil and surface water may also contribute.
Because dated original records establish that P101 was the A pot and P102 was the B pot, and because the same permanent IDs can be read clearly at the second measurement, the team can correct the removable treatment labels if there is no separate evidence that the treatments or contents themselves were exchanged. The revised table should say P101/A: 600 → 585 g, decrease 15 g; P102/B: 620 → 598 g, decrease 22 g. The numbers remain unchanged; their treatment associations have been restored from independent identity evidence.
Now remove the permanent tags and the dated photograph from the invented story. The only clue is ‘the A pot probably lost less mass’. That is not evidence of identity; it simply assumes the answer. The team should mark both treatment assignments unresolved. It can still report two measured changes of 15 g and 22 g for unidentified pots if their paired baseline measurements are genuinely linked, but it cannot claim which treatment produced each decrease.
What counts as a genuinely independent identifier?
A permanent written code on the pot, recorded in a dated setup log, is usually stronger than an easily moved paper card. A distinctive feature documented before the mix-up can sometimes help, but similarity between two objects makes visual inference weak. A position in a tray may be informative only if the tray layout did not change. A student’s memory may suggest where to look, but should not silently override contrary records.
There is another complication: the physical samples might have switched positions as well as labels. A valid identity repair does not necessarily recover which conditions each pot experienced during the uncertain interval. For example, knowing that P101 is a particular pot does not prove that it received the intended lighting treatment throughout the day. The difference between sample identity and treatment history must be preserved.
Diagnose the learner’s reasoning
Ask: ‘A value of 585 g is clearly written. Can we call it Treatment A’s result?’ A strong student says ‘only if we can connect that weighing to the treatment identity with evidence’. Next ask: ‘If Treatment A was expected to lose less mass, can we assign the smaller loss to A?’ The answer is no; that would turn the hypothesis into the means of identifying data.
A third check asks: ‘What did the permanent codes add?’ They made it possible to connect each sample’s earlier and later measurements without relying on the movable treatment cards. The student should also notice that this still does not prove a condition stayed constant if the pots changed location.
Progressive practice: preserve, reason, then report
Guided: repair a documented card exchange
Two labelled boxes have permanent codes B11 and B12. An approved photo shows B11 contained six counters and B12 contained nine. The temporary display cards were exchanged, but permanent codes remained visible. A pupil reads them and assigns six to B11 and nine to B12. The check is not ‘which total seems more likely?’ but ‘which code appears in the documented photograph and current record?’
Supported: calculate without assuming identity
Two unlabelled containers each lose mass. One changes from 150 g to 143 g, a decrease of 7 g. The other changes from 170 g to 159 g, a decrease of 11 g. If their earlier-later measurements are reliably paired but treatment cards are missing, record the mass changes as Container 1 and Container 2. Do not name the conditions. A pupil must explain precisely which comparison remains impossible.
Independent: uncertain conditions after relocation
Give learners a table of four containers, a setup sketch and three timestamped photographs. One photograph is ambiguous and two labels are missing. Ask them to reconstruct only the mappings supported by evidence, flag intervals without known treatment history and propose a better coding method. Award full credit for a well-justified ‘unknown’ rather than an unverified reconstruction.
How parents and teachers can build better science habits
A teacher can ask each group to assign a stable identifier before an experiment and keep it separate from a treatment label. Show pupils how to write ‘P101 — Treatment A — window shelf’ in the setup log and photograph the tray only when allowed. Each row should carry identity, condition, unit and time. Build a five-minute end-of-session identity check into practical work.
A parent can help by asking, ‘How do you know that this week’s number belongs to the same item as last week’s?’ This question applies to seedling journals, reading logs and home energy tracking. The answer should point to a stable identifier or a clear method rather than to a guess based on the result.
Observable progress is a pupil who keeps raw data intact, finds the last reliable point in the record, explains exactly which conclusions remain supported, and designs a prevention plan. Repeating this skill with unfamiliar samples shows transfer better than solving the same two-pot story twice.
What to do when the experiment cannot be rescued
Sometimes labels are the only identity evidence, and they are lost. That is an actual limitation, not a reason to fabricate a neat table. Say: ‘Two outcomes were measured, but we cannot reliably connect them to treatments after the labels were exchanged. Therefore we do not draw a treatment comparison from those results.’
If time permits, a teacher may help the group repeat the relevant measurements with stable IDs. If repetition is impossible, the project can still discuss the procedural weakness and what would change next time. Such a report may be educationally more valuable than a polished but untrustworthy conclusion.
In sensitive research settings, sample mix-ups can have serious implications. Classroom exercises must not be treated as instructions for clinical, forensic or safety-critical sample handling. Follow the qualified institution’s procedures in those settings.
Frequently asked questions
May we erase the wrong labels from the original notebook?
Preserve the original entry and add a dated correction note. The revision should identify the independent evidence that supports it. Silent overwriting makes later review unnecessarily difficult.
What if we can still calculate the average?
You may calculate an average for a clearly defined set of reliably paired numerical measurements, but not present it as the average for a named treatment when the sample-to-treatment assignment is uncertain. Correct arithmetic cannot repair an unknown identity.
Can photographs always prove a sample’s identity?
No. Photos may lack a visible stable ID, be taken at the wrong time or show similar objects. A useful photo must be tied to the relevant moment and actually display the distinguishing evidence. Our guide to comparing plant photos explains a related image-evidence pitfall.
What if someone changes the spreadsheet after we discover the mix-up?
Preserve the current version, record the original entries and repair only independently justified mappings. The Core Aim guide to conflicting spreadsheet edits explains how to keep the evidence while correcting a shared file.
The most important result may be an honest limitation
Science is not merely collecting numbers. It is preserving which observation belongs to which object and under which conditions. A thoughtful student learns that one unexplained label is enough to invalidate a particular comparison, but not a reason to invent a solution.
For more science learning, see the verified secondary science learning guide, the eduKate Education Hub, the Core Aim learning hub and the Clementi diagnosis-first teaching reference.