Did you know that a science laboratory can correctly report not detected even when it cannot establish that a substance is completely absent? The difference is not a trick. It comes from the limits of what a method can detect and measure. Three small phrases—not detected, below the detection limit and zero—can change what a student is allowed to conclude from the very same results.
Here is the useful answer: zero means the value is being reported as zero under the measurement or calculation being discussed. Not detected means the test did not find a detectable signal under its stated conditions. Below the detection limit adds a threshold: the method cannot reliably distinguish a signal at that low level. A non-detect is not proof of absence, and it should not be silently replaced with zero in a graph or an average. This is a reading skill that becomes a scientific reasoning skill.
A fictional water-test report you can actually interpret
Imagine a classroom simulation of a water-quality investigation. The chemical, method and figures below are invented for teaching, not readings from a Singapore reservoir and not a public-health finding. A report states the test’s detection limit for a hypothetical dissolved marker is 0.5 milligrams per litre (mg/L).
- Sample A: 0.9 mg/L — the report gives a measured value above the stated limit.
- Sample B: <0.5 mg/L, reported as not detected — the method did not reliably detect the marker at or above that threshold.
- Sample C: 0.7 mg/L — a reported measurement above the stated limit.
Which sample has no marker? You cannot identify one from this report. Sample B might contain no marker, or some concentration too small for this test to resolve reliably. The notation <0.5 does not mean ‘equals 0.5’, and ND is not the number zero. It is a report of how the result relates to a method’s limit.
A student who changes B to 0 without explanation has invented an exact number the experiment did not produce. A student who changes B to 0.5 has also converted a bound into a precise measured value. Both make an uncertainty disappear on paper.
Detection is not the same job as quantification
Scientists often distinguish a detection limit (DL) from a quantitation limit (QL). Detection concerns whether the method can reliably tell that something is present. Quantification asks whether it can report how much is present with a specified level of measurement reliability. Different laboratories and disciplines may use more specific terms or definitions, so read the report’s own method notes rather than assuming that every number labelled ‘limit’ means the same thing.
The US Environmental Protection Agency’s technical guidance warns against unjustifiably treating non-detects as zero and explains why detection and quantitation limits matter for decisions. Read the EPA guidance on chemical-concentration data near detection limits. It is a scientific source for understanding the words, not a substitute for Singapore-specific drinking-water regulation or an individual laboratory’s instructions.
Five report words students should recognise
- Analyte: the substance or property the test is intended to measure. Ask: What exactly are we testing for?
- Sample: the particular portion collected and analysed. One sample does not automatically represent every place or time.
- Detection limit: a method-specific lower boundary relevant to reliably detecting the analyte. Look for the reported units.
- Quantitation limit: a limit relevant to measuring an amount with stated reliability; it may be higher than the detection limit.
- Non-detect (ND), not detected, <DL: reporting language about the method’s ability to establish a signal or numerical result. The specific notation must be interpreted using the laboratory legend.
Notice the word families: detect → detection → detectable → undetected; quantify → quantification → quantitative; measure → measurement → measurable. They overlap, but they are not interchangeable. Undetected describes what the method did not establish. Absent is a much stronger claim about what is there.
The trap in an average
A class asks for the average marker concentration across Samples A, B and C. The detected values are 0.9 and 0.7 mg/L, and the third result is <0.5 mg/L. Can you calculate an exact three-sample mean from those three entries? Not without an additional treatment rule or information.
For a simple bounds exercise only, if the real concentration in Sample B is nonnegative and below 0.5 mg/L, the ordinary mean of the three underlying values would lie between (0.9 + 0 + 0.7) ÷ 3 ≈ 0.533 mg/L and (0.9 + 0.5 + 0.7) ÷ 3 = 0.700 mg/L; the upper boundary is not included under a strict <0.5 rule. This interval is a teaching deduction from the hypothetical numbers, not an uncertainty statement supplied by an actual laboratory. It illustrates why writing 0.600 or 0.700 as though it were directly measured can be misleading.
In real environmental analyses, qualified scientists select statistical treatments appropriate to the measurement process, number of non-detects and decision purpose. Arbitrarily assigning each ND a fixed value is not a universal method. Preserve the notation and consult the method guidance when doing calculations that matter.
What this has to do with Mathematics and G3 Science
In lower-secondary Science, a careful reader separates the observation from the inference: ‘the test did not detect X’ is an observation about the instrument and protocol; ‘X is absent everywhere’ is an unsupported generalisation. In Secondary 3–4 G3 Science, laboratory questions and data-handling tasks benefit from the same precision. Mathematics supports reading inequalities, measurement units, intervals, rounding and missing-value handling. These habits also matter later in applied science, quality assurance, environmental work, medicine and engineering, although professional conclusions need specialist methods.
There is also an English comprehension lesson. Compare: ‘No marker was detected in the sample’ and ‘There is no marker in the water.’ The first sentence talks about the result of a test on a sample. The second asserts something broader about the water itself. Changing the subject, verb or scope can quietly transform a careful report into an overconfident claim.
eduKateSG’s guide to blank data cells versus zero deals with the separate problem of missing records. A non-detect is different again: a test was performed, but the outcome is constrained by a measurement limit. For the wider route from words to precise explanations, see How English Works: Vocabulary and Meaning.
Four diagnostic questions to ask of a science table
- What was tested? Identify the analyte, sample location, collection time and method.
- What do the marks mean? Find the legend for ND, <, J, U or other qualifiers; do not guess their definitions.
- What is the limit, in what units? The same numerical figure can mean different things in mg/L, µg/L or a percentage.
- What conclusion is justified? Separate a measured number, a bound and a statement about health or safety. One lab row cannot by itself justify a sweeping safety claim.
Ten-minute practice for home or classroom
Ask learners to sort eight cards into four groups: exact reported value, bound or qualifier, missing record and inference that needs more evidence. Example cards: ‘0.7 mg/L’, ‘<0.5 mg/L’, ‘ND’, ‘not tested’, ‘the concentration is zero’, ‘sample collected on Monday’, ‘laboratory unable to quantify’, and ‘all rivers are safe’. Make students defend the sorting in a short sentence.
Then teach retrieval without looking: What is a non-detect? Why isn’t it automatically zero? Why do units matter? The next day, give a new example about light sensors that fail to detect a very dim signal. A week later, ask students to explain why an online search that returns ‘no result’ is not proof the information never existed. This last example is an analogy, not a statement that search engines use laboratory detection limits.
Common misconceptions
‘Not detected’ means a negative result, so the substance is absent. A negative test result may be accurate for the method and still leave open values below its capability. Respect what the test can conclude.
‘Below detection limit’ tells me the concentration equals the limit. No. A value preceded by the less-than sign is a bound, not a measured concentration equal to that number.
‘ND is missing data.’ Not necessarily. The measurement attempt may have been performed; a non-detect contains information about the method’s result and threshold. A blank untested cell is different.
‘Every non-detect can be changed to half the limit for averages.’ That is one possible convention in certain contexts, not a universal scientific rule. A method-sensitive analysis should document its choice and uncertainties.
Frequently asked questions
Can a device truly measure zero? Some instruments can report a zero under their operating definitions, calibration and resolution. The key question is what that zero means for the instrument and situation. A displayed zero is not unlimited proof of perfect absence.
Is an ND result good news? It may be encouraging in some contexts, but the significance depends on the substance, the limit, the application and the applicable standards. Do not make a health judgement from the letters ND alone.
What should a PSLE-level learner take away? Learn the safer everyday idea first: ‘I didn’t find it’ is not always the same as ‘It isn’t there.’ More advanced students can connect that distinction to detection thresholds.
Does an experiment that reports ND fail? No. A clearly documented non-detect can be a valid and useful result if the method and limits are appropriate to the question being asked.
One sentence worth remembering
Report what the test shows, not what you wish it proved. The vocabulary of detection, measurement and absence helps a student read Science carefully, calculate more responsibly and communicate uncertainty without hiding it.