Why Science? | Units of Measurement Ontology and Comparable Scientific Units

Science learning becomes powerful when a value can be read the same way by a pupil, a laboratory, a database and a computer program. The OBO Foundry describes the Units of Measurement Ontology, or UO, as metrical units for use with the Phenotype and Trait Ontology. Its official record lists use in the Ontology for Biomedical Investigations, the Statistical Methods Ontology and the HUPO Proteomics Standards Initiative mass-spectrometry vocabulary. UO gives units stable identifiers and definitions, but it does not decide whether an instrument was calibrated, a value was copied correctly or a conversion is scientifically appropriate. This guide supports Primary Science, PSLE Science, Secondary Science, O-Level Science, O-Level Biology, mathematics, computing and STEM education as a longform evidence-reading route; it is not a laboratory protocol, medical advice, an admission promise or a career guarantee.

Related eduKate reading: UCUM and unambiguous measurement units; Ontology for Biomedical Investigations; PSI-MS mass-spectrometry vocabulary; Information Artifact Ontology; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.

Primary and official sources checked on 13 October 2026: OBO Foundry Units of Measurement Ontology overview, checked 13 October 2026; Official Units Ontology source repository; Canonical Units Ontology OWL release; Units Ontology publication listed by OBO Foundry; 2026 Singapore–Cambridge O-Level Biology syllabus; MOE G2/G3 Lower Secondary Science teaching and learning syllabus; 2026 MOE G2 Computing syllabus. Ontologies, definitions and release products can change, so preserve dated documentation, exact source records and file versions.

How a unit turns a number into interpretable evidence

A bare number is ambiguous. “Five” could mean five seconds, five millimetres, five cells, five degrees Celsius or a category code. A unit gives the numeric value a metrical role, while the measured variable and method explain what the quantity means scientifically.

The OBO Foundry describes UO as metrical units for use with PATO. This makes a practical bridge between a phenotypic quality such as length or mass and the unit attached to its measured value.

The official UO page documents reuse by OBI, STATO and PSI-MS. Shared identifiers matter because the same unit can appear in an experimental design, a statistical result and a mass-spectrometry file without each project inventing a local code.

A unit identifier is more stable than a display symbol. The letter “m” may be interpreted differently outside context, and typography can blur micro, milli and metre. The identifier preserves the intended class while labels serve the reader.

Unit and quantity kind are different. Seconds and metres are both units, yet one measures duration and the other length. A dimensional check should fail before any conversion attempts to combine incompatible quantities.

Scale prefixes change magnitude. A milligram is not a gram with decorative text; losing the prefix changes a value by three orders of magnitude. Data pipelines should store the unit identifier beside every value rather than relying on a column heading that may be detached.

Some conversions use multiplication, while temperature scales can involve an offset as well as a factor. A generic “multiply by constant” rule is not safe for every unit system.

Precision must survive conversion. Writing more decimal places does not create more information than the instrument recorded, and rounding choices can matter near a scientific threshold.

Units do not establish measurement quality. A perfectly coded micrometre value can still come from an uncalibrated instrument, a biased sampling plan or a transcription error.

Counts deserve explicit modelling too. A count of colonies, reads or organisms needs the counted entity, inclusion rule, sampling area or volume and observation method, even when the formal unit is dimensionless.

The official repository distinguishes stable release files from an editor version used during ontology development. Reproducible work should pin the release actually used rather than silently following a changing branch.

The happy promise is controlled translation. When values, UO identifiers, dimensions, methods and uncertainty remain connected, people and software can compare evidence without pretending that every measurement is interchangeable.

Begin with the measurand: the property intended to be measured. Two teams may both report kilograms while measuring different things, so matching units alone never proves matching scientific meaning.

Keep the original reported value and unit alongside any normalised form. This allows a later reviewer to inspect the conversion rather than trusting a transformed number with no history.

Separate unit conversion from calibration. Conversion changes representation under a mathematical rule; calibration relates instrument response to a reference and carries its own evidence and uncertainty.

Reference conditions can matter for concentration, volume and gas measurements. Temperature, pressure, solution basis and specimen preparation may determine whether two values are comparable.

A ratio can be dimensionless yet still have a defined numerator, denominator and reference population. Removing those parts turns a meaningful rate into an unexplained decimal.

Missing units should fail visibly. Guessing from neighbouring rows or a familiar range may make a table look tidy while manufacturing scientific certainty.

Composite units need parsable structure. A label such as milligrams per litre contains a numerator, denominator and prefix that software should not recover from free text alone.

Coordinate and logarithmic scales require more than ordinary units. Decibels, pH and transformed intensities need the reference and transformation definition as well as the displayed symbol.

For classroom experiments, one measurement table can include value, unit, instrument, resolution, repeated readings and uncertainty. This turns “remember the unit” into an evidence habit.

Reader-ready reporting puts units in column headers for scanning and retains machine-readable identifiers at value or field level for reliable exchange.

Unit normalisation can hide source diversity. If every temperature is converted to kelvin, keep whether the original instrument reported Celsius or Fahrenheit and how many significant digits were available.

Error bars and uncertainty intervals have units too. A standard deviation of 2 is uninterpretable unless the underlying variable and unit are known.

Threshold decisions should be performed on validated quantities before display rounding. A value rounded to the cut-off can appear equal even when its stored measurement lies on one side.

A spreadsheet formula that converts units is part of the method. Preserve the formula or code, software version, factor source and tests rather than keeping only the converted column.

Unit aliases and spelling variants improve search but should resolve to one declared meaning. Synonyms must not merge distinct scales merely because readers use the same everyday word.

Dimensional analysis is a powerful low-cost test. It cannot prove a model is correct, but incompatible dimensions can reveal a broken equation or mapping before more expensive review.

Database schemas should not put one unspecified unit on a whole mixed column. Use validated field contracts or explicit unit references that can change only where the model permits.

When an ontology term becomes obsolete or is replaced, migration should document which records changed, whether conversions were involved and whether downstream calculations were rerun.

Before combining datasets, create a unit inventory: variable, unit identifier, scale, allowed range, missingness code, reference conditions, precision and source. This often exposes semantic mismatch early.

UO works best as one layer in an evidence system. Ontology identifies the unit; metrology, calibration, experimental design, provenance and statistical reasoning establish whether the value deserves comparison.

A four-stage evidence ladder for ontology-backed measurements

Stage	Question	Evidence	Responsible output
Observe	What was measured, counted or derived?	Specimen, variable, instrument, method and time	A bounded observation
Qualify	Which unit and scale apply?	UO identifier, symbol, definition and dimension	A machine-readable quantity value
Convert	Can the value be transformed safely?	Declared factor, offset, precision and reference conditions	A traceable converted value
Compare	Do the records answer the same question?	Compatible dimensions, methods, uncertainty and context	A responsible comparison