Units of Measurement

A comprehensive collection of units of measurement covering 138 physical quantities and 11 measurement systems. Available in both JSON and JSONL formats.

Files

The jsonl/ directory contains JSONL files (one JSON object per line). The json/ directory contains the same datasets as JSON arrays. The jsonl/focused/ and json/focused/ directories hold curated subsets derived from the canonical dataset.

units_of_measurement.jsonl

The comprehensive, merged dataset. This is the file most users will want. It combines the SI and UOM datasets below into a single superset with a unified schema covering all fields.

Fields:

The combination of unit and property is unique across all entries.

Ontology Fields:

Where matches exist, external_ids and ontology_metadata provide cross-references to standard ontologies:

external_ids may contain:

ontology_metadata may contain:

si_units.jsonl

SI base units, SI derived units with special names, and non-SI units accepted for use with the SI (per the SI Brochure, 9th Edition). Each unit includes full SI prefix expansions (all 24 prefixes from quecto through quetta). Uses US English spellings ("meter", "liter").

Fields: unit, prefix, symbol, property, alternate_unit (optional), system

uom.jsonl

Units parsed from the Rust uom crate by Mike Boutin. Covers 117 physical quantities across multiple measurement systems, with conversion factors relative to the SI coherent unit for each quantity.

Fields: unit, symbol, plural, property, conversion_factor, conversion_offset (optional), reference_unit, system

units_with_ontologies.jsonl

Intermediate annotated copy of units_of_measurement.jsonl produced by scripts/annotate_with_ontologies.py. Contains the same records with external_ids and ontology_metadata fields populated where matches were found. Used as input to scripts/apply_ontology_annotations.py, which merges the annotations back into the canonical dataset.

ontology_crosswalk_base_units.jsonl

Seven SI base units with full cross-references across all three ontologies (UO, OM, UCUM). Useful as a quick reference for how the ontology identifiers map to fundamental units.

Fields: unit, dataset_properties, om (object), uo (object), ucum_code, sources

Measurement Systems

Conversion Factors

Each entry's conversion_factor is the multiplier to convert one unit to the reference_unit for that property. For example:

• kilometer: conversion_factor: 1000.0, reference_unit: "meter" -- 1 km = 1000 m
• liter: conversion_factor: 0.001, reference_unit: "cubic meter" -- 1 L = 0.001 m^3
• pound: conversion_factor: 0.4535924, reference_unit: "kilogram" -- 1 lb = 0.4536 kg

For temperature units with a conversion_offset, the conversion to the reference unit (kelvin) is: value_in_kelvin = value * conversion_factor + conversion_offset.

Dimension

The dimension field encodes each unit's SI base-quantity exponents as an object. The keys correspond to the seven SI base quantities:

Only non-zero exponents are included. For example, velocity (m/s) is {"L": 1, "T": -1} and power (W = kg·m²/s³) is {"M": 1, "L": 2, "T": -3}.

Dimensionless quantities — angle, solid angle, ratio, logarithmic ratio, and information — have an empty object {}. This follows SI convention: the radian and steradian are dimensionless derived units (dimension 1), and information units (bit, byte) fall outside the SI dimensional framework.

Usage

Install

npm install units-of-measurement   # Node.js
pip install units-of-measurement   # Python

Python

from units_of_measurement import load

units = load()  # all entries from the merged dataset

# Find all length units
length_units = [u for u in units if u["property"] == "length"]

# Convert 5 miles to meters
mile = next(u for u in units if u["unit"] == "mile")
meters = 5 * mile["conversion_factor"]  # 8046.72

# Canonical name & dimension vector for an acceleration unit
accel = next(u for u in units if u["unit"] == "meter per second squared")
accel["canonical_unit"]  # 'meter/second²'
accel["dimension"]       # {'L': 1, 'T': -2}
accel["quantity"] == accel["property"]  # True

# List all measurement systems
systems = sorted(set(u["system"] for u in units))

# Load a specific dataset
si = load("si_units")   # SI entries
uom = load("uom")       # uom entries

JavaScript

const { load } = require('units-of-measurement');
// or: import { load } from 'units-of-measurement';

const units = load(); // all entries from the merged dataset

// Find all mass units
const massUnits = units.filter(u => u.property === 'mass');

// Convert 10 pounds to kilograms
const pound = units.find(u => u.unit === 'pound' && u.property === 'mass');
const kg = 10 * pound.conversion_factor; // 4.535924

// Canonical representation + dimension vector
const accel = units.find(u => u.unit === 'meter per second squared');
console.log(accel.canonical_unit); // 'meter/second²'
console.log(accel.dimension);      // { L: 1, T: -2 }
console.log(accel.quantity === accel.property); // true

// Load a specific dataset
const si = load('si_units');
const uom = load('uom');

Raw Data (no package needed)

# Download as JSON array
curl -LO https://raw.githubusercontent.com/duncanscott/units-of-measurement/main/json/units_of_measurement.json

# Download as JSONL
curl -LO https://raw.githubusercontent.com/duncanscott/units-of-measurement/main/jsonl/units_of_measurement.jsonl

# Query JSON with jq — all Imperial length units
jq -c '.[] | select(.property == "length" and .system == "Imperial")' json/units_of_measurement.json

# Query JSONL with jq
jq -c 'select(.property == "length" and .system == "Imperial")' jsonl/units_of_measurement.jsonl

Focused Lists

jsonl/units_of_measurement.jsonl is the single source of truth, but you can build smaller slices directly from it by running:

python3 scripts/generate_focused_lists.py

This script reads the canonical JSONL and filters it into jsonl/focused/:

• si_base_units.jsonl keeps the eight SI base (and interval) units by checking the existing property/unit fields (e.g., (length, meter)).
• property_summary.jsonl aggregates counts per property, listing the observed system and reference_unit values and tallying how many entries already have external_ids.
• biomedical_units.jsonl includes any record that already carries a UO CURIE or a UCUM+OM pairing; no separate metadata files are required because it reuses external_ids and ontology_metadata.
• uo_units.jsonl lists every unit with a Unit Ontology identifier (pulled from external_ids.uo), along with the stored UO label/definition.
• ucum_units.jsonl lists every unit with a UCUM code (from external_ids.ucum) and includes the associated OM metadata so you can track the UCUM source.

Because each subset is derived purely from the fields in the canonical JSONL, re-running the script after any data change keeps the focused lists in sync without maintaining additional sources.

JSON Schemas

The schema/ directory contains JSON Schema (draft 2020-12) definitions for every JSONL dataset. To validate all files against their schemas:

pip install jsonschema
python3 scripts/validate_schemas.py

See schema/README.md for the full schema-to-file mapping and details on shared schemas.

Data Sources

• SI Brochure (9th Edition, 2019) -- Bureau International des Poids et Mesures (BIPM). Source for SI base units, derived units, and non-SI units accepted for use with the SI.
• uom -- Units of Measurement Rust crate by Mike Boutin, licensed under MIT / Apache-2.0. Source for conversion factors, plurals, and extended unit coverage across multiple measurement systems. See THIRD-PARTY-LICENSES for the full license text.
• Unit Ontology (UO) -- OBO Foundry ontology of units. Source for external_ids.uo CURIEs and ontology_metadata.uo labels/definitions.
• Ontology of units of Measure (OM) 2.0 -- Source for ontology_metadata.om URIs, labels, and definitions.
• UCUM -- Unified Code for Units of Measure (ISO 11240, HL7/FHIR). Source for external_ids.ucum codes, derived from the OM-to-UCUM mapping.

Acknowledgments

This project was built with assistance from Claude (Anthropic) and Codex (OpenAI).

License

This project is licensed under the MIT License.