EN
Back to the archive

The encyclopedia · Software & IT · Technical decision · 1875–2019

The metric system won by being decimal and tied to nature, then a treaty made it global

France's decimal measurement system beat royal standards because units came from nature and scale by ten; the 1875 Metre Convention made it global.

Bureau International des Poids et Mesures (BIPM) · General Conference on Weights and Measures (CGPM)

The solution

Before the metric system, units varied by region and ruler, so trade and science could not compare measurements. The system that became SI made units easy to scale: twenty-four decimal prefixes from 10^30 to 10^-30 turn one base unit into any magnitude.

SI's base units are defined by natural constants, such as the speed of light and atomic frequencies, so the definitions do not depend on a particular king, artifact or country. NIST calls SI, commonly known as the metric system, "the international standard for measurement."

The political lock came on May 20, 1875, when 17 countries, including the United States, signed the Metre Convention in Paris, creating BIPM, the CGPM and the CIPM: permanent bodies whose job is to keep the world's measurement definitions in agreement.

Why it worked

  • Decimal prefixes make conversion a matter of moving the decimal point.
  • Natural-constant definitions hold for anyone, anywhere.
  • One treaty body prevents national units from drifting apart.
  • A common system lets technical documents cross language barriers.
What it achievedDecimal units defined by nature, kept by one treaty bodyclever

What can be applied

A standard wins when its units scale simply, anchor to something universal rather than an authority, and are kept by a body everyone trusts; comparability then becomes a public good nobody leaves.

Aftermath

SI is now the international standard for measurement, used in international commerce and scientific research worldwide; the Metre Convention's anniversary, May 20, is celebrated around the globe as World Metrology Day.

Sources

spotted an error? The archive wants to know.

Related cases