Time Zones and UTC, Explained
Time zones look like a tidy grid of 24 slices. They are nothing of the sort. They are a political artefact with roughly 38 offsets in active use, several of which are not whole hours, and understanding why saves a surprising amount of grief.
Before standard time, noon was local
For most of history, noon meant the moment the sun stood highest where you were standing. Every town kept its own clock, and towns a few dozen miles apart genuinely disagreed by several minutes. Nobody minded, because nothing moved fast enough for the discrepancy to matter.
Railways ended that. A timetable spanning many towns is incoherent if each town defines its own noon, and the consequences of incoherent timetables on shared track are not merely inconvenient. Britain's Great Western Railway standardised on London time in 1840, and "railway time" spread across the network through the 1840s. In North America the railroads imposed a system of standard zones on 18 November 1883 — a date sometimes called "the day of two noons", because clocks in many cities were reset mid-day.
Note the sequence: the railways did this themselves, ahead of governments, because they needed it. Legal recognition followed years later.
Greenwich and the prime meridian
With zones came the question of where to count from. The International Meridian Conference met in Washington in October 1884 and adopted the meridian through the Royal Observatory at Greenwich as the world's prime meridian, largely because a majority of shipping already used charts based on it. The choice was pragmatic rather than principled, and France abstained, keeping Paris time as its legal reference for decades afterwards.
GMT, UTC, and why the name is strange
Greenwich Mean Time is astronomical: it is derived from the rotation of the Earth. The Earth, unhelpfully, is not a precise clock — its rotation wobbles and is very gradually slowing.
Coordinated Universal Time replaced it as the world's reference. UTC is maintained by atomic clocks, which are vastly more stable, and is then nudged to stay within a second of astronomical time by the occasional insertion of a leap second. It took its modern form in 1972.
The acronym is a deliberate compromise. English would give CUT, French would give TUC; the standards bodies chose UTC so that neither language won. In everyday use people say GMT and mean UTC, and for most purposes the difference is under a second and irrelevant. For anything involving logs, finance, or coordination, use UTC and say UTC.
The grid that is not a grid
Offsets are set by governments, and governments optimise for trade, politics, and daylight rather than geometry.
- Half-hour offsets. India is UTC+05:30. Iran is UTC+03:30. Newfoundland is UTC−03:30. South Australia is UTC+09:30.
- Quarter-hour offsets. Nepal is UTC+05:45. New Zealand's Chatham Islands are UTC+12:45.
- Zones that ignore geography. China spans roughly five geographic zones and observes one, UTC+08:00, so the sun rises well after 9am in the far west during winter. Spain sits mostly at Greenwich's longitude but keeps Central European Time.
- Offsets past the 12-hour mark. Kiribati's Line Islands are UTC+14:00, which is why the date line on a map is a zigzag rather than a straight line, and why two inhabited places can be on different calendar dates for a full day.
The practical consequence: never assume an offset is a whole number of hours, and never assume it is stable. Countries change their minds, sometimes with only weeks of notice.
The database underneath all of this
Because offsets are political and change, software does not hardcode them. It consults the IANA Time Zone Database (still widely called the tz or Olson database), a continuously maintained record of every zone's current rules and its historical ones going back decades.
This is why correct systems identify a zone as America/New_York rather than EST. The identifier names a place and its rule set; the abbreviation names only one of that place's two seasonal states, and does so ambiguously — CST alone could mean central time in North America, China, or Cuba. Zone identifiers use a region and a representative city, and the city is chosen for historical stability, which is why the list contains America/Argentina/Buenos_Aires and other apparent oddities.
Rules that prevent most mistakes
Almost every timezone bug reduces to one of a few errors.
- Store instants in UTC; convert only for display. An instant is unambiguous. A wall-clock reading is not.
- Store a zone identifier, not an offset. An offset records what was true on one date. The identifier survives rule changes.
- Never do arithmetic on local wall-clock time. "Add 24 hours" and "same time tomorrow" are different operations on a clock-change weekend, and both are sometimes correct depending on what you meant.
- Treat a future meeting as a zone plus a local time. Storing it as a fixed UTC instant means it silently moves for participants if a government alters the rules before the date arrives.
- Name the zone when you write a time to a human. "15:00 UTC" and "3pm your time" are both fine. "3pm" alone is not.
A note on scheduling across zones
The hard part is rarely the conversion — it is finding a window that is not unreasonable for anyone. Three participants spread across the Americas, Europe, and East Asia frequently have no overlapping working hours, and the honest answer is to rotate the inconvenience rather than pretend an option exists.
The World Time page shows several cities at once with a 24-hour overlap strip and marks the hours where working days actually coincide; the Tools page converts a specific instant between any two zones.
Clock changes add a second layer to all of this — see Daylight Saving Time, Explained.