Ask someone how a UTC offset gets its number, and most people will say something like: divide the world into 24 slices, 15 degrees of longitude each, and count how many slices you are from Greenwich. That answer is correct for exactly one moment in history, 1884, and wrong for almost everywhere on the map today. Real offsets are not calculated from longitude. They are looked up in a table that a small group of volunteers updates by hand, several times a year, based on what governments actually decide to do with their clocks.

That distinction matters more than it sounds. A weather app, a bank’s settlement system, and a flight booking engine do not run a formula when they show you a local time. They query a database of named rules, cross-reference it against a specific date, and apply whatever offset was legally in force on that date in that place. Understanding how that lookup actually works, rather than the tidy 15-degree story, explains why two cities at nearly the same longitude can differ by two and a half hours, why some countries use offsets ending in 45 minutes, and why software occasionally gets it wrong in ways that cost real money.

The 15-degree theory, and why it was already a compromise in 1884

The idea of slicing the globe into 24 equal hourly zones came from Sandford Fleming, a Scottish-Canadian railway engineer who missed a train in Ireland in 1876 because a printed timetable used a.m. instead of p.m. for a departure time, an ambiguity that local solar time made worse across every region running its own clock. Fleming spent the next eight years lobbying for a single worldwide standard. His proposal reached Washington, D.C. in October 1884, when delegates from 25 nations gathered for the International Meridian Conference at the invitation of President Chester A. Arthur. Over three weeks, 41 delegates debated and ultimately adopted the Greenwich Observatory as the zero-degree reference line by a vote of 22 to 1, with San Domingo (now the Dominican Republic) voting against and France and Brazil abstaining. France did not officially drop its own Paris meridian on maps until 1911.

Even at that founding moment, the 15-degree formula was already understood as an approximation, not a strict rule for civil timekeeping. The resolutions fixed the prime meridian and defined a universal day starting at midnight in Greenwich, but they left individual nations free to choose how, or whether, to align their local clocks to the resulting hourly bands. Most of the world did not adopt standard time zones for decades after the conference. Writing on the conference’s hundredth anniversary, one historian estimated that even a full century later, roughly half the world’s population still lived somewhere without a formally observed standard time.

Political geography overtook geometry almost immediately

Longitude tells you where the sun is overhead. It has never told you what a government will print on its official gazette. China Standard Time runs eight hours ahead of Coordinated Universal Time across the entire country, despite spanning roughly 60 degrees of longitude, a width that would justify four or five separate zones under Fleming’s original scheme. In the far west of the country, solar noon can fall well after 1 p.m. by the official clock, a gap of nearly two hours between the sun’s position and what a wall clock reports. A closer look at why China settled on a single national offset shows that the decision was administrative, not astronomical: one clock makes railway schedules, television broadcasts, and government coordination simpler across a vast territory, even at the cost of a visible mismatch between clock time and daylight in the western provinces.

This pattern repeats worldwide. A national offset is chosen for trade alignment, historical colonial ties, or the convenience of having one clock for one government, and geometry is adjusted around that decision rather than the other way around. Comparing how several single-zone countries arrived at their current offsets makes the pattern easy to see: economic integration, not longitude, decided the outcome in most cases.

Where the 45-minute and 30-minute offsets come from

Nepal’s case is worth working through because it shows the calculation the 15-degree rule was originally meant to produce, before politics rounds the number off.

Kathmandu sits at roughly 85°19′ east longitude. Divided by 15, that works out to 5 hours, 41 minutes, and 16 seconds ahead of UTC, the country’s genuine solar-based offset. From 1920 until 1986, Nepal instead used Indian Standard Time, UTC+5:30, for administrative convenience with British India and its successor state. On January 1, 1986, the government advanced its clocks by 15 minutes to UTC+5:45, a deliberate quarter-hour offset that both distinguishes Nepal from India on paper and sits closer to true Kathmandu solar time than the half-hour offset it replaced.

An offset ending in a quarter hour is not a rounding error. It usually marks the exact point where a government decided that being a few minutes off from geometry was worth more than being identical to a larger neighbor. worldtimedata

How a computer actually looks up an offset

No mainstream operating system, database, or programming language calculates offsets from a formula at runtime. Nearly all of them, including Linux, macOS, iOS, Android, Java, Python, and most SQL databases, pull from the IANA Time Zone Database, a public-domain dataset better known by its older name, the Olson database, after Arthur David Olson, who began compiling it in 1986. Paul Eggert has coordinated the project since the mid-2000s, and the Internet Assigned Numbers Authority formally took over hosting duties in 2011 after a copyright lawsuit briefly threatened the database’s availability. It now ships several updates a year and covers roughly 600 named zones.

Each zone is not a single offset number. It is a small history file. Entries follow a Region/City naming convention, such as America/New_York, Europe/Paris, or Asia/Kathmandu, chosen specifically because political names like country borders shift more often than city locations do. The America/New_York file, for instance, does not just say “UTC−5.” It records that the city ran on 4 hours, 56 minutes, and 2 seconds behind Greenwich Mean Time as pure local solar time until November 18, 1883, when railroads switched it to a clean 5-hour offset, then layers in every subsequent daylight saving rule change, including the shift to a March start date that took effect in 2007 under the U.S. Energy Policy Act of 2005. Querying a date in 1995 and a date next Tuesday against that same file returns two different, correctly calculated offsets, because the software is reading history rather than computing geometry. For a broader sense of how that reference point behaves independent of any single country’s rules, see how UTC functions as the anchor every named zone is measured against.

The sign-flip trap almost every developer hits once

One corner of the same database catches out programmers with surprising regularity. The Etc area includes fixed-offset zones such as Etc/GMT+5 and Etc/GMT-5, intended for platforms or ships that need a constant offset with no daylight saving rules attached. Their signs are backward from what anyone would guess.

The reversal exists because the tz database follows an old POSIX convention in which positive signs apply west of Greenwich, the opposite of the ISO 8601 standard almost everyone actually uses for writing offsets. PostgreSQL’s own bug tracker has fielded the same confused report more than once, with maintainers confirming that the database intentionally mismatches ISO output against POSIX-named internal zones because that mismatch is baked into the source data it depends on. The tz project’s own documentation calls the setup “the opposite of what many people expect,” and recommends against ever setting a system clock with a raw string like TZ=’GMT+4′, since the abbreviation it displays is actively misleading about which direction the offset runs.

The extreme edges, and what happens when a country moves the map

Offsets are also capped by international convention rather than by anything physical. The database’s widest gap runs between UTC-12 (Baker Island and Howland Island, both uninhabited U.S. territories) and UTC+14, a 26-hour spread that would not exist at all if every country strictly followed the original 15-degree, 24-zone plan. UTC+14 did not exist prior to 1995. Kiribati’s Line Islands, including Kiritimati, sat on the far side of the International Date Line from the rest of the country, creating a 23 to 24-hour gap between the eastern and western halves of a single nation and leaving only four days a week when government offices on both sides could conduct business simultaneously. On December 31, 1994, President Teburoro Tito’s government simply skipped a day in the eastern islands, moving the Line Islands from UTC-10 to UTC+14 and the Phoenix Islands from UTC-11 to UTC+13, bending the International Date Line eastward around the entire country. The move had a side effect the government leaned into: Kiritimati became the first inhabited place on Earth to greet January 1, 2000, and the island was renamed Millennium Island to mark the occasion.

Samoa made a mirror-image decision in 2011, crossing from UTC-11 to UTC+13 to align its business week with Australia and New Zealand rather than the United States, skipping December 30 entirely that year. Neither move required international approval. There is no treaty that fixes the date line’s path; it exists purely by the accumulated pattern of national decisions, which is why it zigzags sharply around Pacific island nations on any accurate map rather than running straight down the 180th meridian.

These jumps also show why offset calculation can never be reduced to a static lookup table checked once. The same city can carry a different correct answer depending on the exact date in question, which is the entire reason the tz database stores rule histories instead of single numbers, and why any system generating “current” local time for a list of world cities needs a live feed rather than a table someone typed out once and forgot to revisit.

What this means for anyone building or trusting a converter

A time zone converter that only stores whole-hour offsets will silently misreport India, Nepal, parts of Australia, and the Chatham Islands. One that hardcodes offsets without a version-controlled rule history will drift out of date the next time a government changes its daylight saving start date, something that happens somewhere in the world most years. And one that treats Etc/GMT+5 and UTC+5 as interchangeable will hand a user a result five hours off from what they expected, in the wrong direction.

The safest working assumption, for a developer or simply for a curious traveler checking a flight connection, is that an offset is not a property of a place on a map. It is the current output of a rule a government wrote down, subject to change without notice, filed under a city name specifically because city names outlast the political borders around them.