People by country list: a researched guide for 2027

There is no list of the countries. There are several lists, maintained by different bodies for different purposes, and they disagree with each other on purpose.

That is not a flaw anyone is failing to fix. A list built to allocate votes in an assembly has no reason to match a list built to code addresses for shipping. The question "is this a country" has no answer until you say which list you are asking about, and a dataset that never names its list has a defect in its first column.

What to take away

• Every country list encodes the purpose it was built for. Name the list and its version before using any column that depends on it.
• Current lists hold current entities. Historical data needs a separate register of former entities with start and end dates.
• Codes are not names and names are not stable. Join on codes, keep the name as text, and record which naming convention the text follows.

Lists built for different jobs

Documentary heritage programs keep their own registers on the same principle, as the nomination criteria for Memory of the World show: what is on a list is a product of who compiled it and why. The families of list you will actually meet:

Kind of list	Built to answer	Typical inclusion question
Membership of an international organization	Who has standing in this body	Has this entity been admitted
Standards code lists	How to write a place in a data field	Does this entity need a distinct code for exchange
Statistical groupings	How to compile and compare figures	Which unit does the compiling agency report on
Sporting and cultural federations	Who may enter under which flag	Which federation is recognized by the governing body
Postal, telecom and currency systems	How to route or denominate something	Does this territory operate its own service

None of these is the master list. Each is authoritative for its own purpose and wrong for the others. A statistical grouping is not a statement about sovereignty; a code list assigns entries to territories that no one considers independent, because the data still has to go somewhere.

Dependencies, territories and disputed entities

Three cases split the lists apart.

Dependencies and overseas territories. Some lists give them their own entry, some fold them into the governing state, and some do both in different tables. The same population therefore appears once, twice or not at all depending on which list a dataset used.

Constituent parts. Where a state is a union of named nations or regions, some purposes list the parts and some list the whole. Both are correct for their purpose, and mixing them double counts.

Contested and partially recognized entities. Lists differ here because the bodies behind them differ, and the difference is a fact about the list maker rather than a settled matter. A dataset that silently adopts one list has adopted its position too, which is worth knowing even when it is the only practical choice.

Codes outlive names, and names outlive codes

Code systems typically offer a two-letter form, a three-letter form and a numeric form. They are not interchangeable in behavior.

Alphabetic codes are mnemonic and tied to a name, so a rename creates pressure to change the code. Numeric codes carry no meaning and survive renames intact, which is why long time series tend to be keyed on them. Retired codes are held in reserve for a period before reuse, which means an old file and a new file can both contain the same code meaning different things.

The names are worse. For one entity you may find a short form, a formal or long form, a common English usage, the name in the country's own language, and a historical name still used in older sources. A name column that mixes these will fail to join, and the failures will be concentrated in exactly the entries whose names are contested.

Store the code as the key, the code system and its version as their own columns, and the display name as text that nothing joins on. The same discipline of keeping raw and normalized values apart is set out on the city data page.

Historical entities need their own register

Current lists are current by design. They contain no entry for a state that dissolved, merged, split or was absorbed, which means historical records cannot be coded against them without loss.

A workable register of former entities carries, for each one, a start date, an end date, the successor or successors, and a note on whether territory moved without the state changing. Published code lists keep something similar: the change history behind ISO 3166-1 records retirements and reassignments rather than deleting them. That last case is common and it is the one that breaks naive mappings: a place can change country without either country ceasing to exist.

With such a register, a record can be coded twice, once to the entity named on the document and once to the modern entity covering that ground, with both stored and neither presented as the truth. Without it, one of those codings is being done implicitly, and nobody downstream knows which. Dating conventions and the risk of false precision in those start and end dates follow the rules on the century page.

What a country column should carry

At minimum, alongside the value itself:

• The code system and version used.
• Whether the value describes the entity at the time of the record or the entity now.
• How dependencies were handled.
• How historical entities were handled.
• What a blank means: not applicable, not established, or not collected.

Five extra columns sounds heavy until the first time someone asks why two files that both say "country" will not join.

Why there is no list of people by country here

The obvious page to build on top of a country list is a list of people assigned to each entry. We do not publish one.

Assigning a person to a country requires a value in a field that, as the country pillar sets out, answers several different questions at once. Nationality, citizenship, birthplace, residence and family origin are separate, they change during a life, and the document that records one says nothing about the others. A list flattens all of that into a heading.

For living people the rule is absolute: we do not state anyone's nationality, ancestry, religion or birthplace. Where someone has described their own background publicly, we quote them and date it. And we publish no ranked tally of people by origin at all, for reasons that hold whatever the underlying list says. Verifying any historical attribution of this kind is covered under historical records.

Bottom line

A country list is an instrument with a purpose stamped on it. Name the instrument, keep the codes and the names apart, give historical entities their own register, and never let a list of places quietly become a list of people.

Common questions

Which list should a project adopt?

The one whose purpose is closest to the project's, chosen deliberately and documented. For data interchange that usually means a standards code list; for comparing published figures it means the list the publishing agency used, because anything else will not reconcile.

Two sources use the same country name for different territory. What now?

Treat the name as ambiguous and resolve it through the code plus the date. Where neither is available, record both readings and mark the field unresolved rather than picking the more familiar one.