Connecticut
Connecticut abolished county government in 1960, and in 2022 the Census went further and retired the counties as a geography, replacing all eight with nine Planning Regions. The state's own voter file did not get the memo: it still sorts all 169 towns under the counties that no longer exist in either sense.
What happened to Connecticut's counties?
State deltas
Every state has hair. Here is Connecticut's.
The rules that make this state different from the last one, what each breaks if you assume the national default, and how the engine resolves it.
Connecticut ended county government in 1960, so the Census counts zero county governments — one of only two states, with Rhode Island. In 2022 the Bureau also retired the counties as county-equivalents and replaced them with nine Councils of Governments, so the 2024 gazetteer lists Planning Regions instead. Any join keyed on a Connecticut county FIPS code against post-2022 census geography finds nothing, and any pipeline that assumes a county tier exists between the state and the town has no rows to put in it.
The county-equivalent tier is optional and versioned by the vintage of the geography it came from, so a state can have none and a jurisdiction key can change meaning between census releases.
The Secretary of the State's October 17, 2025 registration file gives every town a County — Fairfield, Hartford, Litchfield, Middlesex, New Haven, New London, Tolland, Windham — while the Census government-units file for the same state already uses the nine Planning Regions. Both are current and official. They do not map one-to-one, so a consumer that reads the county from the voter file and joins it to census geography is joining a 1960 administrative fossil to a 2022 planning boundary, and the failure is silent because both sides have plausible names.
Geographic labels are stored with the artifact and vintage they came from, and no label is treated as a join key across artifacts without an explicit crosswalk.
1,057,561 of 2,503,192 registrations are Unaffiliated — 42.2%, more than the Democrats at 35.1% and nearly double the Republicans at 21.1%. It is a state of the record, not an organisation, and it is the largest single value in the enrollment column. Software that ranks parties by registration puts a non-party first; software that filters to parties drops 42% of the electorate; and software that treats the largest bloc as a partisan universe is modelling people who declined to join one.
Enrollment is stored as the state's own value with a flag for whether it denotes a party, so the largest bloc can be counted without being ranked among parties.
The published minor-party key carries I (Independence), INDA (Independance) and IT (Independent) as three separate codes. The middle one is a misspelling that has become an official key and persists in the current file. Name-based matching merges at least two of the three; fuzzy matching merges all three; and any of those merges silently reassigns real registrations between organisations that are legally distinct.
Parties are keyed on the state's own code rather than on the display name, and the name is carried as a label that is never used for matching.
Glastonbury spans districts 1 and 2, Middletown 1 and 3, Shelton 3 and 4, Torrington 1 and 5, Waterbury 3 and 5. The state's own file records the congressional district as a list rather than a value for those rows, which is why a parser expecting a single integer drops exactly those five towns and produces a statewide total short by 176,195 registrations. A town-to-district relation modelled as many-to-one is wrong for five of 169, and they include the state's fourth-largest town.
Town-to-district is a many-to-many relation carrying its own rows, and no parser assumes a single value in a field the source publishes as a list.
Connecticut has ten boroughs — Bantam, Danielson, Fenwick, Groton Long Point, Jewett City, Litchfield, Naugatuck, Newtown, Stonington, Woodmont — and every one of them sits inside a town. In Alaska the identical word names the county-equivalent tier, nineteen of them, each containing municipalities. The same token denotes opposite ends of the ladder in two states, and a tier table keyed on the word alone will file Connecticut's sub-municipal boroughs as counties.
Tier labels are resolved per state rather than globally, so a word can occupy a different rank in different jurisdictions without being renamed.
The 2026 calendar sets a Town Committee Primary on March 3, five months before the August 11 state, district and municipal primary and eight months before the November 3 election. It elects party officers rather than public officials, runs on its own absentee and canvass rules, and is a statewide primary day that most models of an election year do not have a slot for. Treating a state as having one primary per cycle misses it entirely.
Primaries are stored per contest series with their own calendar, so a state can run several in a year for different classes of office.
January 6 for the 25th Assembly District, January 13 for the 139th, February 3 for a Fairfield selectman, February 17 for New Haven's 7th Ward, June 23 for a Middlebury selectman, September 15 for New Haven's 9th Ward. Each is a separate document with its own deadlines, and none of them appears in the November 3 state election calendar. A harvester that treats the state calendar as the calendar misses six elections and every deadline attached to them.
The cycle's calendar is assembled from every calendar the state publishes, and a special election carries its own deadline set rather than inheriting the general's.
Federal to borough
What Connecticut actually elects.
Five House seats, a legislature, and then nothing until the town — because there is no county tier left to put anything in.
Congressional map
Where those seats are.
The 5 congressional districts in force for the 2026 cycle — not necessarily the map this delegation was elected under. Connecticut has not redrawn since the 119th Congress was seated.
| District | Population | Median income |
|---|---|---|
| CT-01 | 716,454 | $83,329 |
| CT-02 | 730,400 | $94,194 |
| CT-03 | 726,058 | $81,637 |
| CT-04 | 721,814 | $117,299 |
| CT-05 | 722,450 | $89,288 |
Town ledger
One hundred sixty-nine towns, sorted by a county that has not governed anything since 1960.
Registration below is the Secretary of the State's Registration and Party Enrollment Statistics as of October 17, 2025 — the current published register. All 169 town rows reconcile three ways: the four enrollment columns, the active-plus-inactive columns, and the state's own totals row all sum to 2,503,192. The second column is the Unaffiliated share, which is the largest bloc statewide at 42.2% and ranges from 32.1% to 51.7% by town. Thirteen of 169 shown.
| Town | Feed vehicle | Registered | Unaffiliated share | Status | Last read |
|---|---|---|---|---|---|
| Stamford · largest | PDF / SOVC | 83,512 | 37.4% | Verified | 1 hr |
| Bridgeport | PDF / SOVC | 76,421 | 33.5% | Verified | 1 hr |
| New Haven · lowest unaffiliated | PDF / SOVC | 69,852 | 32.1% | Verified | 1 hr |
| Hartford | PDF / SOVC | 64,167 | 35.9% | Verified | 1 hr |
| Norwalk | PDF / SOVC | 62,787 | 42.3% | Verified | 1 hr |
| Waterbury · split across CD 3 and 5 | PDF / SOVC | 60,757 | 44.5% | Verified | 1 hr |
| Meriden | PDF / SOVC | 37,020 | 49.4% | Verified | 1 hr |
| Glastonbury · split across CD 1 and 2 | PDF / SOVC | 27,309 | 43.1% | Verified | 1 hr |
| Killingly | PDF / SOVC | 12,219 | 50.4% | Verified | 1 hr |
| Sterling · highest unaffiliated | PDF / SOVC | 2,600 | 51.7% | Verified | 1 hr |
| Colebrook | PDF / SOVC | 1,180 | 42.1% | Verified | 1 hr |
| Canaan | PDF / SOVC | 789 | 39.4% | Verified | 1 hr |
| Union · smallest | PDF / SOVC | 652 | 40.0% | Verified | 1 hr |