# Geography methodology — public-source ABOTA atlas

## Result and proper description

**This is a 50-state Census-based locality index, not a claim to enumerate every U.S. local government or every inhabited community.** The verified index contains **28,067 geographic records** with unique official, layer-prefixed identifiers. Counts are geographic entities, **not distinct municipal governments**, and the file name `cities.json` is an interface convenience rather than a legal classification.

The computed components are **19,505 legal Census places**, **8 consolidated-city balance geographies**, **8,391 governmental town/township/plantation county subdivisions in 12 states**, and **163 Hawaii CDPs**. The inputs are the official 2026 national places and county-subdivisions Gazetteer files.[7][8]

Use this description in the atlas: **“2026 Census legal places, supplemented by governmental towns and townships in 12 states; Hawaii shown with Census-designated places.”** Do not label all records “incorporated cities,” imply each has an independent government, or treat a geographic record as evidence that any ABOTA member practices there.

## Vintage and provenance

The latest vintage located and downloaded was **2026**, not the 2025 fallback; the official 2026 page identifies the corresponding geography as matching 2026 American Community Survey publication areas.[1] The official directory lists the place, county-subdivision, state and county archives with September 8, 2026 modification dates.[3] Retrieval timestamps, response `Last-Modified` values, byte sizes and SHA-256 hashes are preserved in `research/raw/geography/manifest.json`.

The four compressed national sources and extracted UTF-8 pipe-delimited text files are retained unchanged in `research/raw/geography/`. The official layout documents identifiers, names, functional status, and `INTPTLAT`/`INTPTLONG` as the latitude and longitude of the internal point.[2] State names and two-letter postal abbreviations come from the state Gazetteer; county/county-equivalent labels for subdivision disambiguation come from the county Gazetteer.[9][10]

The citation ledger is task-specific: `research/geography-ledger.json`. It contains exact official provenance URLs and evidence quotes. Supporting source text and the two downloaded Census reference PDFs are also retained locally. The historical GARM chapter is used only for the Hawaii/incorporation distinction, not for old counts, incorporation thresholds, or old municipality lists.[6]

## Inclusion rules

### 1. Legal Census places, all 50 states

Include every place in the downloaded 2026 file in the 50-state scope that is a legal place rather than a CDP/statistical entity. The raw Census statuses actually represented among these records are A, B, I and N; inactive and nonfunctioning legal places are retained, not silently discarded.[5][7] Verification compares the complete output place-ID set against all raw 50-state place records whose functional status is A/B/C/I/N. This yields **19,505 legal places**, including 32 inactive (I) and 3 nonfunctioning (N) records; the `kind` field flags these exceptions.

This includes Census-designated cities, towns, villages, boroughs, municipalities, city-and-boroughs and less common incorporated/consolidated place designations.[4][7] It does not assume that a place called a “town” is legally an incorporated place in every state. Census treats towns in New England, New York and Wisconsin as minor civil divisions (MCDs), not incorporated places; Alaska boroughs are county equivalents, and the five New York City boroughs are MCDs.[4]

### 2. Eight consolidated-city balances

Retain the eight non-CDP `FUNCSTAT=F` place rows whose official name ends in `(balance)`, each explicitly tagged `consolidated_city_balance`. These are not silently relabeled incorporated municipalities. Census distinguishes the consolidated city from its balance, which excludes separately functioning incorporated places inside the consolidated government.[4] `F` means a fictitious entity created for Census geographic hierarchy purposes, rather than an independent functioning government.[5]

The output includes the actual Gazetteer balance points for Milford, Athens-Clarke County, Augusta-Richmond County, Indianapolis, Greeley County, Louisville/Jefferson County, Butte-Silver Bow, and Nashville-Davidson. It does **not** substitute a county centroid or claim the balance boundary equals the entire consolidated government.[7]

### 3. Governmental MCD supplement in 12 states

Include records from **CT, ME, MA, NH, RI, VT, NY, WI**, and also **MI, MN, NJ, PA**, where the Census source identifies these states as the 12 in which most MCDs serve as general-purpose local governments.[4] Adding the latter four avoids silently losing New Jersey/Pennsylvania municipalities named townships and major Michigan/Minnesota township governments.

Within those states the exact implemented test is: `FUNCSTAT` in `A, B, C, I`, and official name ending in ` town` or ` township`; in Maine also include names ending in ` plantation`. Thus charter townships are included. The pipeline excludes statistical CCDs/unorganized territories, fictitious place-surrogate MCDs, undefined county subdivisions, subordinate NYC boroughs, reservations, grants/gores/locations, and nonfunctioning `N` subdivisions. Every input record, whether included or excluded, has an auditable decision in `research/geography-record-audit.json`.

The result contains **8,391 MCD records**, including 7 inactive units marked `_inactive`, 23 consolidated units marked `_consolidated`, and 2 partially consolidated units marked `_partially_consolidated`. Functional status C represents a government consolidated with another government with one set of officials; B represents partial consolidation with separate officials.[5] The gazetteer records are retained as distinct legal geographies even when they overlap a separately indexed place. No arbitrary name-based deduplication is performed.

**Consequences:** Connecticut's 169 town-geography records coexist with its separate place/balance records; its index count is not the number of distinct governments. Towns and villages with the same short name, nested town/village jurisdictions, and identically named towns in different counties remain separate entries. Official names and county labels make the distinctions visible.

### 4. Hawaii: explicit CDP exception

The Census Bureau does not recognize incorporated places in Hawaii for this place-geography classification and instead recognizes CDPs; this does not mean Hawaii lacks local government or that the City and County of Honolulu lacks municipal functions.[6] The 2026 raw file contains **163 Hawaii place records**, all CDPs; all are retained with `kind=cdp_hawaii_exception`.[7] They are geographic reference localities, **not incorporated cities**, and must remain visibly labeled as such. The Honolulu-area CDP point does not describe the entire City and County of Honolulu government.

All non-Hawaii CDPs are excluded. No community is added merely because it has a ZIP code, postal city label, neighborhood name, or familiar unincorporated settlement name.

## Coordinates, names, identifiers and schema

`data/cities.json` is a JSON **array**, with exactly these fields on every object:

```text
id, state, state_name, name, display_name, kind, latitude, longitude, source_url
```

- `id`: `place:` plus the unaltered seven-character official GEOID, or `subdivision:` plus the unaltered ten-character official GEOID. Leading zeroes are retained. Place GEOIDs combine state and place FIPS; subdivision GEOIDs combine state, county, and subdivision FIPS.[2]
- `state`, `state_name`: official two-letter abbreviation and name; only the 50 states are included. DC, Puerto Rico and other territories are outside this requested scope.
- `name`: human-oriented short label formed only by removing recognized terminal legal-type text and `(balance)`; no guessed city aliases or spelling corrections. Some exceptional names remain unshortened.
- `display_name`: the **exact official source NAME**, followed by state, or by county/county-equivalent and state for subdivisions. This preserves legal type, balance wording, accents and other source characters. Use this field to resolve ambiguous short names.
- `kind`: derived legal/geographic type; suffixes `_inactive`, `_nonfunctioning`, `_consolidated`, and `_partially_consolidated` communicate exceptional functional status. Hawaii CDPs and consolidated balances have dedicated kinds. Raw `FUNCSTAT`, `LSAD` (where present), ANSI code and official name are preserved in the record audit.
- `latitude`, `longitude`: numeric conversions of the exact official internal-point values; every output coordinate was compared back to its raw source row. These are representative geographic internal points, **not lawyer office coordinates, geocoded street addresses, surveyed civic centers, or necessarily geometric centroids**.[2]
- `source_url`: exact downloadable national ZIP URL for the source layer. The row is located by the official GEOID after removing the layer prefix.

`data/states.json` is a 50-object JSON array containing exactly `code` and `name`.

GEOIDs are stable official keys for this source vintage, not a promise of permanent cross-year identity: Census warns of changes in boundaries, names, creation and dissolution of entities between vintages.[1] This dataset is point-only; no polygon boundary or office-to-jurisdiction spatial containment is provided.

## Known geographic boundaries and limitations

1. **Not every civil-town government nationally.** Governmental MCDs outside the 12 supplemented states are omitted, including those in Illinois, Indiana, Kansas, Missouri, Nebraska, North Dakota, Ohio and South Dakota. Census notes that MCDs in those eight states usually have a lesser governmental role but can still be active governments.[4] Therefore do not claim “every U.S. municipality” without this qualification.
2. **Not every populated place.** Non-Hawaii CDPs and unincorporated communities are deliberately excluded; the file cannot stand in for every postal city or all locations where attorneys may have offices. It is not a residential-address coverage map.
3. **Not unique governments or nonoverlapping territory.** Consolidated place/town legal records and nested jurisdictions can overlap. No population totals should be summed over the index. Shared or nearby representative points do not prove duplicate entities.
4. **Not all county-level governments.** Alaska county boroughs and other county governments are not added merely because they exist; qualifying place records such as Anchorage municipality and Juneau city and borough are retained from the place source.[4][7]
5. **Not full consolidated-city polygons.** Eight balance records are explicitly retained as balance geographies, not silently promoted to full consolidated governments.[4][7]
6. **Not a live state incorporation register.** Coverage is the 2026 Census vintage as downloaded, without independent reconciliation to every state's subsequent incorporations/dissolutions. Some Census labels differ from everyday terminology. No current legal status is inferred beyond the source.
7. **No lawyer research or membership conclusions.** This work establishes a locality denominator only. Presence in the index establishes nothing about ABOTA membership, plaintiff/defense alignment, office presence, or whether five qualifying attorneys exist for a side.

## Verification and reproduction

Parser, indexing, validation and export tests were each added before implementing their missing behavior and observed to fail first; the final suite, including the separate postal reference-point export, passes five tests. One initial test assumed Amherst was a Massachusetts Census “town”; raw-source inspection showed `Amherst Town city`, so the fixture was corrected to the verified `Lexington town` without altering official geography to fit an assumption.

The executed verification confirmed:

- 28,067 output records and 28,067 unique prefixed GEOIDs;
- all 50 requested states present and no DC/PR records;
- complete legal-place set reconciliation against raw functional-status records;
- complete MCD-selection set reconciliation: 8,391 IDs matched;
- every coordinate finite, in range and exactly equal numerically to its source value;
- no non-Hawaii CDP included;
- per-state component sums and kind totals reconcile to the full index;
- all four raw ZIP SHA-256 checksums, ZIP CRC checks, and extracted text equality verified;
- rejection tests exercise malformed GEOIDs, NaN coordinates, changed source coordinates, duplicate IDs and a missing state.

Run from the project root (stdlib Python 3.11+; no third-party packages needed):

```sh
python3 -m unittest discover -s tools -p 'test_geography*.py' -v
python3 tools/geography_build.py
```

The build consumes the preserved raw files; it does not re-download or silently update vintage. It writes only geography-owned output files. Test output is saved in `research/geography-test-results.txt`; validation results in `research/geography-verification.json`; component totals in `research/geography-counts.json`; detailed source decisions in `research/geography-record-audit.json`. To refresh source data, retrieve the exact archive URLs in the manifest, preserve a new dated snapshot, and review schema and inclusion decisions before rebuilding.

## Independent postal reference-point supplement

The separate `data/postal_centroids.json` object adds 33,791 official ZCTA internal points from the national 2026 ZCTA Gazetteer, whose ZCTA geography uses 2020 Census blocks.[1][13] It does not change this city/locality index, infer municipal aliases, or verify a lawyer's municipality. Consumer maps must label it as a **postal-area reference point with municipal match unverified**. Not every valid ZIP has a corresponding ZCTA, including possible PO-box/unique-ZIP gaps; absent codes remain unmapped.[14] See `research/geography-postal-methodology.md` for the full coverage and interpretation rules. Rebuild that separate artifact with `python3 tools/geography_build.py --postal-only`.

## Per-state counts

Computed from the saved index; “legal places” includes inactive/nonfunctioning legal place records. “MCD supplement” may overlap place governments. All figures are index records, not a census of distinct municipal governments.

| State | Name | Legal places | Balances | MCD supplement | Hawaii CDPs | Total |
|---|---|---:|---:|---:|---:|---:|
| AK | Alaska | 149 | 0 | 0 | 0 | 149 |
| AL | Alabama | 465 | 0 | 0 | 0 | 465 |
| AR | Arkansas | 501 | 0 | 0 | 0 | 501 |
| AZ | Arizona | 92 | 0 | 0 | 0 | 92 |
| CA | California | 483 | 0 | 0 | 0 | 483 |
| CO | Colorado | 273 | 0 | 0 | 0 | 273 |
| CT | Connecticut | 29 | 1 | 169 | 0 | 199 |
| DE | Delaware | 57 | 0 | 0 | 0 | 57 |
| FL | Florida | 411 | 0 | 0 | 0 | 411 |
| GA | Georgia | 536 | 2 | 0 | 0 | 538 |
| HI | Hawaii | 0 | 0 | 0 | 163 | 163 |
| IA | Iowa | 939 | 0 | 0 | 0 | 939 |
| ID | Idaho | 199 | 0 | 0 | 0 | 199 |
| IL | Illinois | 1,294 | 0 | 0 | 0 | 1,294 |
| IN | Indiana | 565 | 1 | 0 | 0 | 566 |
| KS | Kansas | 625 | 1 | 0 | 0 | 626 |
| KY | Kentucky | 414 | 1 | 0 | 0 | 415 |
| LA | Louisiana | 305 | 0 | 0 | 0 | 305 |
| MA | Massachusetts | 58 | 0 | 293 | 0 | 351 |
| MD | Maryland | 157 | 0 | 0 | 0 | 157 |
| ME | Maine | 23 | 0 | 459 | 0 | 482 |
| MI | Michigan | 533 | 0 | 1,240 | 0 | 1,773 |
| MN | Minnesota | 856 | 0 | 1,776 | 0 | 2,632 |
| MO | Missouri | 947 | 0 | 0 | 0 | 947 |
| MS | Mississippi | 300 | 0 | 0 | 0 | 300 |
| MT | Montana | 126 | 1 | 0 | 0 | 127 |
| NC | North Carolina | 552 | 0 | 0 | 0 | 552 |
| ND | North Dakota | 355 | 0 | 0 | 0 | 355 |
| NE | Nebraska | 528 | 0 | 0 | 0 | 528 |
| NH | New Hampshire | 13 | 0 | 222 | 0 | 235 |
| NJ | New Jersey | 324 | 0 | 240 | 0 | 564 |
| NM | New Mexico | 105 | 0 | 0 | 0 | 105 |
| NV | Nevada | 19 | 0 | 0 | 0 | 19 |
| NY | New York | 594 | 0 | 933 | 0 | 1,527 |
| OH | Ohio | 923 | 0 | 0 | 0 | 923 |
| OK | Oklahoma | 600 | 0 | 0 | 0 | 600 |
| OR | Oregon | 241 | 0 | 0 | 0 | 241 |
| PA | Pennsylvania | 1,012 | 0 | 1,544 | 0 | 2,556 |
| RI | Rhode Island | 8 | 0 | 31 | 0 | 39 |
| SC | South Carolina | 271 | 0 | 0 | 0 | 271 |
| SD | South Dakota | 310 | 0 | 0 | 0 | 310 |
| TN | Tennessee | 344 | 1 | 0 | 0 | 345 |
| TX | Texas | 1,230 | 0 | 0 | 0 | 1,230 |
| UT | Utah | 255 | 0 | 0 | 0 | 255 |
| VA | Virginia | 227 | 0 | 0 | 0 | 227 |
| VT | Vermont | 40 | 0 | 242 | 0 | 282 |
| WA | Washington | 281 | 0 | 0 | 0 | 281 |
| WI | Wisconsin | 608 | 0 | 1,242 | 0 | 1,850 |
| WV | West Virginia | 229 | 0 | 0 | 0 | 229 |
| WY | Wyoming | 99 | 0 | 0 | 0 | 99 |
| **All 50** | | **19,505** | **8** | **8,391** | **163** | **28,067** |

## Sources

[1] https://www.census.gov/geographies/reference-files/time-series/geo/gazetteer-files.2026.html
[2] https://www.census.gov/programs-surveys/geography/technical-documentation/records-layout/gaz-record-layouts.2026.html
[3] https://www2.census.gov/geo/docs/maps-data/data/gazetteer/2026_Gazetteer
[4] https://www.census.gov/programs-surveys/popest/about/glossary/geo-terms.html
[5] https://www2.census.gov/geo/pdfs/reference/FunctionalStatusCodes.pdf
[6] https://www2.census.gov/geo/pdfs/reference/GARM/Ch9GARM.pdf
[7] https://www2.census.gov/geo/docs/maps-data/data/gazetteer/2026_Gazetteer/2026_Gaz_place_national.zip
[8] https://www2.census.gov/geo/docs/maps-data/data/gazetteer/2026_Gazetteer/2026_Gaz_cousubs_national.zip
[9] https://www2.census.gov/geo/docs/maps-data/data/gazetteer/2026_Gazetteer/2026_Gaz_state_national.zip
[10] https://www2.census.gov/geo/docs/maps-data/data/gazetteer/2026_Gazetteer/2026_Gaz_counties_national.zip
[13] https://www2.census.gov/geo/docs/maps-data/data/gazetteer/2026_Gazetteer/2026_Gaz_zcta_national.zip
[14] https://www.census.gov/programs-surveys/geography/guidance/geo-areas/zctas.html