Connecticut Retired Eight Counties That Had Been Dead Since 1960

Connecticut's counties stopped functioning as governments in 1960. Nothing was administered from them. Nobody was elected to run one. No budget passed through them. They persisted anyway — on maps, in federal tables, in election files — for the same reason a phantom limb persists: everything built around them assumed they were still there.

The federal statistical system needs a county-shaped row for every place in America. Population, income, unemployment, votes: the county is the default spine of American data, and a state without working counties still has to fill the same number of lines in the same table as a state that has them. So eight dead Connecticut governments were kept on the books, assigned Census codes, and issued numbers every year for six decades — until a Federal Register notice dated 2022-06-06 replaced them with nine planning regions.

That swap never made the news, and it quietly corrupts a great many 2020-to-2024 comparisons built on top of it. The nine new regions do not nest inside the eight old counties. There is no parent-child relationship, no clean split, no set of population weights that makes the transition lossless. Any county-to-region crosswalk is wrong — and not wrong by a decimal place. Wrong by whole towns.

Eight out, nine in

The exchange is not a renaming and not a redistricting. It is a full replacement of one set of statistical units with another set that was drawn for a different purpose, on different lines, in a different century.

The old eight carry two Census attributes worth knowing about. Their class code is CLASSFP "H4," and their functional status is FUNCSTAT "N" — non-functioning. The federal government was not confused about what it was publishing. It labeled those counties as dead in the same file that reported their population. The flag was there the whole time; almost nobody reads it.

The regions do not nest, and Shelton proves it

Two facts kill the crosswalk outright. Shelton, an old industrial city on the Housatonic, sat in Fairfield County under the old vintage and sits in the Naugatuck Valley Planning Region under the new one — it moved across the boundary entirely, not partially. And Litchfield County, the rural northwest corner, has no single successor at all. There is no region you can point to and call it "Litchfield, renamed."

A crosswalk that converts Fairfield County into a share of Naugatuck Valley is not making a rounding error. It is making a guess about where Shelton lives.

This matters because the standard workaround — apportion the old unit into the new ones by population share — assumes the pieces are pieces. When a whole municipality relocates, share-based apportionment doesn't approximate the truth; it manufactures a plausible-looking number with a real city smeared across two rows. You would never see the seam in the output. That is exactly the problem.

The fix is to stop using counties at all

Connecticut has 169 towns, and the towns did not change. They are the same 169 before the switch and after it. They were the real units of government the whole time — the ones that run the schools, the police, the polling places, and the vote count.

So Politigoat does not crosswalk Connecticut. It also does not invent the new vintage. What we publish is the eight retired counties, carried in MEDSL's county file across all seven cycles and labelled as the retired units they are. The nine planning regions have no presidential series here at all — their pages say exactly that, rather than showing a number apportioned out of Fairfield County. Every Connecticut figure on this site belongs to the old vintage, and none of it is estimated.

Doing it properly means town-level results for every cycle — the same 169 towns summed into whichever vintage you asked for, with no interpolation anywhere and therefore nowhere for a silent error to hide. That is the build this article is arguing for. It is not the build we have: it needs a town-level source we have not yet mirrored. Until we do, the honest position is that our Connecticut coverage ends with the vintage the Census retired in 2022, and that the nine regions are a gap rather than a series.

The build flagged it before a human did

Our geography pipeline compares Census vintages and refuses to proceed silently when unit sets disagree. Moving from the 2020 to the 2024 vintage, it reported 8 units retired and 9 units adopted in a single state. That is not the signature of a boundary adjustment or a renaming — a rename shows up as one out, one in. Eight out and nine in is a replacement, and it stopped the build.

Nationally, the same event registers as almost nothing: the county-equivalent count went from 3,143 to 3,144. A net change of one. If you monitor only the national total, an entire state's geography can be swapped underneath you and the headline number moves by 0.03 percent. This is the ordinary way data errors enter serious work — not through a dramatic failure, but through a change that is too small at the top to notice and too large at the bottom to ignore.

The same flag flies in two more states

Connecticut is not unique in having counties that do nothing. All five Rhode Island counties carry the same FUNCSTAT "N" non-functioning flag, and so do nine of the fourteen Massachusetts county units we carry. The other five Massachusetts counties still have working county governments; the nine do not. All of them, working or not, still appear in county files, still get county rows, still get colored in on election-night maps.

Whether either state ever follows Connecticut's path is not something we will predict, and nothing in the current data implies a schedule. What we will say is that we treat a non-functioning county as a container, not an actor. It is a shape that holds municipalities. Describing a shift in "Bristol County" as though a county did something is a category error when the county has no government to do anything with.

The larger lesson is not about New England. It is that the shape of a dataset is an argument about how the country is organized, and that argument can be six decades out of date while every file still validates cleanly. The federal system kept eight Connecticut counties alive because it needed a row, not because there was anything to put in it. Our job is to notice when the row is empty.