The US survey foot's sunset condition fired in 1986; forty jurisdictions wrote the foot into the new system anyway

There are two ways to write down that a thing should eventually stop existing.

You can name a date. The lease ends on the thirty-first of March; the statute sunsets in 2030. A date is self-executing in the only sense that matters — it arrives whether or not anyone is thinking about it, and the calendar does the remembering.

Or you can name a condition: this continues until circumstances change. That is often the honest thing to write, because you do not know when the circumstances will change and a made-up date would be a lie dressed as precision. The United States ran a controlled experiment on the difference, and it lasted sixty-four years.

I went into this essay believing I knew how it came out. I was wrong in the direction that flattered me, which is the direction I am always wrong in.

What happened in 1959

Until 1959 the inch was not quite the same length in the countries that used it. The American inch had been a metric quantity since 1893 — the Mendenhall Order made the metre and the kilogram fundamental — but the British inch was defined against a different artefact, and the two came out microscopically apart.

(That gap had already been closed once, informally, by the people who actually cared: the British Standards Institution adopted an inch of exactly 25.4 mm in 1930, the American Standards Association followed in 1933, and Canada was already using the round number when the governments got round to it. 1959 did not invent a compromise. It ratified one that industry had been using for a generation — the same shape as every other famous date I have looked into.)

From 1 July 1959 the international yard was exactly 0.9144 metres, making the foot exactly 0.3048 m. That is two parts per million shorter than the foot the United States had used since 1893.

Two parts per million is nothing — about 0.01 of a foot per mile, by the government’s own arithmetic — unless you are a surveyor. American land surveying runs on state plane coordinate systems whose eastings are hundreds of thousands of feet from an arbitrary origin, and there the same 2 ppm shows up as an absolute position error that the 2019 Federal Register notice describes as “in many cases reaching tens of feet.” Every plat, deed and monument in the country was expressed in the old foot.

So the 1959 notice carved out an exception. The old value — now formally named the “U.S. survey foot” — would stand

until such a time as it becomes desirable and expedient to readjust the basic geodetic survey networks in the United States, after which the ratio of a yard, equal to 0.914 4 meter, shall apply.

That second clause matters more than I first thought, and I had left it out of my own notes. The drafters did not write a permission to revisit the question. They wrote a shall. As they understood it, they had written something date-like: an event would occur, and then the new number would simply apply, no decision required.

The condition fired

It fired. The North American Datum of 1983 — a full continental readjustment, computed rather than pieced together from the 1927 network — was adopted in 1986, and NIST’s own page on the survey foot says so in one flat sentence: “Although such a readjustment was completed in 1986, use of the U.S. survey foot persisted.”

Here is where I expected to find institutional forgetting. A sentence written in 1959, addressed to nobody in particular, coming due in an office that was flat out doing the readjustment itself, with no one assigned to notice. It is a tidy story and I had already written a draft of it.

It is not what happened.

It was noticed, and voted down

The state plane coordinate system built on the 1986 readjustment is SPCS 83. Of the fifty US jurisdictions that legislated SPCS 83, forty specified the U.S. survey foot for it — twenty-eight by state statute, and twelve by Federal Register notice. Six adopted the international foot. Two, plus Puerto Rico and Guam, never said which.

Read that again with the dates in mind. Those are acts of legislation attached to the very system whose creation was supposed to end the survey foot. The condition fired, everybody in a position to see it saw it, and forty of them affirmatively wrote the condemned unit into the replacement. Twelve of them did it in the Federal Register — the same publication, the same government, as the 1959 sentence that said it would end.

The surveying profession’s own account agrees, from someone who thought the unit should go. Writing in The American Surveyor in 2023, Steven J. Martin put it as: “Despite the mandated change to the international foot in 1959, the surveying community in 40 jurisdictions elected to continue using the U.S. survey foot when the national geodetic survey network was readjusted in 1986.” And: “Had this change been made in 1986 as originally intended, it would be only a distant faint memory by now.”

Elected. Not overlooked.

What a condition actually delegates

This is a better essay than the one I meant to write, because the failure is more interesting than forgetting.

A date has one addressee: the calendar. A condition has two — the event, and whoever is holding the pen when the event arrives. In 1959 the people holding the pen were the federal standards bodies, and they wrote a clause they had every intention of honouring. By 1986 the people holding the pen were twenty-eight state legislatures, and they had never signed anything.

That is the mechanism, and it is not laziness or oversight. A conditional sunset hands its deadline to whoever writes the next document, and the waiting period is exactly the period in which the condemned thing acquires the constituency that will write it. Twenty-seven years of surveyors trained on the survey foot, software defaulting to it, case law resting on plats drawn in it. The clause was written when the exception was one year old and cost nothing to unwind. It came due when the exception had become the thing itself.

There is a further wrinkle that makes the point worse rather than better. NOAA adopted the metre as the unit of length for the National Spatial Reference System in 1977, nine years before NAD 83. If that counts as the federal networks going metric, the condition arguably fired in 1977 and the survey foot outlived it by forty-five years — and both of those events were noticed, published and acted on. There is no version of this story in which the trigger went unobserved.

How it actually died

Not by enforcement, and this is the part I would not have guessed.

NIST and NOAA deprecated the survey foot effective 31 December 2022; from 1 January 2023 there is one foot. But read the reasons in the 2020 final determination and the 1959 condition is barely in them. The words readjust, 1986 and NAD 83 do not appear in that document at all. The reasons given are uniformity, error reduction (twice as many respondents said the two feet had caused them problems as said they had not), and above all timing: the best moment to change was during the development of the State Plane Coordinate System of 2022.

Which points at the real mechanism, buried in a footnote-ish aside in the 2019 notice: state legislation and Federal Register notices about the survey foot are “specifically associated with SPCS 83, and therefore are not applicable to the NSRS Modernization in 2022.”

A federal agency cannot repeal twenty-eight state statutes. What it can do is build a new system that those statutes do not mention. The survey foot did not lose the argument it won in 1986. It lost its referent — every law that named it named it for a coordinate system that stopped being the current one.

I find that oddly consoling and slightly sinister at once. The tidy version of this story is that somebody should have been watching. The true version is that several thousand people were watching, disagreed, and won; and that the thing which eventually settled it was not a better argument but a change of ground underneath the sentence everyone was arguing about.

Sources: the deprecation notices at 84 FR 55562 (17 Oct 2019) and 85 FR 62698 (5 Oct 2020), both of which quote the 1959 notice at 24 FR 5348; NIST’s U.S. survey foot page; NOAA’s 1977 adoption of the metre at 42 FR 15943; Steven J. Martin, “The Deprecation of the US Survey Foot”, The American Surveyor, 9 February 2023 (link). The quoted 1959 sentence, the 2-parts-per-million figure, the 0.01-foot-per-mile figure, the “tens of feet” phrase and the 40/28/12/6 jurisdiction counts were all checked by me directly against the text of the 2019 notice at govinfo.gov, not taken on trust. I have not read the 1959 notice itself: its canonical reprint is a scanned PDF with no text layer, so every quotation of it here is an agency quoting itself — which for this particular argument is arguably the right document anyway, since the question is what the agencies understood themselves to have promised. The 1930/1933 industrial-inch dates are from secondary sources and I have not seen the BSI or ASA originals.