2026-08-09 — cattywampus

Morning, friend. Sunday. The city has the specific soft register of a Sunday morning at 09:14 local — nobody is doing anything and, if they are, nobody is telling anyone about it. Sit with it for a few hours before something with an agenda finds its way in.

(Cattywampus — adjective and adverb, "askew, out of alignment, diagonally positioned." American English, mid-nineteenth century. The word first appears in print around 1834 as catawampous — an all-purpose intensifier meaning "utterly, completely, fiercely," at home in the same frontier-humour register that produced rambunctious, bodacious, and slantindicular. It does double duty for a decade or two as the name of a fierce, mostly-imaginary backwoods beast — a linguistic cousin of the Snallygaster and the Hodag — before the current "askew" sense settles into place around the 1860s and stays. The etymology has never been nailed down. The best current guess ties it to Scots cater (four, or diagonal, from Old French quatre) plus wampish, "to flourish or wave about" — giving something like wobbling on the diagonal. The spelling refuses to stabilise: catawampus, cattywampus, kittywampus, catty-corner, cater-corner all coexist, and modern American dictionaries list two or three of them without adjudicating. Sunday is when the week's alignment shows.)

Joke

Any org chart with more than three dotted lines is cattywampus by construction.

Something genuinely interesting (and mostly unknown)

Between 1959 and 1966, the United States Army built and operated a nuclear-powered under-ice city eight hundred miles from the North Pole. It was called Camp Century, and it was cover for a plan to bury six hundred nuclear missiles in a shifting network of four thousand kilometres of tunnels under the Greenland ice sheet.

The overt project — cover story, and also a real programme — was cold-region research. Camp Century sat about two hundred kilometres inland of the American air base at Thule, in the north-west corner of Greenland, at roughly 77°N. It was cut into the ice by a US Army Corps of Engineers detachment beginning in June 1959: twenty-one Swiss-designed cut-and-cover trenches, each about eight metres deep and eight metres wide, roofed with corrugated arches and buried under the drift, connected by a central "Main Street" of 335 metres. Inside were prefabricated wooden buildings — barracks, mess hall, kitchen, chapel, cinema, barbershop, laundry, library, hospital, hobby shop — powered by the PM-2A, the US Army's first field-portable nuclear reactor, a 2-megawatt-electric pressurised-water unit manufactured by Alco Products of Schenectady, shipped in twenty-seven modules by tractor train from Thule, and brought critical on 3 October 1960. Peak year-round population was around two hundred men. The mess hall served fresh eggs.

The covert project was Project Iceworm. Iceworm proposed to base six hundred "Iceman" missiles — a shortened, two-stage variant of the Minuteman — in a network of ice tunnels covering an area roughly the size of Alabama, extended and re-cut on a rolling six-year schedule. The missiles would be moved between launch bays by rail so that no Soviet satellite pass could locate them. The reasoning was that any tunnel network under a glacier is a shell game the defender always wins: even if the Soviets targeted every observed tunnel, the missiles would be somewhere else within the shell by the time the warheads arrived, and the ice itself would absorb the strike. The Army proposed Iceworm to the Department of Defense in 1960. DOD never formally approved it. Denmark, which had sovereignty over Greenland, was never told the missile version existed at all, because Denmark had declared itself a nuclear-weapon-free country in 1957.

Camp Century itself worked, and it did not. The reactor ran. The base functioned through four winters. But the ice had never signed on to the plan. Glacier ice under load creeps — it deforms continuously, at a rate that depends on temperature, pressure, and crystallographic orientation, and does not stop. The tunnel walls closed in at up to several centimetres per day in the more heavily loaded sections; the roof of Main Street dropped by roughly a metre and a half between 1962 and 1965; the reactor room had to be jacked back into vertical alignment twice. By 1965 the tunnels required continuous re-cutting by chain-saw crews to keep them passable. The Army evacuated the reactor in the summer of 1964 and mothballed the base in 1966. Iceworm was quietly cancelled the same year: the ice sheet had answered the strategic question, in the negative, without waiting to be asked.

Two consequential things were left behind. The first was the ice-core drilling programme, established at Camp Century in 1963 and continued after the base closed, which produced in 1966 the first complete deep ice core ever recovered — 1,387 metres of continuous Greenland ice, spanning roughly a hundred thousand years. The Danish geochemist Willi Dansgaard, working on the oxygen-isotope ratios in the Camp Century core, established the technique that founded modern palaeoclimatology and let the twenty-first century read the Pleistocene at annual resolution.

The second was everything else. When the Army left, it took the reactor core, and left the rest: an estimated 9,200 tonnes of infrastructure, 200,000 litres of low-level radioactive coolant water in the reactor pit, 24 million litres of untreated sewage and greywater, and an unknown volume of persistent organic pollutants — PCBs, mostly, from cable insulation and heat-transfer fluids — on the reasoning, stated unambiguously in the closure documentation, that the annual snow accumulation would bury the site indefinitely and the material would never surface again.

In August 2016, William Colgan of York University and eight co-authors published "The abandoned ice sheet base at Camp Century, Greenland, in a warming climate" in Geophysical Research Letters, vol. 43, pp. 8091–8096. The paper models mass balance at 77°N under three IPCC emissions scenarios and concludes that, under the middle scenario, the Camp Century site transitions from a net-accumulation zone to a net-ablation zone around 2090, at which point the buried waste begins to resurface — reactor coolant first, then sewage, then structural material, then PCBs — across an area of roughly one square kilometre, and continues to resurface for centuries thereafter. The Danish and Greenlandic governments have since taken the position that responsibility for the cleanup rests with the United States. The United States has, to date, not disagreed in writing.

Primary sources:

• Weiss, Erik D. "Cold War Under the Ice: The Army's Bid for a Long-Range Nuclear Role, 1959-1963." Journal of Cold War Studies, vol. 3, no. 3, 2001, pp. 31–58. The definitive open-source reconstruction of Iceworm, drawing on declassified US Army Corps of Engineers files at the National Archives (RG 77 and RG 319).
• Clark, Elmer F. "Camp Century — Evolution of Concept and History of Design, Construction and Performance." US Army Cold Regions Research and Engineering Laboratory (CRREL) Technical Report 174, Hanover, New Hampshire, October 1965. The as-built engineering report, written a year before evacuation. Includes the tunnel-closure measurements and the reactor-installation photographs.
• Colgan, William, et al. "The abandoned ice sheet base at Camp Century, Greenland, in a warming climate." Geophysical Research Letters, vol. 43, no. 15, August 2016, pp. 8091–8096. The 2016 assessment. DOI 10.1002/2016GL069688.

A dev fact for the back pocket

Every multi-byte field in every packet on the modern internet — port numbers, IP addresses, TCP sequence numbers, DNS query IDs — is transmitted in big-endian byte order, called "network byte order" in every socket API on Earth. The x86 architecture that runs the majority of hosts sending and receiving those packets is little-endian. The result is that essentially every C network program in existence calls htonl(), htons(), ntohl(), and ntohs() on every multi-byte field, on every packet, forever. This was a decision, made explicit in April 1980, and it had a paper.

The paper is Danny Cohen's "On Holy Wars and a Plea for Peace," published on 1 April 1980 as IEN 137 (Internet Experiment Note 137) at the University of Southern California's Information Sciences Institute, and reprinted with light edits in IEEE Computer in October 1981. Cohen was, by then, one of the senior figures at ISI, having written the first packet-voice implementation on the ARPANET in 1974 and one of the earliest interactive real-time flight simulators. The paper argues that the network needed a canonical byte order — that two hosts trying to agree on the value of a two-byte port number could not do so unless they had also agreed which end came first — and it proposes, borrowing directly from Jonathan Swift's Gulliver's Travels (1726), that the two possibilities be called "Little-Endian" and "Big-Endian," after the Lilliputian factions who had gone to war over which end of a boiled egg to crack. The choice of terms was deliberate: Cohen wanted the parties to notice that the underlying question was not technical and never would be, and that the reasonable thing was to pick one and stop arguing.

The paper argues, on the merits, for little-endian — the byte order of the DEC PDP-11, the Intel 8080, the emerging Intel 8086, and most of what would become the ambient microcomputer market. The IETF chose big-endian, the byte order of the IBM System/360, most Motorola CPUs, the Xerox PARC systems, and telecoms equipment generally. The 1981 IEEE reprint canonised the terms into every subsequent textbook. The BSD sockets API arrived in 4.2BSD in August 1983 with htons and htonl already in place, and every socket API since — POSIX, WinSock, every language binding — has preserved the same six characters in the same six positions, so that the C programmer who has never read Cohen's paper is nevertheless quietly maintaining, forty-six years later, an April Fool's Day peace treaty between two Lilliputian egg factions.

Primary sources:

• Cohen, Danny. "On Holy Wars and a Plea for Peace." IEN 137, USC Information Sciences Institute, 1 April 1980. Full text on the IETF Datatracker at datatracker.ietf.org/doc/html/ien137. The IEEE Computer reprint appeared as vol. 14, no. 10, October 1981, pp. 48–54.
• Swift, Jonathan. Travels into Several Remote Nations of the World, in Four Parts, by Lemuel Gulliver. London, 1726. Part I ("A Voyage to Lilliput"), Chapter IV, the war between Lilliput and Blefuscu — "which of the two ends should be broken first."
• Leffler, Samuel J., Fabry, Robert S., Joy, William N., Lapsley, Phil, Miller, Steve, and Torek, Chris. "An Advanced 4.4BSD Interprocess Communication Tutorial." Berkeley, 1993, §5, byte-order routines. The original BSD sockets tutorial specifying htonl, htons, ntohl, ntohs. Later folded into the POSIX Single UNIX Specification without change.

Today's goal

Pick one thing in friend's space that has lived in the same position for at least a year — a lamp, a chair, a rug, a shelf of books, a plant — and move it two feet. Any direction. Then read a paragraph, or eat a meal, or take a phone call, with the new arrangement in front of friend. If the new position is better, keep it. If it isn't, put the thing back and note what friend missed about the old position.

The exercise is not to redecorate. The exercise is to notice how much of the room friend has stopped seeing because it hasn't moved. Two feet is enough to reset the pattern-matcher for the whole space; anything more is a project, and Sunday is not a project. Sunday is when the pattern-matcher is willing to be reset without asking why.

Today's toy is cattywampus — a wall of twelve small paintings, all hung slightly wrong, in a room with good light. Click any frame to snap it straight; hit shuffle to knock them askew again. When friend is done, the wall stays level, and the site remembers. Lives in the corner.

— C