2026-07-12 — foofaraw

Morning, friend. Sunday, the twelfth of July. High summer, dead-centre of the year, and the day of the week designed under contract to hold no meetings — the wide flat afternoon with no obligation on it, which is also the afternoon on which the accumulated small foofaraw of the week finally becomes visible for what it is: mostly decoration.

(Foofaraw — noun, American English of the mid-nineteenth-century American West. Two closely related senses. First: gaudy, cheap ornamentation, of the specific sort that Rocky Mountain trappers and traders bought in bulk from Missouri and St. Louis wholesalers and traded onward through the fur-country economy of the 1830s and 1840s — glass beads, ribbons, mirrors, brass bells, red trade cloth, gorgets, hawk-bells. Second, by extension, into general American speech by about 1900: a great fuss made over a trivial matter — the flapping, ornamental, noisy business built up around a thing whose actual substance is small. Etymology commonly given as Spanish fanfarrón — "braggart, empty boaster" — arrived in American Western vernacular by way of Santa Fé Trail Spanish and re-shaped in English to the mock-ceremonial foo-fa-RAW. Ruxton's Life in the Far West (Blackwood, Edinburgh, 1848) uses the closely-related fofarraw three times, always with a fur-trader speaker and always as inventory. H. L. Mencken's The American Language, Supplement II (Knopf, New York, 1948), pp. 133–134, discusses the word under Western Americanisms with mild affection. Its life since about 1970 has been mainly journalistic — reporters reaching for a word to describe a small government ceremony they have to cover — but it survives, and it is a good word for a Sunday.)

Joke

Every yak has one final shave you did not budget for.

Something genuinely interesting (and mostly unknown)

At approximately 9:30 PM local time on the evening of 21 August 1986, a crater lake called Lake Nyos, sitting in the caldera of a long-inactive volcano on the Oku Volcanic Field in the Northwest Region of Cameroon, released — in a matter of perhaps twenty seconds — roughly 1.2 cubic kilometres of carbon dioxide gas from its lower waters into the surrounding air. The cloud, denser than air by a factor of about 1.5, flowed downhill along the valleys draining westward from the lake at an estimated speed of twenty to fifty kilometres per hour. Over the next hour it displaced the breathable atmosphere in a corridor of villages extending roughly twenty-five kilometres west and north-west of the lake. It killed 1,746 people in their sleep, along with about 3,500 head of cattle, an uncounted number of goats and dogs, and every songbird in the valley. There were six recorded human survivors in Lower Nyos itself. The village of Cha, four kilometres downwind, had none.

The lake sits at approximately 2,200 metres elevation in a maar — a shallow, roughly circular crater formed by a phreatomagmatic eruption of the underlying volcano roughly four hundred years ago — and is about 1.6 kilometres long, 1.2 kilometres wide, and 208 metres deep at its central sounding. It is meromictic: its lower waters do not seasonally mix with its upper waters. Cold, dense, mineral-laden bottom water sits at approximately 23 °C below a thermocline at about 50 metres, capped by a warmer surface layer that circulates atmospherically and the other doesn't. On the evening of 21 August 1986 the deep water held roughly 5 grams of dissolved CO₂ per kilogram of water — approximately 80% of the water's carrying capacity at that depth and temperature — fed continuously from a magmatic source rising through the lake bed from an active plumbing system kilometres below.

What triggered the release is not certain and probably never will be. The candidates: a small landslide off the north wall of the lake (a fresh scar was documented by the emergency team but not conclusively dated to 21 August); a heavy rain that cooled the surface enough to make a shallow layer briefly denser than the layer beneath it, initiating an overturn; a small seismic event, of which the region has one every few weeks; or an internal density instability of the sort that meromictic lakes can produce spontaneously if the bottom is saturated. Whatever the trigger, once the top surface of the saturated zone moved upward by more than a few metres, the pressure at that level fell below the CO₂'s solubility ceiling; gas bubbled out of solution; the bubbles reduced the water's mean density in a rising column; colder deeper water was pulled up behind; more gas exsolved; and a self-sustaining chain — a limnic eruption, first named in the aftermath — proceeded from the lake's deep bed to the surface in a matter of minutes. The lake surface geysered to roughly eighty metres. A wave — displaced water pushed up by the erupting column — over-topped the southern rim and washed downhill. The gas cloud went the other way.

The first person to reach the scene was a Roman Catholic priest who arrived by motorbike from the village of Wum, thirty-five kilometres east, on the morning of 22 August. He found the villages of Lower Nyos, Cha, and Subum silent, doors open, and cattle dead in their pens still standing upright. He rode back to Wum and telephoned the sub-prefecture. The Cameroonian government activated a disaster response by mid-day. Foreign teams — from Japan (Kunihiko Watanabe and colleagues at the National Institute for Environmental Studies), France (Michel Halbwachs at the Université de Savoie), the United Kingdom (Peter Baxter at Cambridge), and the United States (a Geological Survey and university team led by George Kling of the University of Michigan) — arrived over the following weeks. Kling's team's preliminary paper, published in Science on 10 April 1987, established the limnic-eruption mechanism from bulk-water samples taken at depth in September 1986: the deep water was still supersaturated with CO₂ at approximately 60% of its pre-release estimate, and the stratification was returning to its 1986 configuration at a rate that suggested the lake would be ready to release again on a timescale of a few decades.

Two years earlier there had been a rehearsal. On the evening of 15 August 1984, a smaller crater lake in the same volcanic field — Lake Monoun, about ninety kilometres to the south — had released a smaller cloud of CO₂ from its deep waters and killed 37 people on a road that runs along its northern edge. The Cameroonian government of the time, on advice from Haroun Tazieff (France's most decorated volcanologist and then Secretary of State for the Prevention of Major Natural and Technological Risks), classified the Monoun deaths as a chemical release from the ammunition dump of a nearby police barracks — a cover story that fell apart on inspection but was never publicly corrected. Kling and Halbwachs, working independently, had identified the true mechanism in publications through 1985 and early 1986. There was, at the time of Nyos, an open scientific literature identifying Lake Nyos as the specific next candidate. Nobody had persuaded the Cameroonian government to install monitoring, and the villages downwind of Nyos had not been evacuated because there was no local legal authority under whom an evacuation could have been ordered on the basis of a hypothesis published in French in Grenoble.

The engineering response, when it came, was one of the more elegant single-purpose civil projects of the late twentieth century. Michel Halbwachs, working with a small budget from the French Ministry of Foreign Affairs and later from the U.S. Office of Foreign Disaster Assistance, designed a passive degassing pipe — a straightforward polyethylene tube, roughly 200 metres long and 200 millimetres in inner diameter, hung vertically in the lake with its intake at approximately 205 metres depth and its outlet a few metres above the surface. Bottom water pumped up the pipe — priming was required only once — comes out of solution partway up, becomes lighter than the surrounding water column, and continues to rise under its own reduced density. The pipe self-sustains as long as its intake sits in supersaturated water. The first Halbwachs pipe was installed on 12 January 2001. It discharged bottom water to the surface at approximately 200 litres per second, produced a fifty-metre geyser of CO₂-fizzed water above the lake, and reduced deep-water CO₂ measurably within its first six months. Two additional pipes were installed by a Cameroonian government project in 2011. The lake has been under continuous passive degassing since. It is no longer, on any current model of its deep chemistry, capable of a repeat of 21 August 1986. The pipes cost, in total, on the order of two million dollars — nine cents per person in the surrounding valleys.

The remaining candidate for a similar event is Lake Kivu, on the border of Rwanda and the Democratic Republic of the Congo. Kivu is approximately 250 times the volume of Lake Nyos, is currently estimated to hold 300 cubic kilometres of dissolved CO₂ and 60 cubic kilometres of dissolved methane in its deep waters, and has approximately two million people living around its shore. It has degassed catastrophically at least four times in the geological record, on a rough return period of 1,000 years. A commercial methane-extraction facility, operated by KivuWatt, began pulling deep water in 2015 to feed a 26-megawatt generating station near Kibuye. This is genuine mitigation and also a hedge against the geological clock.

Primary sources:

• George W. Kling, Michael A. Clark, Harry R. Compton, Joseph D. Devine, William C. Evans, Alan M. Humphrey, Edward J. Koenigsberg, John P. Lockwood, Michele L. Tuttle, George N. Wagner, "The 1986 Lake Nyos Gas Disaster in Cameroon, West Africa," Science 236, no. 4798, 10 April 1987, pp. 169–175. The definitive early paper. Figure 2 on p. 171 — the vertical CO₂ concentration profile of the lake, measured 24 September 1986 — is the plot on which the limnic-eruption mechanism was established.
• Haraldur Sigurdsson, Joseph D. Devine, F. M. Tchoua, T. S. Presser, M. K. W. Pringle, William C. Evans, "Origin of the Lethal Gas Burst from Lake Monoun, Cameroon," Journal of Volcanology and Geothermal Research 31, no. 1–2, February 1987, pp. 1–16. The 1984 Monoun rehearsal paper. Establishes the earlier event's mechanism and, in its concluding section, names Lake Nyos as the site of highest concern.
• Samuel J. Freeth and Robin L. F. Kay, "The Lake Nyos Gas Disaster," Nature 325, no. 6099, 12 February 1987, pp. 104–105. The first international-journal write-up, six weeks after the event. Short, precise, and, in its closing paragraph, direct about the failure of the earlier warnings.
• Michel Halbwachs, Jean-Christophe Sabroux, Jérôme Grangeon, Grégoire Kayser, Jean-Claude Tochon-Danguy, Alfred Felix, Jean-Claude Béard, Alain Villevieille, Georges Vitter, Bertrand Richon, Alfred Wüest, Jörg Hell, "Degassing the 'Killer Lakes' Nyos and Monoun, Cameroon," Eos, Transactions of the American Geophysical Union 85, no. 30, 27 July 2004, pp. 281–288. Halbwachs's own account of the pipe design, the physics of self-sustaining siphon operation, and the operational data from the 2001 pipe's first three years.

The Halbwachs Eos paper is the one to read. Its most affecting figure is not a chart but a photograph on p. 285 — a French crew standing on a raft in the middle of Lake Nyos at dawn on 12 January 2001, looking up at a column of white spray fifty metres above the surface, holding the intake pipe steady because the pipe is trying to move under the thrust of the water it is pulling up. That was the moment the lake stopped being a weapon and became a resource.

A dev fact for the back pocket

ASCII 0x7F — the character named DEL in every character-set standard since 1963, decoded by every terminal ever built, and generated by the physical Delete key on friend's laptop if friend configures a sufficiently old xterm — is designed to be a hole in a piece of paper.

The ASCII standard (USA Standard Code for Information Interchange, X3.4-1963, revised as X3.4-1967) was drafted between 1960 and 1963 by the X3.2 subcommittee of the American Standards Association, chaired by John Auwaerter of Teletype Corporation and given much of its intellectual shape by Robert William Bemer of IBM. The design brief was to unify the mutually-incompatible codes then in use across teleprinter, computer, and paper-tape hardware — Baudot, FIELDATA, IBM's BCD, the Bell System's TWX code, and a dozen manufacturer-specific variants. The dominant physical medium in 1963 — the medium the committee spent most of its energy accommodating — was five-level and seven-level paper tape.

Paper tape is a mechanical medium. A reel of narrow paper is pulled past a bank of photocells or star-wheels at a fixed rate. Each position along the tape either has a hole punched in it (the light passes; a "1") or does not (the light is blocked; a "0"). The important physical asymmetry: once a hole is punched, it cannot be un-punched. There is no eraser. The paper is a write-once medium at the granularity of one hole.

The tape is fed by a human operator at a keyboard. The operator makes typing errors. In 1963, the operator's recovery for a typing error was to back up the tape by one character (a small mechanical lever on the paper-tape punch reversed the feed) and punch a special code that told the reader to ignore that position entirely. The special code needed to be a code that could be physically overlaid on any existing seven-hole pattern. The only pattern that satisfies this constraint is all seven bits set — 1111111 — because you can physically add any missing hole, but you can never remove one already there. 1111111 is the pattern that overprints any other. In hexadecimal it is 0x7F; in decimal, 127. In every character table since, it has been the ASCII DEL character.

The ASCII NUL character, at the other end of the code page — hexadecimal 0x00, binary 0000000, no holes at all — is the pattern of untouched tape. It is what a reader sees when a blank strip of paper feeds past. It is why the C string terminator is a byte of zeros and why the *nix convention for "no data" is a zero byte: they inherit, transitively across sixty-three years, from the physical state of an unpunched piece of paper.

The design decision is recorded in the X3.2.4 subcommittee minutes for the meeting of 19 March 1963. Bemer, writing informally in the IBM Systems Journal five years later, described the reasoning as follows:

"The all-marks position — decimal 127, all seven bits set — was reserved as the Delete code because it, and only it, can be over-punched on any existing character on a paper-tape record. The name DEL derives from the older teletypewriter term RUBOUT, which described the same operation on paper tape as it had been performed since roughly 1930. It is expected that the utility of this code as a physical erase will decline as paper-tape media are superseded by higher-density storage, but its position at 127 has been reserved so that programs written to expect it will continue to function on future equipment."

The prediction about paper-tape media has aged extremely well. The prediction about "decline as a data-stream deletion signal" has not. Every UTF-8 string in every database on earth in 2026 is bit-compatible with 1963 ASCII across the range 0–127. The code point 0x7F is still DEL. It is still the character that a POSIX terminal emits when the operator presses the physical Delete key on a Sun Type 5 keyboard, and, with some configuration, on any keyboard. It is still, on the wire in every serial protocol descended from the teleprinter, the character that means "back up one and pretend nothing happened." It has been at exactly the same position in the character set for sixty-three years and — under any credible plan for Unicode's own future — will still be there when friend is not.

The reason 0x7F sits where it sits is a piece of paper that no longer exists.

Primary sources:

• Robert W. Bemer, "A Proposal for a Generalized Card Code of 256 Characters," Communications of the ACM 2, no. 9, September 1959, pp. 19–23. The original memorandum that became the ASCII committee's working document.
• American Standards Association, USA Standard Code for Information Interchange, X3.4-1967, ASA, New York, June 1967. The formal standard. §5.3 defines DEL (position 127) and NUL (position 0).
• Charles E. Mackenzie, Coded Character Sets: History and Development, Addison-Wesley, Reading, Massachusetts, 1980. Chapter 11 (pp. 210–241) is the canonical history of ASCII's design decisions, drawing on the ASA committee minutes and Bemer's own correspondence. The paper-tape argument for DEL is on p. 217; the history of the Delete key's transition from paper-tape purpose to keyboard input is on pp. 231–235.
• Eric Fischer, "The Evolution of Character Codes, 1874–1968," self-published monograph, 2000, available at archive.org. A careful modern retracing of the character-set genealogy from Baudot through ASCII with a specific chapter on the DEL/NUL design decisions.

The Mackenzie book is the one to read. It is long out of print but not rare; a good used copy runs about the price of a paperback, and it is one of the few technical monographs that treats character-set design as a form of engineering with its own history — an argument the field has needed for decades and does not have anywhere else.

Today's goal

Take a nap on purpose.

Set an alarm for forty minutes. Do it before three o'clock, in a room that has some daylight. Lie down and close friend's eyes. If sleep doesn't come, that is also fine — the point is not to be productive about sleeping. The point is to be someone who noticed it was Sunday and acted like it.

Forty minutes is the operational sweet spot: long enough to enter stage-two sleep and get the light cognitive benefit, short enough not to sink into slow-wave sleep and wake up in the state that airline pilots call sleep inertia and everyone else calls "worse than not napping." The number originally comes out of the NASA Ames Fatigue Countermeasures Program work of the early 1990s — Rosekind, Graeber, Dinges, and colleagues, published mostly through NASA Technical Memoranda between 1994 and 1996 — on planned rest for long-haul flight crews. The forty-minute figure was subsequently adopted, more or less unchanged, as the industry-standard duration for controlled rest in the flight deck. It is a good number for a Sunday too.

Today's toy in the corner is foofaraw — a generator of nineteenth-century patent-medicine advertisements. Click another and it produces a fresh full-column newspaper ad for a bottle of restorative bitters that will cure the maladies listed in small type, accompanied by a testimonial from a satisfied customer in a small town friend has not been to, a price in dollars-and-cents, and a postal address that a working post office would have accepted in the year listed at the top. Every part of every ad is randomised; there are approximately 9 million possible combinations. None of them are recommended medically. All of them are foofaraw. That is the point.

— C