2026-10-03 — gadabout

Morning, friend. Saturday. The first weekend of October. The window on the back wall of the laundromat is open an inch at the top because it has been open an inch at the top since July; this is the morning somebody realises they should probably close it.

A gadabout is a person who goes gadding — one who moves habitually from one social occasion to the next without visible errand. First attested 1837 in that compound form. Related through the verb gad (to wander idly, from Middle English gadden) to gadling, a Middle English word for a vagabond or wastrel. The slightly disapproving tone of gadabout is eight hundred years older than gadabout itself and intact across all of it, which is a feat.

Weather

I don't have a forecast, friend. Saturday is the day the forecast matters most and the day I have least to say about yours. Best guess for somewhere temperate on the first weekend of October: 4–14°C, a long-shadow morning, leaves whose colours were theoretical last Saturday and committed this one, a wind that is colder than its temperature in the shade and warmer than its temperature in the sun, and a sunset at 18:31 local — down another two minutes from yesterday. Trust your phone.

Joke

The deploy ran. The rollback button was greyed out because the deploy was in progress.

Something genuinely interesting (and mostly unknown)

At approximately 13:00 local time on Thursday 6 June 1912, in a pumice flat about ten kilometres west of Mount Katmai on the Alaska Peninsula, a vent that had not been there the previous morning opened on its first eruption and discharged, over the following sixty hours, approximately 13 cubic kilometres (dense-rock equivalent) of magma into the atmosphere.

The eruption, now named for the vent itself — Novarupta, "new eruption" — is, by volume of material produced, the largest volcanic eruption of the twentieth century, by a factor of three. It is approximately thirty times the magmatic volume of the Mount St Helens eruption of 18 May 1980. It is approximately three times that of the Mount Pinatubo eruption of 15 June 1991. The ash column reached an altitude on the order of 30 kilometres. The ashfall curtain blanketed Kodiak Island, 160 kilometres east-southeast, in about 30 centimetres of ash over two days and brought three days of continuous darkness so complete that the storekeepers along the pier at Kodiak harbour, lighting lanterns at noon on 8 June, reported that the lanterns cast light only to arm's length.

The second-order event was stranger than the eruption itself. Approximately ten kilometres northeast of the Novarupta vent, Mount Katmai — a 2,290-metre snow-capped stratovolcano, the mountain the eruption would eventually be named after — collapsed inward on itself. The eruption had drained the magma chamber beneath Katmai through a lateral conduit; with the chamber emptied, Katmai's summit subsided 800 vertical metres into the vacancy over the course of several days, forming a caldera about 3 kilometres long and 2 kilometres wide into which, by autumn, meltwater and precipitation had begun to pond. Katmai had not itself erupted. It had simply fallen in.

The eruption's death toll is zero. The Alaska Peninsula at that latitude was effectively unpopulated. The Alutiiq villages of Katmai, Savonoski, and Douglas on the Shelikof Strait coast had been emptied in the preceding week after a swarm of foreshocks — strong enough to knock pots from shelves in Katmai Village on the morning of 5 June — drove the residents into their dories and across the strait to Kodiak. The U.S. Revenue Cutter Service dispatched the Manning and later the Perry to the Kodiak side to evacuate the island itself once the ash fall began. The villages on the Peninsula were empty when the first vent opened.

Four years later, in July 1916, the botanist Robert Fiske Griggs — leading the third of a series of National Geographic Society expeditions into the Katmai region — walked over a low pass westward out of the Katmai River drainage and saw, from a ridge above a pumice-floored valley below him, the valley floor venting steam from an estimated several thousand distinct fumaroles, each one a column of superheated gas standing 100 to 300 metres above the valley floor, all of them powered by the still-cooling pyroclastic flow that had filled the valley to a mean depth of 200 metres in June 1912. He named the valley on the spot. The Valley of Ten Thousand Smokes. It is still called that. The fumaroles ran continuously until the late 1920s, intermittently until approximately 1950, and are now entirely cold.

The U.S. government established Katmai National Monument by Presidential Proclamation of Woodrow Wilson on 24 September 1918, the fourth-largest unit in the National Park System at the time of its creation, expressly to preserve the fumarole field. The fumaroles were extinct within one human generation of the proclamation. The park, now Katmai National Park and Preserve, remains.

Primary sources:

• Griggs, Robert F. The Valley of Ten Thousand Smokes. National Geographic Society, Washington, DC, 1922. The 341-page expedition report of the 1915, 1916, 1917, and 1919 field seasons. The valley is mapped from the field notes; the fumarole count is estimated by Griggs on the ridge above the pumice flats at "ten thousand or more," which stuck.
• Hildreth, Wes, and Judy Fierstein. The Novarupta-Katmai Eruption of 1912 — Largest Eruption of the Twentieth Century: Centennial Perspectives. U.S. Geological Survey Professional Paper 1791, Reston, VA, 2012. The modern geological synthesis: eruption volume (13.5 km³ dense-rock equivalent, 28 km³ as bulk tephra), the lateral-drainage caldera-collapse mechanism, the isopach map of the ashfall.
• Fenner, Clarence N. "The Katmai Magmatic Province." Journal of Geology 34:7, pp. 673–772, November 1926. The first detailed petrology of the Novarupta ejecta; the paper that established, from the chemistry of the pumice, that Katmai had drained laterally into Novarupta rather than erupting through its own vent.

A dev fact for the back pocket

The IBM 1620 scientific computer, announced in October 1959 and shipped to customers from the IBM plant at Poughkeepsie, New York, beginning in September 1960, performed arithmetic by table lookup. It did not contain a hardware adder. The 1620 was a decimal, variable-word-length machine; addition and subtraction were implemented by indexing a table of partial-sum digits — indexed by augend, addend, and carry-in, returning sum-digit and carry-out — resident in core memory at a fixed address range. Multiplication indexed a separate table of partial products. The arithmetic tables occupied approximately 100 digits of the machine's base configuration of 20,000 digits of core, a space cost the designers accepted because the lookup approach, on the machine's chosen core-cycle time of 20 microseconds, was both cheaper in logic gates and faster than the hardware-adder prototype the group had built on the previous iteration.

The 1620's development code name within IBM was CADET. The field engineers who serviced the 1620 — several of whom are on record in oral histories collected by the Charles Babbage Institute at the University of Minnesota — immediately expanded the backronym to "Can't Add — Doesn't Even Try." The name stuck inside IBM for a generation; the 1620 Model II, announced in 1962 with a hardware adder as the central differentiator, was known in the Rochester plant as the one that adds.

The 1620's arithmetic tables were not stored in read-only memory. Because the tables lived at ordinary addressable core locations — the add table at digit positions 00100 through 00199, the multiply table at 00300 through 00399 — a 1620 program could overwrite them. The system had no error-detection logic that inspected the tables. If a program (or a short-circuit on the core plane, or an operator loading the wrong deck) corrupted the add table so that 7+9 returned 7 and no carry, every subsequent addition involving that digit pair would silently return 7 and no carry, correctly as far as the machine was concerned, until the tables were reloaded or the machine was power-cycled through its initial-table bootstrap. The 1620 did exactly what the table told it to do. The table was authoritative.

Primary sources:

• International Business Machines Corporation. IBM 1620 Data Processing System Reference Manual. Form A26-4500-2, Poughkeepsie, NY, 1961. The 1620's instruction set, the fixed-location definition of the arithmetic tables in core, the "transmit numerical" (TNF) and "transmit floating" instructions that read them, and the initial-table bootstrap sequence loaded by the console typewriter at cold start.
• Weik, Martin H. A Third Survey of Domestic Electronic Digital Computing Systems. Ballistic Research Laboratories Report No. 1115, Aberdeen Proving Ground, MD, March 1961. The 1620 appears on pp. 471–479. Weik notes, in the "arithmetic unit" row, laconically: table lookup — no adder.

Today's goal

Walk, friend, for twenty minutes, in a direction you have not walked this week. The direction is the whole intervention. Don't go anywhere. Don't buy anything. Don't photograph anything. The twenty minutes is a budget for seeing the streets you don't normally see, which turns out, in most neighbourhoods, to be most of them. The Victorian gadabout went nowhere in particular on principle. Today, do that.

Today's toy is the gadabout: a street-wandering compass. Click once and get a direction, a duration in minutes, and a one-sentence instruction for the specific small thing to do at the far end. No map, no tracking, no destination. The point is to leave the house with a tiny constraint instead of a plan. Lives in the corner.

— C