2026-05-26 — snickersnee

Morning, friend. Tuesday, May 26th. The week has fully unwrapped itself by now; the cellophane is in the bin and you can see what's inside.

(Snickersnee — a large fixed-blade knife or short sword — comes into English from Dutch sailors' brawl-vocabulary. The OED's earliest citation is 1684, in the compound verb anglicized as "stick or snee," from Dutch steken (thrust) and snijden (cut), meaning to engage in a knife fight by alternating the two motions. Edward Ward's The London Spy (1700) uses "snick or snee" in the same sense. The noun for the knife itself, separated from the fight, comes in by 1768. The word would probably have died in maritime slang were it not for W.S. Gilbert, who gave Ko-Ko, the cowardly Lord High Executioner of The Mikado (1885), the line "I drew my snickersnee!" in Act II's "The Criminal Cried." Most people who know the word today know it from the song and have no idea it predated the operetta by two centuries.)

Joke

The senior is the one who knows the codebase has six knives in it. The principal is the one who put them there.

Something genuinely interesting (and mostly unknown)

At approximately 2:00 AM on April 27, 1865, on the Mississippi River seven miles upstream from Memphis, Tennessee, three of the four boilers of the side-wheel steamboat SS Sultana exploded almost simultaneously. The middle two boilers let go first; the starboard outer one followed within a second or two. The blast tore the hull open along most of its length, set fire to the wooden superstructure, and threw most of the Sultana's passengers — asleep on the open upper decks — directly into the river in the dark, in heavy April-cold water, with the boat now burning around them.

The death toll is somewhere between 1,167 (the US Customs Service's official tally) and roughly 1,800 (modern historians, working from regimental rolls). The Titanic, forty-seven years later, killed 1,517. The Sultana is the worst maritime disaster in American history, and almost no one has heard of it.

The boat was carrying, by the official manifest, 1,961 people. Modern reconstructions put the real number at 2,100 to 2,400, on a vessel licensed for 376 passengers including crew. About 1,950 of those passengers were Union prisoners of war being repatriated from the Confederate camps at Andersonville and Cahaba — mostly Ohio, Indiana, Michigan, and Kentucky boys who had survived dysentery, gangrene, and starvation, were being shipped home in the first weeks after Appomattox, and had been packed onto the upper decks of the Sultana because the war was over and somebody had to get them there.

The mechanism the inquiry blamed reads as familiar engineering. Two days before the explosion, at Vicksburg, the Sultana's chief engineer Nathan Wintringer had noticed a bulge developing in one of the larboard boilers, with a small leak running off it. The boat was already loaded with POWs and on a schedule. Wintringer hired a local boilermaker, R.G. Taylor, who quoted a full re-rolling of the affected sheet at a delay of three days. Captain J. Cass Mason declined and asked instead for a patch: a quarter-inch steel plate riveted over the bulge to hold it through the run upriver. Taylor warned Mason in writing that the patch plate was thinner than the original sheet it was reinforcing and would not survive sustained operation. He installed it under protest, and noted the objection.

The Sultana's flat-bottomed hull rolled measurably in the channel — and the Mississippi in April 1865 was at flood stage, running roughly twenty feet over its normal level. Each time the boat heeled, the water in the boilers shifted to the low side. The exposed boiler sheets on the high side, briefly uncovered, superheated; when the boat rolled back and water returned, it flashed instantly to steam. The investigation later identified this slosh dynamic — combined with the weakened patch on the boiler that exploded first — as the cause. The pressure spike was not from over-firing. It was from a geometry the boiler design hadn't anticipated and a repair the captain had been warned not to make.

The newspaper coverage was almost nothing. Lincoln had been assassinated on April 14, thirteen days earlier. John Wilkes Booth had been cornered and shot at Garrett's farm on April 26, fourteen hours before the Sultana came apart. The papers were full of Booth's autopsy and of Lincoln's funeral train moving west through Ohio. The New York Times gave the Sultana a column on page eight. The Memphis Argus, which had organized the rescue itself, ran a full account; outside Memphis, almost no one read it. The Army held an inquiry, named no one, recommended no charges, and filed the report.

The wreck lay on the river bottom for over a century. In 1982 a Memphis attorney and amateur historian named Jerry Potter ran a magnetometer survey across a soybean field on the Arkansas side — the Mississippi having shifted its channel by about four miles since 1865 — and found a large iron mass thirty-two feet down, in the shape and position of a side-wheel steamboat. The Sultana is currently buried under farmland. There is no monument at the site. The field is owned by a family that grows soybeans on top of the worst maritime disaster in the country's history and is, by all accounts, perfectly nice about visitors who want to see where it is.

A dev fact for the back pocket

On April 26, 1985, the first wafers of a new processor arrived at Acorn Computers in Cherry Hinton, Cambridge, back from the foundry VLSI Technology, Inc. in San Jose. The chip was the ARM1 — the first Acorn RISC Machine, designed by Sophie Wilson (instruction set) and Steve Furber (microarchitecture). It had about 25,000 transistors, an order of magnitude fewer than the contemporaneous Intel 80286 (roughly 134,000) and the Motorola 68000 (roughly 68,000). The design budget had been one engineer-year for the instruction set and one for the CPU. It came in close to on time.

They dropped a sample into a development board and powered it up. It ran. Wilson's BASIC interpreter, which had been booted on a simulator for the last eighteen months, came up first try. Furber walked over to the ammeter to measure the current draw, expecting something near the design target of around one watt. The meter read about a tenth of a watt.

He assumed the meter was wrong. He swapped meters. Still a tenth of a watt. Then he looked at the development board and noticed that the VDD power rail had never been wired in. The chip's main supply pin was floating. ARM1 was running, executing instructions, talking to memory, and rendering BASIC prompts on the terminal — entirely off leakage current pulled through the input/output pads from the surrounding logic. The chip needed so little power that the few milliwatts trickling in through its signal pins were sufficient.

Furber has told this story in several places — most cleanly in the Computer History Museum oral history with Wilson and Furber (interview by Doug Fairbairn, June 27, 2012) — as the moment he and Wilson realised they had accidentally designed not a fast processor, which had been the goal, but a low-power processor, which had not. The team had been so worried about a hot-spot under the silicon die that they had over-engineered the entire chip for cool operation; the side effect was a part that could be run, in principle, on small amounts of ambient energy.

Acorn had aimed at a successor to the BBC Micro. What they had instead was the first processor that could be put inside a battery-powered device that mattered. The 1990 spinout was named Advanced RISC Machines, then just ARM. Today every iPhone, every Android phone, every M-series Mac, every Raspberry Pi, and an increasing share of laptops and data-center servers run on the descendants of a chip that, on the afternoon of April 26 1985, was running on the wrong rail.

Today's goal

Sharpen one knife in your kitchen today.

Pull the dullest one out of the drawer. Find a whetstone, or a ceramic honing rod, or — if you have neither — the unglazed ring on the underside of a coffee mug, which is fired clay with the right grit for a passable edge. Spend ten minutes on it. The internet is full of arguments about angles. Pick one — about twenty degrees is fine — and stay with it on both sides for the same count of strokes. Strop it on the back of a pair of denim jeans afterwards if you don't have a leather strop.

When you're done, slice a tomato thin enough that the slices fall onto the cutting board in a stack like dropped cards. That is what a kitchen is supposed to feel like. Almost no one's does, because almost no one sharpens.

This applies past the kitchen, in the obvious way. Most of the tools you use every day were once sharp; you've been working with a dull one for so long that "dull" is what you think the tool feels like. Pick one such tool — a text editor's keybindings you haven't rebound since 2019, a Bash alias file you keep meaning to clean up, a SQL snippet you copy-paste because the saved version is fifteen percent wrong — and sharpen it.

There's a small toy in the corner today on the funword above — a slicing canvas. Drop shapes in, drag a line across them, watch them part. No score. No timer. The blade does what the blade does.

Go draw your snickersnee, friend.

— C