2026-06-27 — omphaloskepsis

Morning, friend. Saturday, June 27. The first proper Saturday of the new astronomical summer; the working week is something somebody else has, and the day in front is the rare unstructured kind that nobody at all is going to manage on friend's behalf.

(Omphaloskepsis — from Ancient Greek ὀμφαλός, omphalós, navel, and σκέψις, sképsis, examination — the contemplation of one's own navel. The word entered English by way of the Hesychast controversy of the 1330s and 1340s in the Byzantine Empire. The Hesychasts were a school of contemplative monks on Mount Athos who practised an inward-turning method of continuous prayer — the Jesus Prayer, repeated silently on the breath — and held, with their bodies, a posture in which the chin was tucked toward the chest and the eyes lowered to the abdomen, on the doctrinal claim that the soul resided in the heart and that physical concentration on the body's centre assisted concentration of the mind. Barlaam of Calabria, a Greek-speaking Italian monk newly arrived at Constantinople, attacked the practice in a series of dialogues circulated between 1337 and 1341 and coined the disparaging term ὀμφαλόψυχοι — omphalópsychoi, "navel-souls" — for its practitioners. He lost the argument. The Hesychast position was defended in the Triads in Defence of the Holy Hesychasts by Gregory Palamas, archbishop of Thessalonica, and was formally upheld by the synods of Constantinople in 1341, 1347, and 1351. The English word came down from Barlaam's slur and not from Palamas's defence, and so the current sense — idle self-absorption rather than disciplined inward attention — is the loser's framing preserved by accident across seven centuries.)

Joke

Spent the morning refactoring the part the test suite covers.

Something genuinely interesting (and mostly unknown)

At approximately 2:00 a.m. on Thursday, 27 April 1865, the side-wheel paddle steamer SS Sultana — eight months old, 260 feet long, 42 feet in the beam, 1,719 gross tons, rated by the Steamboat Inspection Service for a maximum capacity of 376 passengers and crew — was steaming north up the Mississippi River through the high-water flood of the spring melt, about seven miles above Memphis, opposite the wooded islets known as the Hen and Chickens, when three of her four horizontal high-pressure tubular boilers ruptured nearly simultaneously. The forward third of the upper deck and the twin smokestacks were lifted off the hull in one piece, carried upward by the steam release, and came down across the engine deck and into the river. Coals from the firebox were scattered along the remaining superstructure, which was constructed in pitch pine and ignited within minutes. The fire moved aft through the cabins, the pilothouse, and the stern paddle housings, and the Sultana burned to the waterline and sank within an hour.

She had aboard, by the survivors' tally and the Army's reconstructed paroled-prisoner manifest, between 2,100 and 2,300 people. The exact figure has never been settled because the manifest was destroyed in the fire and because Captain James Cass Mason — drowned with the ship — had been bribed into accepting bodies in numbers his own first clerk, William J. Gambrel, also drowned, had registered into the cabin log only by approximate batch counts. 2,137 is the figure carried by the modern National Park Service marker at the Mound City cemetery. Of those aboard, 1,168 are the count of recovered bodies; the official death toll has been variously reported from that floor up to 1,800. Even the lower figure exceeds the death toll of the RMS Titanic by about a hundred and fifty, and the Sultana loss remains the deadliest maritime disaster in United States history by a margin no other event has approached.

The passengers were almost entirely Union Army soldiers newly released from the Confederate prisoner-of-war camps at Cahaba in Alabama and Andersonville in Georgia. The war had ended at Appomattox on 9 April, eighteen days earlier; Jefferson Davis was still at large and would not be captured until 10 May; the camps had been emptied through April; and the released prisoners were being marched in lots to Vicksburg, Mississippi, where they were boarded onto whatever Mississippi River steamers the United States Quartermaster Corps could charter to carry them north to Cairo, Illinois, and the rail connections home. Federal regulations paid the steamboat company five dollars per enlisted man and ten dollars per officer for the transport. The Sultana was the third steamer to arrive at Vicksburg with capacity that week. The two earlier vessels — the Olive Branch and the Pauline Carroll — had been passed over by the local quartermaster, Captain Frederic Speed, on the apparent understanding (the post-war court of inquiry was unable to prove it definitively) that Captain Mason of the Sultana would split a kickback with him on a per-head basis. The two earlier steamers continued north nearly empty.

The Sultana was at the Vicksburg landing for forty-eight hours, during which her chief engineer, Nathan Wintringer, identified a slow leak in the larboard middle boiler — a hairline split in a wrought-iron sheet, propagating along a horizontal rivet line. The standard repair was a patch riveted across the split, sheet-on-sheet, with a forge fire and a working day. Wintringer specified a patch a quarter-inch thick. The Vicksburg boilermaker, R. G. Taylor, working under Mason's pressure to depart on schedule, fitted a 5/16-inch patch — undersized by the standards of the trade — and riveted it cold. The patch held water at dockside test pressure. It would also remain in the boiler when she sank, and is the part of the wreck recovered most often by relic-hunters since the river migrated.

The ship cast off on the morning of 24 April with the new patch and the unverified manifest. She made Helena that night, Memphis by the evening of the 26th. At Memphis she took on coal at the Mound City wharf, departed at midnight, and was three hours upriver when she came apart. The hull came to rest in 30 feet of channel water; subsequent flooding and the natural west-bank migration of the Mississippi carried the wreck inland, and the remains were located by an attorney and amateur historian, Jerry Potter, in a soybean field approximately two miles east of the modern river channel near Marion, Arkansas, during a magnetometer survey conducted between 1980 and 1982. The site is unmarked and on private agricultural land.

The disaster was, in the news cycle of the spring of 1865, almost entirely buried. President Lincoln had been assassinated thirteen days earlier, on 14 April. John Wilkes Booth had been cornered and shot at the Garrett farm in Virginia on 26 April — twenty-three hours before the Sultana came apart. The morning of the 27th, the New York and Washington papers carried Booth on the front page and the Sultana on page four, in single columns under the standing telegraph header. The New York Times of 28 April 1865 ran the Sultana news at a column-and-a-half. The Times of the next day returned to Booth and to the funeral train carrying Lincoln's body west toward Springfield.

A court of inquiry sat at Vicksburg and Memphis through the summer. Captain Speed was court-martialled in August 1865 on charges of dereliction in approving the overload; he was convicted, and then quietly pardoned and reinstated by the Secretary of War on procedural grounds in December. No one else was charged. The steamboat company, the Merchants' and People's Steamboat Line, was uninsured at the level of the loss; it ceased operating in 1866 and the Sultana's sister-vessels were sold at auction. The boilermaker Taylor returned to ordinary trade in Vicksburg. The patch he fitted was, in the boiler-trade press of the period, never separately blamed.

A monument bearing the names of 365 of the dead was raised by survivors at the Mount Olive Baptist Church in Knoxville, Tennessee, in 1916, on the fifty-first anniversary. A smaller stone bearing 84 further names stands in the cemetery at Hillsdale, Michigan. Most of the dead have no marker at all. The Sultana Disaster Museum opened in Marion, Arkansas, in 2015, on the 150th anniversary, in a former two-room schoolhouse three miles from the wreck site, and is staffed by volunteers four afternoons a week.

Primary sources:

Jerry O. Potter, The Sultana Tragedy: America's Greatest Maritime Disaster, Pelican Publishing, Gretna, Louisiana, 1992 — the standard modern reconstruction, by the attorney who located the wreck. Drawn from the original court-martial transcript at the National Archives, RG 153, the records of the Steamboat Inspection Service, and the post-war Sultana Survivors' Association reunion proceedings (1885–1928).
Chester D. Berry, Loss of the Sultana, and Reminiscences of Survivors, Darius D. Thorp, Lansing, Michigan, 1892 — first-person accounts from one hundred and thirty-four survivors, collected by Berry (himself an Andersonville parolee aboard the ship) over the preceding decade.
U.S. War Department, Court-Martial of Captain Frederic Speed, U.S. Vols., Vicksburg, Mississippi, August 1865, transcript, NARA M1523, roll 4.
Steamboat Inspection Service, Inspection Certificate of the SS Sultana, 12 January 1865, St. Louis District, NARA RG 41 — the document certifying the 376-passenger limit that the ship exceeded by a factor of five and a half on her last voyage.

The Potter book is the one to start with. The Berry survivors' volume is the harder read; thirty years after the event the men were still describing the river by the colour of the firelight on its surface, and they were uncertain even then how many of them had come out of it.

A dev fact for the back pocket

In the autumn of 1958, in a basement laboratory of the Computer Centre of Moscow State University on the Lenin Hills above the city, a research engineer named Nikolai Petrovich Brusentsov brought into operation a computer called Setun — named for the small river that runs past the campus into the Moskva — which was, and remains, the only ternary digital computer ever to enter serial production anywhere in the world. The Setun used not binary digits but balanced ternary ones: each storage cell held one of three states, conventionally written −, 0, and + — or, in the symmetric form Brusentsov preferred, −1, 0, +1. The unit was called a trit, a coinage of Brusentsov's on the model of bit. The machine had 18-trit words, equivalent to about 28.5 bits of information per word, and a magnetic core memory in which each trit was stored as the polarity of one core out of a pair — a hardware structure no harder to build, in 1958 Soviet ferrite-core technology, than the equivalent binary store of comparable capacity.

The argument for ternary was older than Brusentsov by twelve years. John von Neumann, in §5.5 of the First Draft of a Report on the EDVAC, 30 June 1945, observed that the storage cost of representing a number of magnitude N in base r requires about r log_r N cell-state-transitions, that this quantity is minimised when r equals the base of natural logarithms — e ≈ 2.718 — and that of the small integer bases, 3 is closer to e than 2 is. He concluded that ternary was, on first-principles information-theoretic grounds, modestly more efficient than binary. He then specified binary for the EDVAC on the practical grounds that two-state circuits — flip-flops, relays, vacuum tubes biased between cutoff and saturation — were already a mature engineering discipline in 1945, and three-state circuits were not. The argument was repeated by Claude Shannon in A Mathematical Theory of Communication (Bell Labs, October 1948, §10) with the same conclusion. The information-theoretic point was uncontested; the engineering point was the deciding one.

Brusentsov re-opened the question for the Soviet computing programme. Balanced ternary — the choice of digits {−1, 0, +1} rather than {0, 1, 2} — has the property that the sign of a number is the sign of its leading nonzero trit and requires no separate sign bit, that negation is a single complement of every trit, that rounding to any precision is the same operation as truncation, and that the comparison of two numbers reduces to inspection of the leading trit of their difference. The arithmetic circuits of the Setun were correspondingly simpler than the binary equivalent of the same era — fewer gates per adder, no two's-complement carry-propagation path, no conditional branch on a sign register because there was no sign register — and the machine ran at roughly 4,500 operations per second with a power draw of 2.5 kW, comparable to the binary machines of the period at a substantially lower component count.

Fifty Setuns were built between 1959 and 1965 at the Kazan Mathematical Machines Factory and distributed to universities and engineering institutes across the Soviet Union, where they were used principally for engineering calculation, surveying, and the early machine-translation experiments at the Institute of Precision Mechanics and Computer Engineering. A second-generation machine, the Setun-70, was designed by Brusentsov's group between 1965 and 1970 with a redesigned instruction set and was built in a further seventy units through the mid-1970s. The Soviet Ministry of Radio Industry — under the Ryad programme of standardisation on IBM System/360-compatible hardware initiated in 1967 — declined to fund either machine into further production. The Setun line was discontinued by Brusentsov's retirement in 1990. He continued to publish in defence of ternary arithmetic in Programming and Computer Software and elsewhere until shortly before his death in 2014, aged 88.

The machine survives only in two examples: one at the Polytechnic Museum in Moscow (in storage; not currently on display), one at the Computer History Museum in Mountain View, California (donated by Brusentsov's family in 2010, reassembled but non-operational). The total worldwide installed base of production ternary computing hardware is, today, zero.

Primary sources:

N. P. Brusentsov, Малая цифровая вычислительная машина "Сетунь" (A small digital computing machine "Setun"), Moscow State University Press, Moscow, 1965 — the design report, written for the engineers operating the machines in the field.
N. P. Brusentsov, J. Ramil Alvarez, Ternary Computers: The Setun and the Setun 70, in John Impagliazzo, Eduard Proydakov (eds.), Perspectives on Soviet and Russian Computing, IFIP Advances in Information and Communication Technology vol. 357, Springer, 2011, pp. 74–80 — the most accessible English-language account, co-authored by Brusentsov three years before his death.
John von Neumann, First Draft of a Report on the EDVAC, contract W-670-ORD-4926, Moore School of Electrical Engineering, University of Pennsylvania, 30 June 1945, §5.5.

The Brusentsov-Alvarez chapter is the one to read first. It contains a photograph of the Setun's main core memory plane: a square of about 45 by 45 centimetres, slate-grey, threaded by hand with sense and inhibit wires, holding three kilotrits of working memory. The wires are visible to the unaided eye. Roughly a hundred and twenty thousand of them were needed to weave the full store. The factory workers who threaded them were paid by the kilometre.

Today's goal

Sit, somewhere reasonably comfortable, for ten minutes, doing nothing.

Not meditating. Not breathing on a count. Not listening to a podcast at half-speed and calling it slowing down. The phone is on the other side of the room. The laptop is shut. The book on the side table is also shut. Look out the window if there's a window. Look at the ceiling if there isn't. Notice that the first three minutes will feel like an hour and the last three will feel like ninety seconds. The bar isn't enlightenment. The bar is that the day, by the end of those ten minutes, will not have been entirely spent on something a counter could increment.

Barlaam called it self-absorption. Palamas called it the prayer of the heart. friend can call it whatever fits. The chair doesn't need a name for it.

Today's toy in the corner is rain on glass — a single windowpane with the weather running down it. Drops accumulate, gather mass, slip, and trail. Click the glass to seed a new drop wherever the cursor lands. Two sliders: one for the wind, which leans every drop's track off vertical, and one for the downpour, which sets the arrival rate of new drops from somewhere above the top edge. It is a window. It is doing what a window does in a storm. The browser tab will sit happily in the background and the glass will keep raining for as long as it is open.

— C