Tidetable

The ball that falls at one

Wed, 04 Nov 2026 00:00:00 GMT

Every afternoon at five minutes to one, a red ball climbs halfway up a mast on the roof of Flamsteed House in Greenwich, and stops. At two minutes to one it climbs to the top. At one o’clock exactly, it falls.¹ That is the whole performance, and it has run since 1833.

On the Thursday I went to see it there were about forty of us in the courtyard below, most of us holding up phones to film a ball doing what balls do.

I had come to look at a clock with no face. A time ball tells one time a day, to anyone who can see it, and it does so without a sound. It does not tick, it does not strike, it has no hands. For five minutes it gets ready, and then for less than a second it tells you something precise.² It was built for people who needed that second more than they needed anything else the clock could have said.

A clock for ships

The people it was built for were on the river. In the 1830s the Thames below Greenwich was full of ships about to leave for months at sea, and every one of them carried a chronometer: a heavy, carefully sprung watch in a gimballed box, set to the time at Greenwich and kept on it, as nearly as possible, for the whole voyage.

The reason is longitude. A navigator can find local noon from the sun, the moment it stands highest, wherever the ship happens to be. If the chronometer says that local noon arrived at a quarter past two in Greenwich, the ship is two and a quarter hours of the earth’s turning to the west, and since the earth turns fifteen degrees an hour, that is thirty-three and three-quarter degrees west of Greenwich. Every four minutes of difference is one degree. The method is simple. Everything depends on the watch.

So the watch has to be right, or at least wrong in a known way.³ A second of error in the chronometer is fifteen seconds of arc in the position, which on the equator is about a quarter of a nautical mile.⁴ A watch that drifts by a few seconds a day will, after a month, have quietly moved the ship by miles, and miles are the difference between a harbour mouth and the rocks beside it.

Before the ball, a captain who wanted to check his chronometer against Greenwich had to send it ashore, or send a mate with a pocket watch to compare against the observatory’s clocks and row back with the result. Neither is a good thing to do with a delicate instrument. The time ball turned the problem around: instead of bringing the watches to the time, it sent the time out to the watches, over open water, at the speed of light.

The idea came from a naval officer, Robert Wauchope, who put up the first one at Portsmouth in 1829. Four years later the Astronomer Royal, John Pond, installed the one at Greenwich, on the roof of Flamsteed House, the oldest building of the observatory, and with breaks for storms and repairs it has dropped ever since.

Why one, and not noon

Noon would seem the obvious hour. It is the hour astronomers measure, the hour the sun itself marks. That is exactly why it was not chosen. At noon the astronomers of Greenwich were busy at their instruments, timing the sun as it crossed the meridian, and nobody could be spared to work the ball. One o’clock gave them an hour to finish, check the clock and get to the roof.

The odd hour stuck. Time balls in other ports followed Greenwich and dropped at one, while those in the United States, starting with the Naval Observatory in Washington in 1845, mostly dropped at noon.⁵ Nobody in a hurry would design a public time signal around the lunch hour of a small staff. But that is what the signal was: not an abstraction handed down from the sky, but a promise made by particular people who would be at the top of a building at a particular moment.

I like the five minutes of warning more than the drop itself. At five to one the ball rises halfway, and that is the instruction: stop what you are doing, find the mast, have the chronometer open and the pencil ready. At two minutes to one it goes to the top, and you watch it. The signal is not the moment the ball lands but the moment it starts to move. Anyone who looks away at one o’clock and back again has missed it, and there is no second chance until tomorrow.

The gun and the ball

Some cities preferred noise. In Edinburgh a gun has been fired from the castle at one o’clock since 1861, and it still is, except on Sundays and a few holidays. A gun works in fog, which a ball does not, and it reaches people who are not looking at the right roof. It has one fault, which anyone who lives in Edinburgh learns sooner or later: sound is slow.

Sound covers about 343 metres a second in air at room temperature. Stand a kilometre from the castle and the bang arrives three seconds late; at two kilometres, six. For setting a kitchen clock this does not matter. For rating a chronometer it is a disaster, unless you know your distance and subtract.⁶ The railways had a noise of their own.

The startings and arrivals of the cars are now the epochs in the village day. They go and come with such regularity and precision, and their whistle can be heard so far, that the farmers set their clocks by them, and thus one well-conducted institution regulates a whole country.

Henry David Thoreau, Walden (1854), “Sounds”

Thoreau was writing about the line that ran past Walden Pond, in Massachusetts. The farmers there had no observatory; they had a whistle, which was late by however far they lived from the track. Edinburgh solved the same problem by doing both: a time ball on the Nelson Monument on Calton Hill, which ships in the Firth of Forth could see, and the gun for everyone else.

A ball is seen, not heard. From across the river at Greenwich the light takes millionths of a second to arrive, which is no delay that any sailor could have measured. That is what I meant by honest. The ball has nothing to correct for. Whoever sees it begin to fall sees it at the same moment, wherever they stand, as long as they can see it at all.

The ball has its own faults. It cannot be seen at night, in fog or in heavy rain, and it is not raised on very windy days, because a ball travelling up and down a mast in a gale is a danger to the mast and to the people beneath it. A gun can be fired in any weather. A ball has to be believed, which is perhaps why so few of them are left.

Wires

The telegraph changed what a time ball was. Until about 1850 a ball could only report the time of the clock beside it. With a wire, it could report the time of a clock a long way off, and do so without anyone on the spot needing to know what time it was.

The best example is at Deal, on the Kent coast, where ships waited in the Downs, the sheltered water between the shore and the Goodwin Sands, for a wind to take them down the Channel. In 1855 an old Admiralty signal tower on the seafront was given a time ball, and from then on it dropped at one o’clock when an electric signal arrived from Greenwich, about a hundred kilometres away. The keeper at Deal raised it at five to one and at two minutes to one, exactly as at Greenwich, and then waited for the wire to let it go.⁷ It worked, on and off, until 1927.

The railways had already found that one time was easier to run than many. The Great Western Railway kept London time on its line from November 1840, and in 1847 the Railway Clearing House recommended that all its member companies do the same. For a while Britain ran two times at once: the railway’s, and the one the sun gave each town. The clock on the Exchange in Corn Street, Bristol, still has two minute hands, about ten minutes apart, one for Bristol and one for London. Greenwich time did not become the legal time of Great Britain until 1880.

There was even railway time observed in clocks, as if the sun itself had given in.

Charles Dickens, Dombey and Son (1848), chapter 15

Dickens wrote that as a complaint, about a London neighbourhood torn up for a railway. Read now, it is a fair description of what the time ball and its wire had done: the sun had not given in, but it had been overruled by a clock in a building on a hill.

What a ball is for now

Radio made the time ball unnecessary. From 1924 the BBC broadcast six short pips before the hour, and a ship could hear Greenwich anywhere it could pick up the signal, in fog, at night, without a mast in sight. Chronometers were rated by radio, then by satellite, and now most of the world takes its time from atomic clocks without thinking about it, as part of the phone it was filming the ball with.

Yet the balls keep falling. At Lyttelton, the port of Christchurch in New Zealand, a time ball station built in 1876 was damaged in the earthquakes of 2010 and 2011, and its tower came down in an aftershock that June. It was rebuilt, and reopened in November 2018.⁸ Nobody in Lyttelton needed it to set a chronometer. They rebuilt it anyway, the way you repair a clock that belonged to a grandparent: not because you need it to tell you the time, but because it is the one that always did.

The most famous time ball in the world does not tell the time at all. Since the last night of 1907 a ball has been lowered down a pole in Times Square, New York, to mark the start of the new year. It is a time ball turned upside down: it moves slowly for a minute and arrives at midnight, instead of waiting at the top and leaving at one. But the grammar is the same. Everyone looks at one thing, at one moment, and agrees.

Five minutes of warning

I stayed in the courtyard after the drop. Most of the crowd left at once, the way people leave a firework display. The ball sat at the bottom of its mast, and a man near me checked his watch against it, and then, I think, against his phone, and nodded to himself as though both had passed.

What the ball gives is not really the time. Anyone in that courtyard had the time to the nearest millisecond in a pocket. It gives something rarer: five minutes in which a number of strangers agree to watch the same thing, and then one moment that is exactly the same for all of them. The warning matters more than the drop. A signal you have to prepare for is a signal you are likely to catch.

I have started keeping my own small version of it. At five to one on a working day I stop what I am writing, and at one I start again. It does not make me more punctual. It makes the day have a middle, which is more than most clocks manage.

Notes

The times are those the Royal Observatory gives for its visitors. The ball runs on Greenwich Mean Time in winter and on British Summer Time in summer, so in summer it falls at noon by the sun’s reckoning, give or take.
Watch the top of the mast, not the bottom. By the time the ball lands, the moment it was announcing is already gone.
A navigator never reset the chronometer. He wrote down how far it was wrong (its error) and how much that changed each day (its rate), and applied both in his sums. Opening the case to move the hands would have done more harm than any drift.
The earth turns 360 degrees in 24 hours, which is 15 seconds of arc for every second of time. A minute of arc of latitude is one nautical mile, so 15 seconds of arc is a quarter of one. Away from the equator, the lines of longitude close in and the error shrinks with them.
The Washington ball dropped at noon, local time, from 1845. Ships in American ports had the same problem as ships on the Thames; they were simply solved by a different lunch hour.
At about 20 °C. Sound is a little slower in cold air, about 331 metres a second at freezing, so on a January afternoon the gun is later still.
The line was not always reliable: on some days the signal arrived too weak to release the ball, or not at all. For years the wire ran underground without proper insulation, which near the sea is asking for trouble.
I have not been to Lyttelton. What I know of it comes from photographs of the rebuilt tower, which looks, in the way of rebuilt things, slightly more certain of itself than the old one did.

The last train is a promise

Sun, 01 Nov 2026 00:00:00 GMT

The last train out of the city leaves at eight minutes to midnight, and on most nights it carries fewer people than it has doors. I have counted. On a Tuesday in October there were eleven of us across six carriages: two nurses going home from the late shift, a man with a cello case who slept from the first station, a group of students who got off together without saying goodbye, and the rest of us, each in a seat facing backwards, watching the platform lights give up one by one.

By any arithmetic the 23:52 should not exist. A bus would do the job for a tenth of the cost, and a taxi voucher for the nurses would do it for less. The operator knows this; every few years a consultant’s report says so in a table.¹ The train runs anyway, and I have come to think that it runs because it is not really a train. It is a promise printed at the bottom of a timetable.

What the last line says

Read any timetable from the bottom up and it changes character. The early columns are about capacity: how many people, how often, at what speed. The last column is about something else. It tells you how late you may stay out, how far from home a conversation can wander before it has to stop. It sets the outer edge of an evening.

When a railway cuts its last train, the people who complain are rarely the people who used it. They are the people who might have. The woman who has never once missed the 23:52 still plans her dinners around the knowledge that she could. Take the train away and the dinner ends at ten, because now there is a risk, and the risk is enough.

A timetable is a list of times. Its last line is a list of permissions.

I wrote that sentence on the back of a ticket on the Tuesday in October, and I still believe it, although it reads better at midnight than in the morning.

Who it is for

The nurses are the obvious answer, and they matter. A town that wants people to work at night owes them a way home that does not depend on someone waiting up. But the cellist matters too, and so do the students, and so does whoever I am on that train: a person who stayed too long at something good. The last train is the town’s way of saying that staying too long is allowed, within reason, once.

There is a moment, a few stations out, when the driver dims the lights in the empty carriages to save power, and the train goes on through the dark with only our carriage lit, like a room that has been carried out of a house. Nobody speaks. The cellist sleeps. It is the most public place I know in which to be entirely alone.

The cost of a promise

I do not pretend this is an argument a finance committee will accept. A promise is expensive precisely because it is kept on the nights it is not needed. That is what makes it a promise and not a service.² The day the 23:52 runs only when it is full, it will stop being the last train and become one more train that happens to be late.

So I count the passengers, and I do not tell anyone the number. Eleven is a small number. It is also exactly enough people to fill a town with the knowledge that it can get home.

Notes

The most recent one is careful, and fair, and about forty pages long. Its recommendation is in the second paragraph; the remaining thirty-nine pages explain why nobody will follow it.
The distinction comes from my grandmother, who said it about lighthouses, not trains. She had never seen a lighthouse; she meant her brother, who phoned every Sunday whether or not he had anything to say.

The cellist again

Tue, 20 Oct 2026 00:00:00 GMT

The man with the cello case was on the 23:52 again. He slept from the first station, as before, with one hand on the case as if it might leave without him. I have now seen him four times and never seen him awake, and I have decided that I do not want to know where he plays. Some regulars are better left as weather.

Four minutes, a follow-up

Sun, 12 Jul 2026 00:00:00 GMT

After the essay about my kitchen clock, readers sent me their own wrong clocks: fast, slow, and in one case running backwards on purpose. The most common gap is five minutes. The record is twenty-two, held by a reader who says she has never once been late and never once known what time it was. I am printing no names. Every one of these clocks is a household secret.

The kitchen clock is four minutes fast

Sun, 07 Jun 2026 00:00:00 GMT

The clock in my kitchen is four minutes fast. I know this, everyone who has ever stayed in the flat knows this, and yet every morning it does its job: I look at it, I see that I am late, and I hurry, and I am on time. A clock that everyone knows to be wrong should be useless. Mine is the most useful object in the room.

A lie that works

I asked around, and almost everyone I know has a clock like it. My sister keeps her car clock seven minutes fast. A colleague keeps every clock in her house ten minutes ahead except the one in the bedroom, which is right, “so that I can sleep”. The baker in the street behind the market keeps the clock above his oven three minutes fast and has done so for thirty years, though he could not tell me why.¹

What interests me is that the lie keeps working long after it has been discovered. You would think that knowing the clock is four minutes fast would let me simply subtract four minutes and relax. It does not. The subtraction takes effort, and in the first minutes of the morning I do not have effort to spare. The clock is faster than my arithmetic, and so it wins.

Time you can argue with

I think there is something else at work too. A clock that is exactly right is an authority: there is nothing to say to it. A clock that is slightly wrong is a companion. It has a personality. You can argue with it, mutter at it, forgive it. It belongs to the household in a way that the time on a phone, which is correct to the second and the same for everyone, never will.

Public clocks are meant to be right, and they should be. The ferry would be nothing without them. But the clocks we keep at home are allowed to be ours, and being ours, they are allowed to be wrong in our particular way.

I have thought about setting mine right. I have decided against it. I would be four minutes late for the rest of my life.

Notes

When pressed, he said the bread does not know what time it is, and neither, before the coffee, does he.

The station clock stops for a minute

Sun, 05 Apr 2026 00:00:00 GMT

The big clock in the station hall does something I had never noticed before this spring: its second hand sweeps round a little faster than a second, then stops at the top of the minute and waits, perfectly still, for a signal from somewhere else. Then the minute hand jumps and it begins again. It keeps time by being briefly wrong and then agreeing. I can think of worse ways to live.

Reading a tide table

Sun, 01 Feb 2026 00:00:00 GMT

In the window of the harbour office there is a sheet of paper that predicts the sea for a whole year. It lists, for every day, the times of the two high waters and the two low waters, and how high and how low each one will be, to the centimetre. It was printed before the year began. It is, as far as anyone can tell, correct.

I find this one of the most astonishing documents a town keeps, and almost nobody stops to read it.

How the sea is predicted

A tide is not weather. It is arithmetic. The water rises and falls because the moon and the sun pull on it, and because those pulls repeat in cycles that can be measured: the main one, driven by the moon, takes a little over twelve hours and twenty-five minutes, which is why high water arrives about fifty minutes later each day.¹ Add the sun’s smaller cycle, the slow wobble of the moon’s orbit, and the shape of the particular coast, and you can work out the sea years ahead. The office does not observe tomorrow’s tide. It looks it up.

Here the tides are small. On an ordinary day the difference between high and low water is less than the height of a kitchen table, and visitors from the ocean coasts laugh at it. But small is not the same as unimportant. That table’s worth of water decides whether the old steps by the fish market are dry, whether a boat with a deep keel can leave its mooring, whether the children can walk out to the rock with the iron ring on it.

What it does not say

The table has one great silence, and the people who live by it learn it early. It predicts the sea that the moon makes. It cannot predict the sea that the wind makes. A strong wind from the south piles water against the shore and the high tide arrives higher than printed; a cold wind from the north-east, the one that comes down from the plateau in winter, can push the sea away and leave the harbour lower than anyone expected.²

So the table is read with an eye on the sky. The fishermen I asked do not trust it and do not ignore it. They treat it the way you treat a very good friend’s directions: correct about the roads, wrong about the traffic.

I think that is the right way to read most printed things that claim to know the future. Read the arithmetic. Then look out of the window.

Notes

Tidal scientists call these regular cycles constituents, and the main lunar one has the pleasingly dry name M2. A tide prediction adds together dozens of them, each with its own speed and size for one particular harbour.
On those days the harbour office tapes a handwritten note to the corner of the table. It says, every time, the same three words, which I will not translate because they are not polite.

Low water, lower than printed

Sun, 18 Jan 2026 00:00:00 GMT

The cold wind came down from the plateau overnight and by morning the harbour was twenty centimetres lower than the table promised. The old steps by the fish market had a green stripe nobody had seen since last winter. Taped to the corner of the tide table in the office window was the handwritten note, the same three words as always. A boy was copying them into a notebook. I did not stop him.

What readers write in margins

Sun, 07 Dec 2025 00:00:00 GMT

I bought a second-hand collection of essays this week for the price of a coffee. Three different readers have written in its margins over what looks like fifty years: one in fountain pen, all question marks; one in pencil, who underlines and writes “yes”; and one in biro, who corrects the others. The biro reader is always right and I like him least. The book is better for all three of them.

The pilcrow’s long holiday

Sun, 02 Nov 2025 00:00:00 GMT

Look at the start of any paragraph in a printed book and you will see a small empty space: the indent. It looks like nothing, and that is roughly what it is. It is the place where a mark used to be.

The mark is the pilcrow, ¶, the reversed and doubled letter that word processors still show when you ask them to reveal hidden characters. For a long stretch of the history of writing it was not hidden at all. It sat at the head of a new thought, often painted in red, and told the reader that one argument had ended and another had begun.

A letter that became a sign

The pilcrow started as a letter. Scribes writing in Latin marked a new section, a capitulum, with a C, and over the centuries the C acquired a vertical stroke, then a second one, and its bowl filled in, until it no longer looked like a letter at all but like a small black flag.¹ It was often added after the text was written, by a different person with a different ink: the rubricator, whose job was to put in the red.

That division of labour is where the trouble began. The scribe would leave a space at the start of each section for the rubricator to fill. When printing arrived, the printers kept the habit, setting the black text and leaving the space for a hand to add the red marks later. And the hands, increasingly, did not come.² Books went out with the gaps empty. Readers got used to the gaps. Eventually the gap was the mark.

The indent as a ghost

I find this one of the most touching stories in the history of the page. Every indented paragraph you have ever read carries a small absence at its head, the outline of a red mark nobody painted. We have been reading the ghost of the pilcrow for five hundred years and calling it white space.

These pages put a little of the red back. Each paragraph of an essay has its number in the margin in the second ink, and the first words of every essay are printed in red, the way a rubricator would have marked where a text begins. It is not nostalgia. A number is useful: it lets a reader point to the exact place they disagree with me. That, I suspect, is what the red marks were always for.

Notes

The name has its own long story through French and English, by way of the Greek word paragraphos, a mark in the margin. By the time it reached English it had become, among other things, “pylcrafte”. Spelling was a looser business then.
The economics were simple. The black text could be printed hundreds of times in an hour; the red marks had to be painted one by one. One was cheap and one was not, and the cheap one won.

The bell on a Sunday

Sun, 12 Oct 2025 00:00:00 GMT

The bakery behind the market is closed on Sundays. Still, at a quarter past nine, two windows in the street open, and stay open for a minute, and close again. I watched it happen three Sundays running. Nobody was waiting for bread. They were waiting for the bell, out of habit, and when it did not ring they went back to their mornings a little disappointed and, I think, a little reassured that the week had a shape.

A bell for the bakers

Sun, 07 Sep 2025 00:00:00 GMT

At a quarter past nine on weekdays, and half past on Saturdays, a hand bell rings in the narrow street behind the covered market. It is a small brass bell with a wooden handle, the kind a teacher once used to end a break, and it is rung by whichever of the two bakers has floury hands at the time. It means that the second batch is out of the oven. It has meant that, in this street, for longer than either of them has been alive.

Nobody under fifty seems to know this. The younger residents think the bell is a nice touch, a bit of theatre for visitors. The older ones put their coats on.

The first batch and the second

A bakery’s first batch is for people who are going somewhere: workers, the early school run, anyone who eats standing up. It comes out at six and it is gone by seven. The second batch is for people who are staying. It is the bread for lunch, the rolls for the old woman on the third floor who no longer climbs stairs before ten, the loaf a neighbour carries up for her.

The bell exists because the second batch is a different kind of promise. Its customers do not pass the shop on their way anywhere; they have to be called. Before telephones, before anyone had a clock in the kitchen that could be trusted, a sound was the only way to tell a street that something warm was waiting.¹

A clock that is a sound

I like that the bell keeps time without showing it. It cannot be read, only heard, and it is heard differently from every window: sharp at the corner, muffled behind the courtyard, almost not at all on the top floors, where people have always relied on the people below them to pass the news upwards. In that way it measures not just the hour but the street, its depth and its neighbourliness.

When the bakers were asked, a few years ago, whether they would stop ringing it now that everyone has a phone, the elder of the two said that he would stop when the bread stopped being ready. It was a better answer than the question deserved.²

Notes

Kitchen clocks are a later subject in this series, or at least a cousin of it. They are rarely right and almost never trusted, which is not the same thing.
The person asking was me. I was hoping for a quotation about tradition. I got one about bread, which is the same thing said more usefully.

A timetable in a drawer

Sun, 10 Aug 2025 00:00:00 GMT

Clearing out a kitchen drawer, I found a bus timetable from 2009, folded into eight and soft as cloth. Every route on it still runs. Not one of them runs at the same times. The 14 used to leave at seven minutes past the hour; now it leaves at twelve. Somewhere between the two a meeting was held, and five minutes of everyone’s mornings were moved, and nobody remembers agreeing to it.

On answering letters late

Sun, 06 Jul 2025 00:00:00 GMT

A reader wrote to me in March about the ferry essay. She had grown up on the far shore and remembered the horn differently: two short blasts, she said, not one long one, and only in fog. It was a careful, generous letter, three pages, the kind that takes an evening to write. I answered it last week. It is July.

I want to apologise to her here, in public, because she deserved a faster reply. I also want to say why I am not sure she deserved a quicker one.

The speed of the thing

Most messages now arrive asking to be answered in the time it took to send them. A letter like hers asks for something else. It was written slowly, on purpose, and its slowness was part of what it said: I have thought about this, and I would like you to think about it too. An answer written on the same evening would have been polite. It would also have been a kind of refusal, a way of closing the matter rather than taking it up.

So I took it up. I went back to the ferry. I asked the deckhand about fog signals, and he looked at me as if I had asked whether boats float, and then told me, at length, that she was right.¹ I asked two of the old swimmers, who disagreed with each other and with her. I read the harbour regulations, which settle nothing and are very long. By the time I wrote back I had a better answer than the one I would have sent in March, and three new questions to put at the end of it.

A small rule

I do not want to make a virtue of my own disorder. Some letters need answering at once, and I have failed those too. But I have started to keep one rule, borrowed from a friend who keeps a bookshop: answer a quick message quickly and a slow letter slowly, and never the other way round.

The reply, she wrote back, arrived on the morning of the first warm day of the year. She said it was worth the wait. I suspect she was being kind. I am keeping the letter anyway.

Notes

Two short blasts in fog, one long blast when leaving in clear weather. The essay has been corrected, with a note. Corrections are the most useful letters I receive, and the ones I answer fastest.

Two short blasts

Sun, 15 Jun 2025 00:00:00 GMT

A reader who grew up on the far shore wrote to say that in fog the ferry sounds two short blasts, not one long one. The deckhand confirms it. The long blast is for clear mornings. I have corrected the essay and I am grateful: there is nothing like a correction to prove that someone read to the end.

The ferry that waits for no one

Sun, 04 May 2025 00:00:00 GMT

The first ferry leaves at twenty to seven, and it leaves at twenty to seven. I put it that way because people here say it that way, with the same emphasis on the second half, the way you might say of a friend that she is honest and she is honest. The ramp goes up at 6:39 and fifty seconds. If you are running down the quay at that moment, the deckhand will look at you with sympathy and not move.

I have been that person twice. The first time I was furious. The second time I understood that the deckhand’s sympathy was real, and that it was the point.

A boat as a clock

A ferry that keeps perfect time is not a convenience. It is an instrument. People on the far shore set their mornings by its horn, which carries across the water a minute before departure; the bakery opens its shutters when it hears it, and the old men who swim from the rocks in every season use it to decide when they are cold enough to come out.¹ Nobody planned this. The boat was punctual for long enough that the town arranged itself around the punctuality, the way a garden arranges itself around a wall.

If the ferry started waiting for latecomers, it would not become kinder. It would become vague. Every person on the quay would be negotiating with every other: a minute for the woman with the pram, a minute for the boy who is always late, a minute for me. The horn would sound at 6:41, then 6:43, and the swimmers would stay in the water too long.

The kindness of not waiting

What the deckhand is protecting, when he does not lower the ramp, is everyone else’s morning. His sympathy is for me; his loyalty is to the timetable, which is to say to the forty people who were on time. I find this more moving the older I get. Most institutions bend for whoever is in front of them and call it service. The ferry bends for no one and calls it nothing at all.

There is a second boat at 7:20. Everybody knows that too.² Missing the first ferry has never stranded anyone; it has only made them late, and lateness, here, is something you own. You cannot blame the boat. The boat left at twenty to seven.

Notes

There are four of them, all over seventy. They have an argument about the water temperature every morning and they settle it by the horn: whoever is still in when it sounds has won.
The 7:20 is not nearly as punctual. Nobody minds. It is the second ferry; its job is to forgive.

In praise of the footnote

Sun, 02 Mar 2025 00:00:00 GMT

The footnote has a bad reputation, and most of it is deserved. It is where academics hide the arguments they are afraid to make in the open, where lawyers bury the clause that will matter, where a certain kind of novelist goes to be clever at the reader’s expense.¹ When people say they hate footnotes, they usually mean they hate being made to look down at the bottom of the page and find that the real conversation was happening there all along.

I want to defend the footnote anyway, because I think its bad reputation comes from bad use, and good use is one of the small pleasures of reading.

Two voices

A good footnote is a second voice. The main text speaks at the pace of the argument; the note steps beside it and says something the argument has no time for. A source. A doubt. A joke that would break the paragraph if it were told inside it. The reader can listen or not. That choice is the whole point: a footnote is the only part of a page that asks permission.

This is why I set my own notes beside the text instead of under it. On a wide screen they sit in the margin, next to the line they belong to, so the second voice is heard at the same moment as the first.² On a phone, where there is no margin, they go back to the end, like the notes in a paperback. Either way the main text never stops to wait for them.

A rule for notes

I keep one rule, which I break only on purpose: a note may add, but it may not rescue. If a sentence needs its footnote to make sense, the sentence is wrong and the note is a bandage. The test is to read the essay with every note removed. If it still stands, the notes are a gift. If it falls over, they were scaffolding, and scaffolding should come down before anyone moves in.

Read that way, a footnote is an act of trust. It tells the reader: here is something I found interesting, and I believe you will decide for yourself whether you do too.

Notes

I will not name names. You know the kind of book. The notes have notes.
Margin notes are old. Medieval scribes and their readers filled the margins of manuscripts with corrections, commentary and, occasionally, complaints about the cold.