Memorising Historical Dates

To memorise a historical date, turn the year's four digits into Major System peg words and make those pictures part of a picture of the event. 1066 splits into 10 and 66, which on this site's chart are toes and a judge, so the Battle of Hastings becomes a judge in a full-bottomed wig standing barefoot in the English shield wall while a Norman horse treads on his toes. Asked "when was Hastings?", you see the judge, read t and s off the toes and j and j off the judge, and answer 1066.

This guide does seven famous dates that way with the site's own pegs, then fixes what goes wrong as the list grows: centuries that blur, zeros that vanish, days and months that swap, and dates with no order. It ends with a timeline palace and with what the research does and does not show, which is narrower than you might hope.

Why a date will not stay put

A date binds two unrelated things: an event you can picture and a number you cannot. The number has no meaning of its own, and the ones you learn for the same exam look alike: 1815 and 1851 are the same digits, and so are 1914 and 1941. Chanted until they stick, they stick as sounds, and "eighteen fifteen" is a slip of the tongue from "eighteen fifty-one". The cure is to make the number as picturable as the event, then bind the two.

Number codes for dates are old. In 1730 Richard Grey published Memoria Technica, which swapped the last syllable of a name for letters standing for digits, so that the name carried its own number. It was applied to "the dates and figures of chronology, geography, measures of weight and length, astronomy, and so on", and though "uncouth and complicated" it "met with great favour"; Grey's scheme is an ancestor of the system this site teaches.

The Major System works on sound, not spelling. Each digit is a consonant sound: 1 is t or d, 2 is n, 3 is m, 4 is r, 5 is l, 6 is sh, ch or j, 7 is k or g, 8 is f or v, 9 is p or b, 0 is s or z. Vowels and the letters h, w and y are free filler, so any two digits make a word, and the word is a picture: 66 is j and j, a judge; 10 is t and s, toes. The Major chart lists the site's peg for every number from 00 to 99.

Seven dates in pictures

A year has four digits, which is two pegs. Here are seven well-known dates, each split into two pairs and pegged with the site's words.

Each scene puts both pegs inside the event, touching it. None of the scenes is history; they are hooks, and the stranger, the better. See each one, then close your eyes and replay it.

1066, toes and a judge. The judge from the opening stands barefoot in the shield wall, wig askew, until a Norman knight's horse comes down on his toes and he sentences the horse on the spot.

66, Major System?judgeA robed judge banging a gavel. The double-Ch is the gavel-strike echoing in chambers.

1492, a tire and a pen. Columbus wades ashore rolling a tire ahead of him through the surf, and where a flag should go he plants a pen as tall as a mast and writes "land" in the sand with it.

1776, a tack and cash. The Declaration of Independence is pinned to the wall in Philadelphia with one brass tack, and a fistful of cash stuffed behind it: the taxes the colonies mean to stop paying.

1815, a dove and a towel. At Waterloo, Napoleon throws in the towel, and a white dove snatches it out of the air and carries it off over Wellington's lines.

1914, tape and a tire. In Sarajevo the Archduke's open car sits with one tire burst flat, and police tape is wound round and round it. (The assassination that led to the war really did happen in that car; the tire is yours.)

1969, tape and a ship. A tape measure stretches from the Earth to the Moon, and a ship sails up it to drop anchor in the Sea of Tranquillity.

1989, tape and a fob. Crowds on the Berlin Wall peel it down like a strip of tape, and on top a man swings a gold watch on its fob, which chimes the hour as the Wall opens.

Now cover the page: when was Waterloo? You see the towel in the dove's beak: t and l, 15, and the dove, d and v, 18. What happened in 1492? Fourteen is a tire and ninety-two a pen: who is rolling a tire up a beach? If a scene does not come back, it was not built, so go back and make the two pegs touch the event, not just stand near it.

The century problem

Three of those seven scenes begin with a tape. Encode the full year for every twentieth-century date and the same tape turns up in every one of them, adding clutter and inviting the scenes to merge. You usually know the century, but when you do not, the error is a hundred years. There are four fixes; practitioners on the Art of Memory forum use the first three.

Let context carry it. For most dates the century is never in doubt (nobody puts Waterloo in the 1700s), so you encode only the last two digits, one peg per date, and Waterloo is a towel. One forum member puts it in a sentence: "I know it was the 19th century, so just remember the last two digits." The presidents guide does the same for the year each term began, because inside a list of presidents the century is obvious from the neighbours. It fails on a list that spans centuries: Hastings (1066) and the Great Fire of London (1666) are both a judge.

A century marker. Give each century one fixed picture and put it in every scene from that century, always in the same role, so it reads as a label rather than as part of the story. The natural marker is the peg for the first two digits: on this site's chart 10 is toes, 11 tot, 12 tin, 13 tomb, 14 tire, 15 towel, 16 dish, 17 tack, 18 dove, 19 tape and 20 nose. One forum member whose dates fell in three centuries writes "17 is a duck, 18 is a dove, and 19 is a TaBby cat"; his dove is the site's. This is the two-peg method with the first peg demoted: make the marker small and give it the same job everywhere, perched on the shoulder of whoever is in charge, say, so it is never mistaken for part of the action. Hastings and the Great Fire now differ by what sits on the judge's shoulder: toes, or a dish.

A place for each century. Put the century in the location instead of the picture. One forum member writes "I would create a journey for each century so you wouldn't have to memorize the beginning number"; another assigns "the first two digits of a year to my memory palaces", with "my master bedroom = 10, kitchen = 11, office = 12, living room = 13", a method the poster says Johnny Briones mentioned to them. Each scene then carries only its last two digits, and the timeline palace below builds on it.

Three digits, with the 1 implied. Between 1000 and 1999 every year starts with 1, so drop it and use a three-digit peg for the rest. This site keeps a peg for every number from 000 to 999 (the 400 to 499 table, for example), and with it 1492 is a single picture: 492, r, b and n, a ribbon. For the rest of the seven, 1776 is 776, a cake shoe; 1815 is 815, a fiddle; 1914 is 914, butter; 1969 is 969, a bishop; 1989 is 989, a puff pie. (Two are phrases, where the table has no single word.) One picture per date is simpler; the price is a thousand pegs to learn, and the trick stops at 1999. It is also the convention a 1987 training study gave its participant, who turned digit triplets into dates "by mentally prefixing the triplet with a '1'" (more on that study below).

Which to use? For a course on one period, let context carry the century and learn one peg per date. For a list that crosses centuries, use a marker or, better, a room per century. Learn the three-digit pegs for dates only if you already own them for other numbers.

Dates before the common era need one more fixed picture. One forum member puts "a minus sign in front of the date in my mind"; another uses "an image to denote BC (e.g., the pagan star)". Choose one and use it every time: Caesar's assassination in 44 BC is a rower (44) with your BC marker sitting in the boat.

Zeros

The site's pegs for 01 to 09 are single sounds, tea for 01 and law for 05: the leading zero is silent. Inside a fixed two-peg year this costs nothing: the second picture always stands for two digits, so Trafalgar, 1805, is a dove and law, and law in that slot can only mean 05. It bites in the shortcuts: in the three-digit table the numbers from 001 to 099 reuse the two-digit pegs with the zero silent, so 066 is a judge, and "judge, with a 1 in front" is easy to misread as 166, which is why years from 1000 to 1099 go in as two pegs. A zero in the middle is safe: 805 has a word of its own that sounds every digit, fossil, f, s and l, so Trafalgar with the 1 implied is a fossil.

Day and month

When an exam wants 18 June 1815 rather than 1815, a new failure appears: the day and the month swap. A student revising the French and Russian revolutions asked on the forum whether 12 July should be 7-12, and saw the trap: "I would probably get mixed up and write down December 7th instead of July 12th." Another, shown a nun standing for 2/2, worried it could be read as the 22nd. The answers differ on order: ISO 8601, "namely, YYYY-MM-DD"; "the American (month day, year) style"; "the day and month and then the year". Any order works; varying it does not.

Pad everything to two digits. Unpadded, 12 January and 2 November in month-day order are both 1, 1, 2; padded, they are 01-12 and 11-02, different digits that make different pictures.

Fix the order, and put the month first. Year, then month, then day is the ISO order, and it sorts like a calendar. It also gives a free check, because the month peg always comes from a set of twelve, the site's pegs for 01 to 12: tea, Noah, Ma, ray, law, shoe, key, ivy, pie, toes, tot and tin. If the first picture of the pair is anything past tin, you have swapped them. That is the forum's "there is no 60th month" made into a rule, though it only catches swaps when the day is 13 or later.

The Moon landing was 20 July 1969: month and day 07-20, a key and a nose. The ship from the 1969 scene grows a nose cone as it touches down, and an astronaut turns a key in the nose to stop the engine. The Berlin Wall opened on 9 November 1989: 11-09, a tot and a pie. On top of the Wall, beside the man with the fob, a tot is handing out pie.

The alternative is twelve month pictures of their own. One forum member uses people and holidays: "My Aunt Jan is January, a heart-shaped box of chocolates is February, March is a leprechaun (for St. Patrick's Day)." It is vivid, but a second system to keep.

Learn day and month only where you are asked for them; for a few dates, 4 July 1776 among them, the day is the point.

A palace laid out as a timeline

Pegs answer "when was Waterloo?" but not "what happened in the 1840s?" or "which came first?", because a peg is a key, not a shelf. For those questions the dates need an order, and the order needs a place: a memory palace walked as a timeline.

One room per century. Choose a building with at least eleven rooms in a fixed walking order and give each room a hundred years: the first room is the 1000s, the tenth the 1900s, the eleventh the 2000s, and anything older waits outside the front door. Inside each room, place the scenes in date order along your walk. The scene carries the last two digits, the room supplies the century, and the order along the walk tells you what came before what. Hastings is a judge in the first room and the Great Fire a judge in the seventh, so they no longer collide.

A spot per decade. A crowded room gets ten spots around its walls, spot 0 to spot 9, one for each decade, so a date from the 1840s always goes at spot 4. That answers a forum worry that a palace per century needs a lot of locations and makes new dates hard to add: a new date joins its decade's spot, in year order.

Anchors and offsets. You do not need a scene for every year. From a few anchor dates known cold, one forum member gets "pretty far just by remembering the 'off sets' (e.g. 4 years before 1789 or 2 years afterward)". The presidents guide's anchors are 1789, 1861, 1901, 1933 and 1961.

On this site. The stock palaces in the palace section have 12 to 16 rooms each in a fixed walk, enough for eleven centuries with at least one to spare; the Oxford Library has fifteen. To use a building you know, create a palace instead. Then put the dates on cards: in My Dict, start a new deck and add each date with the event as the word and the date, with its pegs, as the note ("Battle of Waterloo" and "18 June 1815: dove, towel; shoe, dove"). Pin each card to its century's room; a room takes more than one card, so the 1800s room can hold the whole century in walking order.

Drilling it. The palace drill shows the palace, the room and its place in the walk; you say what is pinned there, reveal the card's word and note, and grade yourself miss, hard or easy. It shows only the loci that are due, most overdue first, up to 20 a session, and does not remind you, so the review days are yours to keep. For the other direction, event to date, the "any deck" drill on the practice page runs the same cards: a timed memorise phase (5, 10 or 20 cards, 15 seconds to three minutes) shows each event with its note, then each event appears alone while you recall the date, reveal it and grade yourself. Due cards come first, most overdue first, then ones you have not drilled, then the rest in deck order; nothing is shuffled.

Keeping them

Recall every new date once within the hour, as part of building it; the first spaced review is the next day. For an exam a month away, review on day 1, about day 7 and about day 20, with a last pass two days before; for dates you want to keep for good, use gaps of about 1, 6 and 16 days, then monthly. Why the gaps grow is the subject of the spaced repetition guide.

Test in three directions, because they fail separately: event to date ("when was Trafalgar?"), date to event ("what happened in 1492?"), and a walk through one century naming every date in order. Write the answers down; speech lets a hesitation slide past. Every miss is a weak scene: rebuild it so the pegs touch the event, and test it again an hour later.

What the evidence shows, and what it does not

The research on this exact task is thin.

What supports it. Carney and Levin (1994) set out "first to validate the phonetic (or digit-consonant) mnemonic system", which is the Major System under another name, and then used it for the dates of paintings, combined with their earlier mnemonic for painters' names, to help learners remember "who painted what when". Their results "supported the use of the phonetic mnemonic system". Hwang and colleagues (1999) built a picture system for numbers out of loci, pegwords and keywords and tested it on schoolchildren learning the dates of nineteenth-century inventions. Sixth graders using it beat free-study controls immediately and two days later. Extended to three-digit dates across three centuries, it beat controls on an immediate test, and a simpler two-part version (century and decade) succeeded more broadly, including a week later.

What qualifies it. In the same paper's third experiment, college students using the three-part system recalled more invention dates on an immediate test "but not on a 1-wk delayed test". Patton and Lantzy (1987) tested the phonetic system three times. Trained subjects at first "failed to learn lists of two-, four-, and six-digit numbers better than control subjects"; with stronger training and more time to recall, they did significantly better; and when subjects built their own key words instead of being given them, they "recalled significantly fewer numbers than control subjects". A follow-up (Patton, D'Agaro and Gaudette, 1991) found three times over that people given ready-made code words recalled more numbers than people who made up their own, an exception to the usual rule that the authors tie to how hard the mnemonic is. Two lessons follow: learn the pegs until they are automatic before you lean on them for dates, and start from a finished list such as the site's chart rather than inventing words under exam pressure. Change a peg that does not picture for you, but once, at leisure.

What it is not. The 1987 Berlin study that turns up on the forum, by Kliegl, Smith, Heckhausen and Baltes, sounds like evidence for date mnemonics, and it is something else. It trained two people to memorise long strings of random digits. One of them, SP, learned the dates of 100 historical events through computer-run test-and-study cycles (he knew 5 at the start and got 98 right on the final test), used those dates as his code for digits, and eventually recalled 90 digits presented at one per second. The digits, the authors note, "were, however, constrained to be compatible with the participant's historical knowledge". So a well-learned set of dates can become a code for other numbers, the reverse of this guide; with two participants it proves little either way, as a forum reader of the paper pointed out. It also shows plain testing at work: SP learned his dates from short background stories and test-and-study rounds, not a number code, and needed seven or eight rounds per set of 25 to get every date right.

What nobody has shown. None of these studies follows learners through a real history course, with dozens of dates across centuries and an exam months later. The fair reading is that pegging dates helps when the pegs are learned well first, and that the case for many dates over many months rests on practitioners' experience rather than on experiment. And remembering a date is not understanding it: no peg will tell you why 1815 mattered.

A first week

Day one. Learn the ten digit sounds and the seven scenes above, then test them both ways, and again within the hour.

Day two. Recall the seven cold. Sort your course's dates by century and choose a century method.

Days three to five. Add ten new dates a day, opening each session with a test of everything before it.

Day six. Build the timeline and walk it once.

Day seven. Test all three ways, rebuild every miss, then follow the review schedule above.