What Is the Earthquake Clue, and What Has Been Built on It?

Somewhere near the start of Romeo and Juliet, in the third scene of the first act, the Nurse settles into the longest reminiscence of her life and remarks, almost in passing, that it is eleven years since the earthquake. From that single number a date for the whole play has repeatedly been manufactured. An earthquake shook England on 6 April 1580, the severest in living memory, and it was famous for a generation. Add eleven years to 1580 and the sum is 1591. On that arithmetic, generations of readers have been told that Shakespeare wrote his tragedy of the Veronese lovers in or about 1591, when he was twenty-seven and barely visible in the London theatre. The clue is the most quoted piece of internal evidence in the play’s chronology, and it is the subject of this study: what the line says, what the argument built on it assumes, what the rival candidates do to the sum, and what the firmer evidence, the printed quarto of 1597, the company name on its title page, the testimony of Francis Meres, the verse itself, actually establishes.

The play itself invites the error, and the invitation is worth noticing before the error is corrected. No tragedy in the canon counts its own time more feverishly than this one. Its Chorus promises the two hours’ traffic of the stage in the Prologue (line 12); its heroine cries for night to gallop apace and reckons the hours until morning; its friar measures a potion’s sleep at two and forty hours (4.1.105); its action compresses a courtship, a marriage, three killings and a double suicide into five days that the dialogue keeps tallying, a fortnight and odd days here, on Thursday next there, three hours, one hour, a quarter of an hour. A work so saturated in counted time trains its readers to count, and the dating debate is in part the story of that training escaping the fiction’s borders. The Nurse’s eleven years gleams at the critic with the same false promise the Friar’s forty-two hours makes to Juliet: the promise that a number in this play is a measurement. Inside the action, the Friar’s number is exact and the catastrophe happens anyway. Outside it, the Nurse’s number is approximate and the catastrophe is a wrong date printed in a thousand handbooks. The parallel is not an argument. It is a warning about the kind of text under examination, a text in which numbers are instruments of pressure and pathos first and of record second, if at all.

The verdict can be stated at once, because everything that follows is its defence. The earthquake line is genuine period colour and almost certainly a real memory lodged in the text: a quake that a whole nation used as a date-marker was exactly the kind of landmark by which a woman like the Nurse would reckon her past, and there is no reason to doubt that the tremor of 1580 stands somewhere behind her words. But as an instrument for fixing the year of composition the clue fails, and it fails instructively. It requires three assumptions to hold at once: that the Nurse’s earthquake is the English earthquake of 1580 rather than any other; that her eleven years is an exact count rather than a character’s round figure; and that the dramatic present of the fiction coincides with the calendar present of the author. Each assumption is contestable, one is contradicted by the words on the page, which set the shaking in Verona and not in London, and the conclusion they jointly produce, a play of 1591, collides with the external record at almost every point where that record can be checked. The date the evidence supports is the middle of the 1590s, most probably 1595, inside a defensible window running from 1594 to 1596. The earthquake can corroborate nothing and refute nothing inside that window. It dates a memory. It does not date a manuscript.

That verdict is not a counsel of despair, and the wider question of when Shakespeare wrote Romeo and Juliet has its own full treatment elsewhere in this series, where the whole chronology is laid out. The concern here is narrower and stranger: one line, one number, one remembered tremor, and the long scholarly attempt to make them carry the weight of a chronology. The attempt deserves respect before it deserves rejection. Dating an undated play is detective work conducted with almost no documents: no manuscript survives, no letter mentions the writing, no diary records a first night. In that vacuum, a line that appears to name a public event and to count the years since it gleams like a fingerprint. Understanding why the gleam misleads requires taking the clue apart slowly: the speech itself, the event behind it, the arithmetic, the objections, and then the evidence that does the real work.

What Does the Nurse Actually Say in Act 1, Scene 3?

The earthquake line cannot be weighed until the speech that carries it is read whole, because the speech is the first objection. The scene is a domestic interview: Lady Capulet has come to raise the question of marriage, and she begins with her daughter’s age. Juliet, she says, is not yet fourteen. The Nurse seizes the subject with both hands. She will lay fourteen of her teeth on the point, and yet, to her teen be it spoken, she has but four (1.3.15 to 17, in the Oxford text of Jill L. Levenson, whose line numbering is followed here with the small variations other editions show). The exchange that follows is worth quoting at the length the argument requires, because every clause of it matters to the clue:

Even or odd, of all days in the year, Come Lammas Eve at night shall she be fourteen. Susan and she, God rest all Christian souls, Were of an age. Well, Susan is with God. She was too good for me. But, as I said, On Lammas Eve at night shall she be fourteen. That shall she, marry; I remember it well. ‘Tis since the earthquake now eleven years, And she was weaned, I never shall forget it, Of all the days of the year, upon that day, For I had then laid wormwood to my dug, Sitting in the sun under the dove-house wall. My lord and you were then at Mantua. Nay, I do bear a brain. But, as I said, When it did taste the wormwood on the nipple Of my dug and felt it bitter, pretty fool, To see it tetchy and fall out wi’ the dug! Shake, quoth the dove-house; ‘twas no need, I trow, To bid me trudge. And since that time it is eleven years. (1.3.18 to 37)

The scene around the speech sets it a standard it pretends to meet. Lady Capulet’s contribution to the age question is a flat negative, she’s not fourteen, and the Nurse counters with the boast that frames everything she says: faith, I can tell her age unto an hour (1.3.12). Unto an hour: the claim is total precision, and the speech that follows is the performance of that claim, dates, feasts, weather, herbs, architecture, all marshalled to certify a count. Nor is hers the play’s only opening chronology. An act earlier, her master was keeping time by a different instrument, telling Paris that his daughter hath not seen the change of fourteen years and asking for two more summers to wither in their pride before she is thought ripe to be a bride (1.2.9 to 11). Capulet counts in summers, the Nurse counts in earthquakes. The first act of the tragedy is a small contest of calendars, the father’s seasonal patience against the servant’s catastrophic memory, and the contest is not decoration: the whole action turns on whose clock governs Juliet, the household’s slow calendar of ripeness or the fevered one of five days that the lovers impose. The dating debate, seen from inside the play, is that contest continued by other means, with scholars volunteering to be a third contestant, counting in earthquakes like the Nurse and in summers like nobody at all.

Read as a document, the speech is a marvel of specificity. A fixed feast supplies the birthday: Juliet was born on Lammas Eve, the night before the first of August, and her fourteenth birthday falls a fortnight and odd days after the scene. A fixed event supplies the weaning: the earthquake, on which day the Nurse had laid bitter wormwood on her breast to break the child of the dug, sitting in the sun under the dove-house wall while the master and mistress were away at Mantua. The dove-house shook; the child tasted bitterness and fell out with the dug; the Nurse, by her own comic logic, needed no bidding to trudge. And the interval is stated twice, framing the anecdote like a refrain: ‘tis since the earthquake now eleven years at the head of it, and since that time it is eleven years at the tail.

Read as arithmetic, the speech is perfectly coherent, and its coherence is part of the problem. If Juliet will be fourteen on Lammas Eve and the weaning happened eleven years ago on the day of the quake, then Juliet was weaned at something under three years old, a late but unremarkable weaning for the period, and her dead contemporary Susan, the Nurse’s own daughter, belongs to the same small cluster of dates. The numbers dovetail. A dramatist constructing a back-story could hardly do better, and a memory performing itself could hardly sound more authentic. Yet both readings of the passage, the documentary and the theatrical, run through the same narrow gate: everything in the speech is the Nurse’s. The calendar is hers, the feast is hers, the count is hers, and the earthquake is hers, located by her own telling in the garden of the Capulet house in Verona, where the dove-house over her head did the shaking. The play gives no narrator, no chorus, and no document to stand behind her figures. The Chorus of the Prologue, who might have vouched for a date, speaks instead of two households and a pair of star-crossed lovers and says nothing of years at all.

How Old Was Juliet When the Earthquake Struck, on the Nurse’s Own Figures?

Just under three years old: the Nurse’s two numbers, fourteen at Lammas Eve and eleven years since the weaning day, leave a gap of something less than three years between Juliet’s birth and the earthquake, which is precisely the span her anecdote describes, a toddler old enough to taste wormwood, take offence, and be remembered falling out with the dug.

Notice what this internal consistency does and does not do. It certifies that the fiction is well built. A playwright who wanted the audience to feel the solidity of the Capulet household’s past, its feasts and losses and remembered summers, achieved exactly that solidity, and the earthquake is the keystone of the effect: a public catastrophe demoted, in the Nurse’s scale of values, to a way of remembering the day a child was weaned. Editors and audiences have always felt the rightness of it. But internal consistency belongs to good fiction quite as much as to true testimony, and it certifies nothing whatever about London. The numbers count years inside Verona. To make them count years inside the author’s study, the whole apparatus of the clue has to be imported from outside the speech: the identification of the quake, the trust in the count, and the equation of the play’s present with the playwright’s. The speech supplies none of the three.

One more feature of the speech deserves emphasis before the argument built on it is examined, because it will return as evidence. The Nurse insists. That shall she, marry; I remember it well. Nay, I do bear a brain. Her memory is a performance of reliability, complete with a wager, fourteen teeth laid on a fact of which she owns only four, and the wager is the tell. The comedy of the scene depends on a woman whose certainty outruns her apparatus, who tells the same fact three times in twenty lines, who digresses to Susan’s grave and back, and whose precision is real but embodied, stored in sun and wormwood and a shaking dove-house rather than in any register. Shakespeare wrote her that way on purpose. A dating argument that treats her as a parish clerk has mistaken the genre of its only witness.

Two further details of the speech repay attention, because both will matter when the objections are weighed. The first is the weaning itself. Wormwood smeared on the breast was the period’s standard engine for breaking a child of the dug, a bitter herb laid on the nipple so that the taste itself delivered the refusal, and the Nurse’s account of it is a small masterpiece of remembered sensation: the sunny wall, the bitterness, the little fool’s outrage at the betrayal, the falling out with the dug as if the breast were a person who had given offence. A child weaned by this method at something under three was no infant; she was a walking, talking toddler with opinions, which is why the tantrum survived eleven years in the telling. The memory’s centre of gravity is the body, hers and the child’s, and the earthquake enters her story as the day’s public weather, the shaking that accompanied the private milestone. Popular memory works exactly like this in every century: the great event is retained as the frame of the small one. The second detail is Susan. The Nurse’s own daughter, born in the same season as Juliet, dead in infancy, surfaces in the speech for three lines and is dismissed with a proverb’s economy: she was too good for me. The bereavement is the speech’s emotional ballast, the reason its teller is a nurse in another woman’s house telling another woman’s daughter’s age. A memory carrying that much private freight is being offered by the play as a window into a life. It is not being offered as a chronicle, and its freight is precisely what makes it useless as one: every element in it exists because it mattered to the woman speaking, and a chronologer’s needs were not consulted.

There is also the calendar to notice. Juliet’s birthday falls on Lammas Eve, the night before the first of August, and Lammas, the loaf-mass, is the English harvest festival, the first fruits of the wheat gathered in English fields under an English name. Verona keeps no Lammas. The fiction’s household, in other words, already runs on transplanted English time before the earthquake is mentioned at all: the feast that fixes the heroine’s birth is as English as the tremor that fixes her weaning. That symmetry will return in the objections, because it quietly generalises them. If the Englishness of the earthquake makes it a document of the author’s present, the Englishness of Lammas must make the birthday one too, and Lammas yields no year whatever. The calendar of the speech is a servant’s calendar from the author’s own country, worn as naturally in Verona as a nurse’s apron. It proves the play’s world is built from English materials. It was never designed to prove when the building happened.

How Was the Year 1591 Built Out of Eleven Years?

The argument from the clue has a beautiful, fatal simplicity, and it is fairest to state it in its strongest form before touching it. Premise one: the Nurse’s line refers to a real earthquake, the one public tremor of the age that everyone in a London audience would recognise. Premise two: that earthquake is the great shock of 6 April 1580, felt across England and northern France, recorded in chronicles, sermons and a small library of pamphlets, and remembered for decades as the earthquake, the one an English writer would mean when no other was specified. Premise three: the Nurse’s count is exact, so the play’s present stands eleven years after the event. Conclusion: the play’s present is 1591, and since a topical playwright writes in his own present, the composition belongs to 1591. Every element of the argument has surface plausibility. There is no earthquake in the sources, neither in Arthur Brooke’s poem of 1562 nor in William Painter’s prose tale, so the detail is the dramatist’s own insertion, and an inserted contemporary reference is the classic fingerprint by which undated plays are dated. The Henry V chorus that salutes the Earl of Essex pins that play to 1599 by exactly this logic, and nobody doubts it.

The argument circulated for generations in the conditional mood, which is itself diagnostic. A Victorian editor’s formulation, repeated in the old variorum tradition, runs that if the Nurse’s remark be taken literally, the composition of the play must be referred to 1591. The whole case lives inside that conditional, and the conditional is the confession. Its most committed modern champion is Eric Sams, whose revisionist chronology of the canon, The Real Shakespeare, argues for a far earlier start to the playwright’s career than the standard record allows. Sams takes the line at face value and draws the consequence without flinching: the play’s inception may well be datable to early 1591, because the Nurse says it is since the earthquake now eleven years, and a substantial tremor had been felt throughout England on 6 April 1580. He adds, shrewdly, that the reference stands unchanged in the early texts, present in the quarto of 1597 as in its successor, which rules out the suspicion that the line drifted in late. For Sams the clue is a fixed point around which the early career can be rebuilt, and his larger thesis needs it: a Shakespeare already writing masterpieces in 1591 is a Shakespeare whose lost years fill with work.

Against Sams stands the settled practice of the standard editions, which print dates in the middle of the 1590s and treat the clue as colour rather than chronology. Brian Gibbons, in the Arden Second Series edition of 1980, reports the earthquake argument in his account of the date and declines to build on it, settling where the balance of evidence points, in the mid-1590s. Jill L. Levenson’s Oxford edition of 2000 dates the play to 1595 and handles the Nurse’s arithmetic as the period flavour it is. René Weis, in the Arden Third Series edition of 2012, likewise places composition in the middle of the decade. E. K. Chambers, whose Study of Facts and Problems of 1930 rebuilt Shakespeare chronology on a deliberate hierarchy of evidence, external documents first and internal allusions last and least, assigned the play to about 1595 on the combined record and gave allusion-hunting no authority to overrule it. The disagreement is therefore cleanly drawn. Sams: the line is a topical reference, topical references date plays, therefore 1591. The editorial consensus: a topical colouring is not a topical reference of the datable kind, the line belongs to a fictional memory in a fictional city, and the documents place the play half a decade later. The sections that follow adjudicate that disagreement piece by piece, and the adjudication goes against Sams, though not contemptuously: his reading of the line’s texture is right, and his inference from it is wrong, and the distance between those two judgements is the whole lesson of the clue.

One element of Sams’s case deserves separate handling, because it is the strongest card the clue holds and it still does not win the trick. The earthquake line is textually stable. It stands in the quarto of 1597 in old spelling, it stands in the fuller quarto of 1599, and it descends into the Folio text of 1623 with the rest of the speech; no early witness omits it, varies its number, or moves it to another mouth. Sams argues from that stability that the reference was load-bearing from the start, an original topical marker faithfully transmitted. The premise is sound and the conclusion does not follow. Stability proves the line was in the play by 1597, which the first quarto already proves about every line it contains, and it proves that actors and printers found nothing in it to mend. It is silent about 1591 precisely as the rest of the text is silent, and a topical allusion, of all things, is the element a revising playwright would be likeliest to refresh if it had dated: the line’s survival unchanged across a decade of reprinting is at least as consistent with its being timeless colour, a memory with no expiry, as with its being a dateline that everyone politely preserved. The card, examined, is blank on the side that matters.

The method that produced the sum has a pedigree worth a paragraph, because the clue was never an isolated fancy. When Edmond Malone set out in 1778 to ascertain the order in which the plays were written, the first serious attempt at a canon chronology, he worked with almost nothing: a handful of register entries, the publication dates of quartos, and internal allusions, passages where a play seemed to glance at a datable event. Allusion-hunting became the chronologer’s chief instrument for a century, and it scored real successes where the allusion was public, precise and pointed, a statesman’s expedition, a storm, a notorious trial. It also generated an undergrowth of false positives, since a sufficiently determined hunter can find an allusion to anything in a canon of this size, and the earthquake is the undergrowth’s most successful plant: public, precise-seeming, and attached to the most beloved play in the language. The bibliographical turn of the early twentieth century, the work of Pollard, Greg and McKerrow on how the early texts were printed and transmitted, shifted the discipline’s trust from what plays seem to say to what documents can be shown to be, and Chambers’s chronology institutionalised the shift. In that disciplinary story the earthquake clue is a fossil: a Malone-era instrument preserved in a Chambers-era world, still gleaming, no longer load-bearing.

The sum’s handbook career shows the fossilisation in progress. Older reference works frequently printed the play’s date as a range beginning in 1591, a form that looks like caution and is actually surrender: the early terminus is the earthquake’s arithmetic, the late terminus is the quarto’s, and the range between them confesses that the writer could not choose. A range is an honest notation when it marks the spread of a probability. It is a dishonest one when it yokes a document to a conjecture and presents the yoke as a finding, since the two ends of such a range have entirely different kinds of authority, one a fact of print and the other a trope of memory. Much of the clue’s modern survival lives inside those hyphens. Readers who would never assert 1591 outright absorb it as the range’s lower bound, and the lower bound does quiet work in the imagination, pulling the play toward the lost years, toward the apprentice Shakespeare, toward the romance of the masterpiece written by a nobody. Killing the clue therefore requires more than preferring 1595. It requires noticing that the defensible range does not begin in 1591 at all: it begins where the convergent evidence begins, in 1594, and the earthquake’s year is not a bound but a mirage at the desert’s edge.

It is worth pausing on why the 1591 sum proved so durable in textbooks and programme notes long after editors abandoned it. The sum is teachable in one breath, it flatters the detective instinct, and it converts the least documentable fact about the play into the most precise. A date of 1595, held as a probability inside a three-year window, is a harder thing to print on a classroom wall than a date derived from an earthquake in eleven steps a child can follow. The clue’s afterlife is a case study in how a weak argument with a strong shape outlives a strong argument with a weak one, and the series thesis of this study is exactly that the cliché of the known date dissolves, on inspection, into a judgement call among probabilities, in which the famous clue is the least load-bearing element on the table.

The Earthquake of 6 April 1580: What England Remembered

If the Nurse’s quake is the English quake, it is this one, and its profile explains both the clue’s plausibility and its danger. On Wednesday 6 April 1580, in Easter week, between five and six in the evening, an earthquake centred beneath the Dover Straits shook south-eastern England and the facing coasts of France and the Low Countries. Modern seismological reconstruction, including the study made for the Channel Tunnel and the British Geological Survey’s estimate, puts its magnitude at about 5.7 to 5.8, considerable for the region: it was felt from York to Paris, chimneys and pinnacles came down in London, masonry fell at Dover and Sandwich, a section of the cliffs near Dover dropped into the sea, and London’s pamphleteers reported deaths where falling stones struck. The city was at evening prayer in many churches when the shock came, a circumstance that fixed its moral interpretation in advance. This was received, instantly and almost universally, as a warning from God, and print moved faster than the aftershocks.

Thomas Churchyard, soldier and pamphleteer, had an account in print within two days: A Warning for the Wise, a Fear to the Fond, a Bridle to the Lewd, and a Glass to the Good, dated 8 April 1580, describing the churches, palaces and houses quivering and the people tossed as they stood. Arthur Golding, the great translator of Ovid, wrote A Discourse upon the Earthquake, weighing natural causes, winds and vapours imprisoned in the bowels of the earth, against the providential reading and finding the event too universal for any merely local engine. Abraham Fleming gathered the storm into a collection, A Bright Burning Beacon, published on 27 June 1580, reprinting Churchyard and Golding alongside reports by Richard Tarlton, the most famous clown of the age, Thomas Twyne, Francis Schackleton and others. The Privy Council ordered a form of prayer for Wednesdays and Fridays to avert the wrath the quake had threatened, to be used in every parish church in the realm. Gabriel Harvey, writing to Edmund Spenser in the familiar letters published the same year, turned the earthquake into an occasion for mocking the popular and academic methods of accounting for such tremors, which is to say that within months the event had passed from news into furniture: it was a thing clever men joked about in print, confident that every reader would know the reference without a date attached.

The sermon literature extended the pamphlets’ work across years. A shock received as a divine warning generated a preaching genre of its own, in which the earthquake served as the text for repentance long after the rubble was cleared, and Fleming’s collection itself is half devotional manual, its title promising a doctrine of signs and wonders and a discourse of the end of the world alongside the news. This devotional afterlife is the final reason the quake could not function as a chronometer even for its own generation. Events that become moral commonplaces detach from their years. Within five years the earthquake of 1580 was functioning in sermons the way the Flood functioned, as a standing illustration, dateless in use however dated in fact; within fifteen it was the Nurse’s kind of property, a landmark in the landscape of memory, available to any speaker of the playwright’s generation precisely because it belonged to all of them. The clue’s users ask the line to do the one thing its culture had spent fifteen years training the reference not to do: to point at a year and nothing else.

How Strong Was the 1580 Earthquake, and Where Was It Felt?

Strong enough to be the largest English earthquake of the century: modern estimates place it near magnitude 5.8, centred under the Dover Straits, felt across England into France, Flanders and Germany, damaging buildings in London, Dover, Sandwich and a scatter of Kentish and Essex towns, and bringing down pinnacles, chimneys and a length of the Dover cliffs.

For the playwright’s own generation the event had a biographical address. The boy who would write the Nurse’s speech was fifteen in April 1580, in his last years at the grammar school in Stratford, in a town close enough to the shock’s wide field for the tremor to be part of the talk of his schooldays; the reports Fleming gathered came from well beyond the capital, and the quake was a national experience in a way no purely London incident could have been. Harvey and Spenser were young Cambridge men when it struck; Tarlton was the reigning clown of the London stage; Churchyard was a veteran of sixty. The earthquake thus sat in the memory of the entire culture that would build the 1590s theatre, authors, actors and audiences alike, at exactly the depth a dramatist dreams of for a casual reference: universally held, personally felt, and requiring no gloss. A writer reaching for the landmark fifteen years later was not doing research. He was spending the common inheritance of his generation, and his hearers recognised the coin at once. That is the strongest form of the case that the line glances at 1580, and it is fully granted in this study. What the grant purchases is atmosphere. A coin that everyone holds can be spent in any year of the reign.

This is the earthquake as public memory, and the profile matters to the dating debate in two opposed ways. In the clue’s favour: an event of this saturation is precisely the kind that lodges in a play as an unobtrusive marker of time. People dated their lives by it; letters and diaries of the period reckon by the earthquake as by a regnal year, and a dramatist writing a nurse’s reminiscence could hardly find a more authentic nail to hang it on. The playwright himself, baptised at Stratford in April 1564, was fifteen years old, a schoolboy, when the ground moved; the tremor belonged to his own generation’s stock of shared memories, and its presence in the speech has the ease of a thing remembered rather than researched. Against the clue, the same profile cuts the other way. An earthquake this famous was famous for decades, and a famous event can be mentioned at any distance from itself. The Nurse’s line works in a play of 1595, or 1596, or 1597, exactly as well as in a play of 1591, because its function is to make her past feel solid, and a fifteen-year-old memory is no less solid than an eleven-year-old one; indeed the comedy of her certainty improves with distance. The pamphlets prove the quake was memorable. Memorable is the one property that destroys a terminus, since what everyone remembers, anyone may mention at any time.

The aftershocks of the event, in print and in the ground, kept the memory warm through exactly the years that matter to the debate. Lesser tremors followed on the first and second of May, close enough to renew the alarm and to keep the pamphlets selling. John Stow entered the earthquake in his Annals among the year’s public facts, between the doings of parliaments and the deaths of notables, which is to say the event passed into the chronicle record by which the literate nation organised its past. The official prayers ordered for Wednesdays and Fridays made the tremor a recurring appointment in every parish: months after the ground had settled, congregations were still rehearsing its meaning on a weekly rota, a state-sponsored mnemonic of remarkable efficiency. And Harvey’s letter to Spenser shows the final stage of the process, the stage at which a date-marker becomes permanent. Harvey does not report the earthquake; he uses it, as the occasion for a display of wit about the astrologers and natural philosophers competing to explain it, secure that the reference needs no introduction and no date. By the time the letter was a year old the quake had completed the journey from event to idiom. An idiom can be spent at any time. The dramatist who reached for the earthquake in the middle 1590s was reaching for the readiest landmark in the national memory, one his audience had been drilled in by prayer, pamphlet and jest for fifteen years, and the readiness of the landmark is the whole of its theatrical value and the whole of its chronological nullity.

Objection One: The Quake Is in Verona, and the Witness Is a Comic Nurse

The first objection is geographical, and it has never been answered. The earthquake of the speech happened, by the speech’s own staging, in Verona. The Nurse was sitting in the sun under the dove-house wall of the Capulet garden; the dove-house shook over her head; her employers were away at Mantua, which in the fiction’s geography is a journey of some twenty miles. Nothing in the passage reaches toward England. The dating argument therefore begins by silently relocating the event across the Alps and the Channel, on the theory that a Veronese nurse’s earthquake must be the playwright’s English one because the playwright was English. But the fiction is full of such transplantations running the other way, and they are the ordinary weather of Elizabethan drama: English law, English weather, English church bells and English measures fill the Italian cities of the stage without anyone proposing to date the plays by them. The tragedy itself supplies a control experiment in its fourth act, where the hired musicians, waiting in the Capulet house on the morning of the supposed corpse, bandy English ballad tunes with Peter, heart’s ease and the dump, in a passage of pure London street-music culture transplanted without a blink into Verona. No chronologer has ever dated the play by its ballads, though the tunes have histories as recoverable as the earthquake’s, because the ballads cannot be made to yield a year by addition. The earthquake is treated differently for one reason only: it arrives carrying a number, and the number can be added to a famous year. Strip the number away and the quake is exactly what the ballads are, the sound of the author’s own streets inside his invented city, evidence of workmanship and of nothing else. If every English colouring in Verona were a document, the play would be datable by a dozen incompatible clues at once. The quake is local colour inside the story, the day the dove-house shook; the argument needs it to be a dateline outside the story, and the text refuses the promotion.

The second objection is the witness. Even granting the relocation, the count belongs to the Nurse, and the Nurse is written, with great deliberateness, as a glorious unreliable narrator of her own past. Consider the credentials the scene itself issues her. She opens by wagering fourteen teeth on her daughter-by-charge’s age while owning four. She states her central fact three times, loops away to her dead Susan and back, loses her thread and recovers it with a formula, but, as I said, that functions as punctuation rather than as rigour. Her numbers elsewhere in the play have the same texture: sent in the morning to meet Romeo, she returns at noon complaining of the errand as a day’s labour, and the comedy of her errands is the comedy of elastic time. None of this makes her a liar, and the play clearly intends her Lammas arithmetic to be sound within the fiction. It makes her something more dangerous to a chronologer than a liar: a character, whose figures serve the scene’s warmth and comedy, not the archive. The leap from a character’s memory to a real event, the complication this study was charged to confront, is not a small step across a ditch. It is a change of jurisdiction, from the rules of fiction, where eleven years is the right-sounding span for a weaning remembered across a childhood, to the rules of evidence, where eleven years is a count that must be exact to be worth anything at all.

A third strand completes the objection. The number itself, eleven, is doing dramatic work that has nothing to do with 1580. Juliet must have been weaned at an age the audience can picture, old enough for the wormwood joke to land and for the tantrum to be remembered; she must now be on the cusp of fourteen, because the whole scene exists to open the marriage question; and the span between the two must feel like a childhood, long enough for the Nurse’s whole tenure in the household to have accumulated. Eleven years is the figure that satisfies all three pressures at once. A span of fifteen, which the mid-1590s date would require against 1580, satisfies them nearly as well, which is why no audience has ever felt the line jar; but a span chosen to satisfy the scene is evidence about the scene. Where the figure must be exact for one purpose and merely apt for another, method requires assuming the purpose it demonstrably serves. The apt purpose is on the page. The exact purpose is imported.

The Lammas symmetry, flagged when the speech was read, now does its work. Everything in the Nurse’s calendar is English: the harvest feast that dates the birth, the Susan who shares it, the earthquake that dates the weaning. If English colour in Verona were documentary, the calendar would have to be read whole, and read whole it yields no year at all, since Lammas comes round annually and Susan’s age is a fiction’s fact. The clue’s users therefore practise a quiet selection: of all the English materials in the speech, one item, the earthquake, is promoted to evidence, because one item happens to have a famous year attached in the world outside the play. The promotion is driven by the availability of the answer, not by any property of the text. This is the availability error in its purest scholarly form, and recognising it is worth more than the date itself, because the same selection operates wherever a colourful detail with a known year is allowed to outrank the dull documents that do not glitter. The dove-house and the loaf-mass stand on the page in identical type. Only one of them has been mistaken for a chronicle, and the mistake was made by the readers, not by the Nurse.

Her other numbers, elsewhere in the play, complete the portrait of the witness, and they are uniformly of the same manufacture. She swears, on her first entrance, by her maidenhead at twelve year old (1.3.2), a bawdy flourish whose own arithmetic, virginity kept or lost at twelve, has kept editors arguing for centuries without yielding a date for anything, and which is best read as a number of the same manufacture as her eleven years: sized for effect, warm with implication, and innocent of record. Sent at nine in the morning to find Romeo in the second act, she returns at noon and is mocked for a three-hour errand of yards, then complains of her bones as if she had crossed a county: her hours stretch and contract with her grievances. Even her refrain testifies against the clerkly reading. But, as I said, she repeats, at every return from digression, the formula of an oral storyteller re-finding the thread, and the speech’s tripling of its central fact, shall she be fourteen, that shall she, I remember it well, is the rhythm of memory performing itself for an audience of two women and a theatre. Oral memory is conservative of shape and careless of count; it keeps the wormwood and drops the register. A discipline that knows this about every other witness in every other archive has no business exempting the Nurse of Verona because her testimony is charming and her sum comes out even.

Objection Two: Other Earthquakes, Other Sums

Suppose, against the first objection, that the line does remember a real tremor and counts its years with intent. The sum still does not close, because 1580 is not the only earthquake the period offered, and eleven is not the only way to count. The later-date advocates have long pointed to William Covell’s Polimanteia, published in 1595, a wide-ranging treatise on the fall of commonwealths that also carries early testimony to the playwright’s reputation. Covell alludes to an earthquake of 1584, and the arithmetic that rescues the clue for the consensus is immediate: 1584 plus eleven years gives 1595, the very year the external evidence favours. The episode is a perfect emblem of the clue’s weakness. The same line, the same count, and a different candidate event produce the rival date with equal facility; a key that opens two locks eleven years apart is not a key but a coincidence in search of a door. Defenders of 1580 reply that the Dover Straits shock dwarfs any tremor of 1584 in the record, which is true, and irrelevant in the way the first objection predicts: the line cannot certify which quake it means, because in Verona it means a Veronese one, and in London it means whichever its hearers supply.

The 1584 rescue also deserves a fair statement of its own weakness, since fairness to the rival sums is this study’s method. The tremor Covell alludes to left almost no trace in the record beside the mountain of print that 1580 generated: no best-selling pamphlet, no order of prayer, no Harvey jesting about it to Spenser. Nobody supposes that an audience hearing the Nurse’s line would think first, or at all, of 1584. The rescue is therefore not a rival identification but a demonstration, performed with the opposition’s own pieces, that the identification step in the argument is underdetermined: choose the famous quake and the line yields 1591, choose the footnoted quake and the line yields the consensus. A method whose output depends on which input the operator prefers has measured the operator. That the rescue is artificial is precisely its value. It was never offered as history. It is a reductio, and a successful one, since no defender of 1580 has ever explained by what property of the text the famous quake is selected over the obscure one, other than the fame that equally disqualifies it.

What Earthquake Does William Covell’s Polimanteia Point To?

An earthquake of 1584, mentioned in Covell’s treatise of 1595: its importance to the dating debate is purely arithmetical, since eleven years counted from 1584 lands on 1595, showing that the Nurse’s figure can be made to certify the consensus date as readily as the early one, and therefore certifies neither.

The Italian candidate dissolves the sum from the other side. Northern Italy is earthquake country, and the great Ferrara earthquake of November 1570, a destructive sequence felt across the region and reported throughout Europe, stands in the fiction’s own landscape. The Ferrara shocks began in November 1570 and returned in waves for months, cracking palaces and churches, driving the Este court into gardens and improvised shelters, and filling the news networks of Europe with a catastrophe that needed no English pamphlet to be famous: it was the Italian earthquake of the century, in the very region, the Po valley and its bordering cities, where the play’s Verona and Mantua sit. A Veronese nurse whose dove-house shook has, inside her own geography, a candidate of the first magnitude, closer to her garden wall than Dover is by the width of two kingdoms. The English reading of the line must argue that the playwright reached past the famous quake of the play’s own landscape to the famous quake of his own, which is possible, since his hand stocked Verona with English things throughout. But the Italian reading needs no such argument, and the contest between a candidate that fits the geography and a candidate that fits the biography is a contest the text cannot referee, since the text mentions neither Ferrara nor Dover, only a dove-house. Steve Sohmer, in a study of the play’s time-riddles, has argued that the Nurse’s chronicle is exact inside the fiction if her quake is the catastrophe of 1570: the weaning falls in November of that year, Juliet, born on a Lammas Eve, passes her fourteenth birthday in the summer of 1582, and the scene’s mid-July conversation sits precisely where the Nurse’s figures put it. Whether or not Sohmer’s elegant construction is accepted, its exposure value is high. It demonstrates that the line can be wholly, even pedantically, accurate as fiction while telling the chronologer nothing: a sum that yields 1582 inside the story and 1591 outside it cannot be run in both directions, and the direction the clue’s users need is the one with no controls. If the earthquake is Verona’s, its year is the fiction’s business. If it is London’s, its year must be proved, not assumed, and the proof is what the argument was supposed to supply.

Finally there is the possibility the clue’s literature least likes to face: that the line remembers no single event at all. Earthquakes punctuate the chronicles of the period regularly, in England in 1551 and 1580 and by report in 1584, in Italy as a standing condition of the landscape, and a dramatist assembling a nurse’s memory from the common stock of how old people date their lives, by the year of the great wind, the year of the quake, the year of the dearth, might reach for an earthquake precisely because it is the generic landmark, needing no specific tremor behind it. On that reading the search for the right earthquake is a category error from the start, like searching Brooke’s poem for the right balcony. The possibility cannot be proved either, and that is the point: between a documented 1580, a reported 1584, an Italian 1570 and an invented tremor, the line itself casts no vote. A clue compatible with four answers, spread across twenty-five years, has excluded nothing, and evidence that excludes nothing confirms nothing.

One more elasticity remains to be priced in, because it quietly doubles the sum’s tolerance. Early modern counting of intervals was frequently inclusive, reckoning both endpoints into the tally, the convention by which a Friday crucifixion and a Sunday resurrection make three days. A speaker counting inclusively who says eleven years since an event may mean ten full years and a fraction by the modern exclusive count; a speaker rounding a famous interval to the nearest speakable figure may be a year out in either direction without any intention to deceive. The Nurse’s eleven therefore carries an honest tolerance of at least two years even on the clue’s own terms, before the choice among earthquakes is opened at all. A sum whose true output is a band, 1589 to 1593 on the 1580 identification with its tolerances honestly stated, has already surrendered the quality that made it attractive, its precision. What remains is an allusion consistent with the early 1590s and equally consistent, once Covell’s candidate is admitted, with the middle of the decade. Consistency with a conclusion is the weakest relation in evidence. The documents that fix the window do not merely consist with 1595; they exclude 1591. The difference between those two verbs is the difference between the consensus and the clue. For completeness, the conditions under which an earthquake could date a play can be stated, since they show by contrast what is missing here. The event would need to be unique in its generation, so that no rival candidate could be produced. The count from it would need to be uttered in the author’s own voice, a chorus, a prologue, a datable compliment, not in the mouth of a character whose reliability the drama itself ironises. And the resulting year would need to collide with no document. The Nurse’s line satisfies none of the three conditions. Its event is one of at least three, its count is a comic servant’s, and its year is contradicted by the title page of the very book that preserves it.

The Bounds That Actually Hold: Source, Quarto, and Company

Strip the earthquake away and the dating of the play rests on a small set of hard objects, none of them glamourous and all of them stronger than the clue. The floor is set by the sources. The tragedy could not exist before Arthur Brooke published The Tragicall History of Romeus and Juliet in 1562, nor before William Painter’s prose version in the second volume of The Palace of Pleasure in 1566 and 1567 gave the tale its second English form; the playwright worked from Brooke above all, and the debt is verbal and structural throughout. A floor of 1562 excludes nothing that matters, since the playwright was not yet born when Brooke’s poem appeared, but it performs one service: it reminds the chronologer that the tale’s matter is a generation older than any candidate date, and that the earthquake, absent from both sources, entered with the dramatist whenever he came.

The ceiling is set by print and performance together. In 1597 John Danter printed the first quarto of the play, the short early text whose title page announces that it had been often, and with great applause, publicly played. The full study of the first quarto of 1597 and its ‘bad quarto’ label belongs to another article in this series; here the quarto matters as a bound, and the bound is tighter than its own imprint. The title page names the playing company as the Lord Hunsdon’s Men. That name had a short life. Henry Carey, Lord Hunsdon and Lord Chamberlain, died in July 1596; his son George Carey, the second Lord Hunsdon, held the company under his own title until March 1597, when he in turn became Lord Chamberlain and the company resumed its Chamberlain’s Men style. A play billed as often played, with great applause, by a company that bore the Hunsdon name only between July 1596 and March 1597 was necessarily in performance, and in successful performance, inside that window or hard against it. Print followed in 1597. The earthquake line stands in this quarto in substantially its familiar form, as Sams himself stresses, so the clue was part of the play as first printed; what the quarto cannot do is point backward from its applause to a first night six years earlier.

Between Which Dates Must the Play Already Have Existed, on the Title Page Alone?

Between July 1596 and March 1597 the playing company named on the first quarto’s title page, the Lord Hunsdon’s Men, existed under that exact style, and the quarto bills the tragedy as often played with great applause by them, so the play was a going concern on the stage inside that window.

The printing itself tells the same story of a living repertory piece. The 1597 quarto was printed by John Danter, with part of the sheets from Edward Allde’s press, for the bookseller Cuthbert Burby, and it was never entered in the Stationers’ Register, the profile of a play text that reached print through the theatre’s economy rather than through an author’s study. Its title page is a playbill in prose: the tragedy had been often, with great applause, publicly played. Formula though the boast is, the formula has a grammar, and its grammar is present-perfect: the applause is accumulated, recent enough to sell a book on, and attached to the company of the hour. Two years later the second quarto of 1599 advertised itself as newly corrected, augmented and amended, the fuller text superseding the short one while the play was still warm in the repertory. Print history of this shape, a short text cashed in at the height of applause and a long text following within two seasons, is the biography of a current success. It is not the biography of a manuscript that had waited since 1591 for a company to notice it.

The second printed bound arrives a year later. Francis Meres, in the comparative discourse of wit called Palladis Tamia, entered in the Stationers’ Register in September 1598, names the play among the tragedies for which its author was already counted excellent, in company with Richard the Second, Richard the Third, Henry the Fourth, King John, Titus Andronicus and others. The list is the play’s first fame document and is treated in full in the study of Meres’ 1598 list and the play’s early fame; for the chronology its force is simple. By the autumn of 1598 the tragedy was not merely written but canonical enough to anchor a reputation, and Meres lists it without any flavour of novelty. The list’s company also testifies, quietly, to the bound’s firmness. Meres names the tragedy in a set of six, alongside works whose dates the documents fix in the middle and later 1590s, and his survey is a snapshot of a reputation as it stood in 1598, not an excavation of the decade’s beginnings: he is cataloguing the excellences of a living author at the height of his first fame. A play written in 1591 could of course still be excellent in 1598. But the snapshot’s whole texture, a new name weighed against the ancients and found sufficient, belongs to a moment of arrival, and arrivals are not usually seven years old. Between the Hunsdon window and Meres, the documentary picture is of a play established, applauded and in print. Nothing in the record reaches back before 1596, and the shape of the record, a repertory success surfacing into print and canonisation in three quick years, is the familiar shape of a recent play, not of a six-year-old one quietly waiting for its fame. Scholars assembling the title pages and registers that bound the play will set the 1597 quarto beside its 1599 successor and the Meres page in a single view, and a slide-based tool such as the ReportMedic PPT Viewer serves that collation well: the three documents, read together, are the skeleton on which any honest date must hang.

The Mid-1590s Case: Verse, Company, and Repertory

Between the floor of the sources and the ceiling of the Hunsdon window stands the dating proper, and it is done, as Chambers insisted it should be, by the convergence of the internal evidence that can be tested rather than the allusion that can only be imagined. The testable internal evidence is the verse. Across the second half of the twentieth century, editors and chronologers built a battery of stylistic measures, proportions of rhyme and of prose, rates of enjambment and of feminine endings, the pause-patterns within the line tabulated by Ants Oras, the colloquialism counts that track how far the verse relaxes toward speech, and on every one of these measures the tragedy keeps the same company. Its neighbours are Richard the Second and A Midsummer Night’s Dream, with Love’s Labour’s Lost and King John in the same stratum: the lyric, rhyming, sonnet-haunted cluster of the middle 1590s, when the playwright’s line was at its most musical and his structures at their most patterned. The Oxford chronology of Stanley Wells and Gary Taylor, assigning dates by the whole battery, places the play in 1595, and the standard editions converge on the same neighbourhood: 1594 to 1596 is the window the verse keeps indicating, whichever measure is run.

The kinship with A Midsummer Night’s Dream deserves a moment of its own, because the two plays are the decade’s most instructive pair. The Dream burlesques, in the mechanicals’ Pyramus and Thisbe, the precise machinery the tragedy had made overwhelming: separated lovers, a wall, a moonlit assignation, a tomb-side suicide doubled by error. Burlesque of that intimacy presupposes a target its audience knows in its bones. Whether the comedy followed the tragedy by a season or grew beside it in the same repertory winter, the two works breathe one theatrical air, share the sonnet’s climate, the fairy-haunted night, the doctrine that young love is a kind of weather, and editors from Chambers onward have dated them as neighbours for that reason as much as for any statistic. A tragedy of 1591 would make the Dream’s burlesque a revival-house joke at a four-year-old success, possible in principle and supported by nothing; a tragedy of 1595 makes the pair what they read as, twins of one season’s imagination, the same hand treating the same story once as doom and once as farce. Chronology should notice when two works insist on being read together. These two insist.

Which Plays Stand Nearest the Tragedy in the Verse Tests?

Richard the Second and A Midsummer Night’s Dream stand nearest, with Love’s Labour’s Lost in the same stratum: on rhyme proportion, enjambment, feminine endings and pause-pattern the tragedy consistently groups with the lyric plays of the middle 1590s rather than with the work of the decade’s opening years.

The tests deserve a word about their mechanics, since their authority is statistical and statistics invite suspicion. No single measure dates anything. A high proportion of rhyme, taken alone, might mark a young writer’s habit or an old genre’s demand; the tragedy rhymes heavily because it is a lyric tragedy, and genre confounds chronology everywhere in the canon. The method’s strength is convergence under independence. Rhyme proportion falls steadily across the documented career as the verse matures toward the great tragedies’ spareness. Feminine endings, the unstressed eleventh syllable that loosens the line’s gait, rise just as steadily. Enjambment, sense running over the line ending, climbs; the mid-line pause migrates; prose claims a growing share of the whole. Each trend was established on the plays whose dates the documents fix, the anchors at either end of the career, and each is a ratchet that genre alone cannot explain, since the trends run within genres as well as across them. When an undated play is fed to the battery, its measures do not vote once; they vote five and six times, on evidence drawn from different strata of the verse, phonetic, metrical, syntactic. For the tragedy the votes cluster with a consistency that chance does not produce: the profile is that of the middle 1590s, neither of the apprentice work Greene sneered at nor of the Jacobean manner to come. Chambers built his chronology on exactly this principle, that a date should rest on the concurrence of independent witnesses, documentary where possible, stylistic where necessary, and it is the principle, as much as any figure, that the earthquake offends. The clue is one witness, interested, uncorroborated, and testifying about itself. The battery is a panel. Panels outrank witnesses, in chronology as in law.

The internal world of the play corroborates the verse. Its fabric is woven from the sonnet culture of the decade’s middle years: a Prologue in sonnet form, a second chorus in sonnet form, the lovers’ first exchange built as a shared sonnet, Romeo’s early speeches steeped in the Petrarchan conventions the opening scenes then school him out of. The parallel fashion examined in the study of the 1590s sonnet vogue and the play’s verse belongs there in full; here the point is chronological. The vogue for printed sonnet sequences that followed the posthumous publication of Sidney’s Astrophil and Stella in 1591, Daniel’s Delia in 1592 and the swarm of imitators behind them had made the sonnet the period’s dominant lyric instrument by the middle of the decade, and the play’s easy, ironising mastery of the form reads as the work of a dramatist writing inside that fashion at its height. A play of 1591 would anticipate the vogue rather than ride it, quoting a craze in the year of its first kindling with the fluency of a form already ripe for parody. Fluency is not proof, but it is the kind of fact that accumulates, and it all accumulates in one direction.

The institutional record pushes the same way. A composition in 1591 needs a stage, a company and an occasion, and the early 1590s supply none that fit. The London theatres were closed by plague for long stretches between the summer of 1592 and 1594; the companies were dissolving, touring and reforming; the playwright’s own whereabouts in those years are the famous lost years of the biography, with his name attached to no company that can be documented. The company that demonstrably owned the play, the Chamberlain’s organisation in its Hunsdon interlude, was formed in 1594, and the tragedy reads in every fibre as a Chamberlain’s play: written for Richard Burbage’s rising powers, for the clown Will Kempe, whose name survives in the second quarto’s stage directions, for a company investing in the new lyric tragedy its chief writer was inventing alongside the Dream’s fairy machinery and Richard the Second’s ceremonial verse. The print history of the sonnet vogue lets the fashion argument be stated with dates instead of impressions. Sidney’s Astrophil and Stella, composed in the early 1580s, reached print only in 1591, in the pirated edition that opened the flood; Daniel’s Delia followed in 1592, Constable’s Diana the same year, and the sequences multiplied through 1593, 1594 and 1595 until the sonnet had become the decade’s signature small form, the thing a young writer proved himself on and a maturing one parodied. The tragedy’s relation to that timeline is exact. Its sonnets are not the work of a man guessing at a fashion from its first pamphlet; they are the work of a man for whom the form is already an institution, supple enough to be shared between two speakers as dialogue, solemn enough to carry a Prologue, and familiar enough that Rosaline’s Petrarchan Romeo can be mocked from inside the convention by a friend who speaks prose. Institutions of that ripeness are not available in the year of a form’s first printing. They are available in the middle of the decade, when the sequences were stacking up on the bookstalls and every theatregoer had the sonnet’s turns by ear. The play’s sonnet craft is therefore a datable climate, and the climate is 1594 to 1596, the same window the verse statistics indicate and the same window the title page permits.

Occam’s razor is a blunt instrument, but it is the right shape here. The 1595 hypothesis requires a playwright at the documented centre of a documented company writing the play his verse says he wrote. The 1591 hypothesis requires an undocumented premiere, an undocumented company, an unrecorded six-year stage life ending in sudden print, and a number in a comic speech to be a chronicle entry. Chronology, like criticism, should prefer the hypothesis that needs less rescue.

The Career Frame: Where the Play Falls in a Documented Life

A date is also a placement inside a life, and the life, though thinly documented, has a frame into which the two candidate years fit very differently. The frame’s uprights are these. In September 1592 Robert Greene’s posthumous Groatsworth of Wit attacked an upstart crow, beautified with the company’s feathers, who supposed he could bombast out a blank verse and, in a parody of a line from the third part of Henry the Sixth, was named in everything but letters: Shake-scene. The attack proves that by the autumn of 1592 the Stratford man was a London playwright successful enough to be resented, with the Henry the Sixth plays behind him. In 1593 and 1594 his name went onto the title pages of Venus and Adonis and The Rape of Lucrece, dedicated to Southampton, the print debut of a writer presenting himself as a poet. In 1594, after the plague years had dissolved the old companies, the Lord Chamberlain’s Men formed with the playwright among its sharers, the arrangement that would hold, through the Hunsdon interval, for the rest of his career. In 1596 and 1597 the tragedy surfaced in performance and print under that company’s patronage. In 1598 Meres canonised it. The frame, be it noted, contains no event of 1591 at all: that year falls in the documentary silence before Greene’s snarl, in the lost years proper, where the early daters must place their masterpiece.

What Does Greene’s Groatsworth of 1592 Contribute to the Play’s Dating?

It fixes the career’s frame without touching the play: Greene’s attack proves its target was already a resented London playwright by September 1592, with early histories behind him, so a tragedy of 1591 is not impossible in principle, merely unsupported, since no document connects the young Shake-scene to this play, this company or this story.

The frame’s testimony is about shape. A career that runs apprentice histories, a resentful explosion in 1592, two courtly poems offered as first fruits in 1593 and 1594, a sharer’s place in a new company, and then the lyric cluster of the middle decade, Richard the Second, the Dream, King John, the tragedy of the Veronese lovers, is a career with a legible gradient, each stage preparing the next. The tragedy’s position in that gradient, at the point where the poem-writer and the company man fuse, is what the mid-1590s date asserts. The 1591 date asserts a different shape: a spike, in silence, before the gradient starts, a tragedy of the first rank preceding the poems that call themselves first heirs and the histories that provoked Greene, after which the spike’s author subsides into the documented apprenticeship. Biographies can contain spikes, and genius is not obliged to be gradual. But chronology is a branch of evidence, not of romance, and between a gradient that fits every document and a spike that fits one contested line, the evidential choice is not difficult. The lost years are lost because no paper survives from them, and paperlessness is a property of the archive. It is not a permission slip for masterpieces.

What Hangs on the Date: Two Different Shakespeares

The fight over eleven years is not pedantry, because the two candidate dates build two different playwrights. Put the tragedy in 1591 and it becomes a prodigy of the lost years: a love tragedy of fully achieved lyric power written before Venus and Adonis gave its author a printed name, before the history plays of the early decade had run their course, by a twenty-seven-year-old whose company, whose theatre and whose Burbage cannot be produced. On that chronology the play’s mastery is an anomaly demanding explanation, and Sams and the early daters supply one by redrawing the whole career, pulling the canon back and compressing the apprenticeship into invisibility. Put the tragedy in 1595 and it becomes the keystone of the best-documented annus mirabilis in English drama: the year of Richard the Second’s ceremonial grief and the Dream’s wedding machinery, when the Chamberlain’s Men were new, the Globe not yet built, and their house poet was converting the sonnet culture of the moment into drama at a rate no rival matched. On that chronology the play’s mastery is the centre of a pattern, each neighbouring work illuminating the others, the Dream’s mechanicals burlesquing the lovers’ deaths in the same theatrical breath that first staged them in earnest.

What Would a Date of 1591 Do to the Shape of the Career?

It would detach the tragedy from every documented anchor of the playwright’s working life: before his first printed poem, before the formation of the company that owned the play, before the plague closures that scattered the early decade’s companies, leaving a masterpiece with no stage, no company and no record for its first six years.

The reading consequences are as large as the biographical ones, and quieter. Date the play 1591 and its criticism acquires a standing excuse: the double time-scheme, the creaking of the last act’s machinery, the sententious friar, can all be patronised as the marks of a young writer, forgiven in advance by the calendar. The play becomes a promise. Date it 1595 and the excuse evaporates. This is the work of the dramatist of Richard the Second in the same season, and its choices must be judged as choices: the compression of Brooke’s months into days is not youthful impatience but a deliberate engine, the sonnet architecture not a beginner’s scaffolding but a master’s quotation of the age’s lyric system, the Nurse not a lucky sketch but a constructed instrument, memory included. Editors feel the difference in their bones. An early play is edited indulgently, its oddities ascribed to the apprentice or the transmission; a mature play is edited strictly, its oddities defended as intentions. The two dates thus print, in effect, two different texts between the same covers, and the earthquake is the only witness ever called for the indulgent one.

A last stake deserves naming, because it explains the clue’s very existence. The earthquake speech is in the play because of an artistic decision that the chronology can watch being made. In Brooke’s poem, the source the dramatist quarried line by line, the heroine is a girl of sixteen, ripening toward the eighteen of Painter’s prose; the play famously lowers her to the edge of fourteen, a change whose debate belongs to another article in this series. A heroine of sixteen needs no certification of age; her marriageability is visible. A heroine of not-yet-fourteen needs her age established, defended and dramatised, in a household where a mother proposes a bridegroom and a father counts summers. The Nurse’s speech exists to do that establishing. Its calendar, its feast, its weaning, its earthquake, are the machinery by which an invented girl’s invented age is made solid enough to bear the plot’s weight, and the machinery was built because the design required it, at whatever date the design was executed. The clue, in other words, is a by-product of the play’s boldest alteration of its source, generated by the needs of the fiction at the point of maximum fictional pressure. Evidence manufactured under that kind of pressure, to certify an age that exists only inside the story, is the last material in the play from which to extract a fact about the world outside it. The dating debate has spent four centuries reading the certificate as if it were the register.

The repertory consequences run parallel. A play of 1595 enters a London theatre rebuilding itself after the plague years, hungry for repertory that could hold a stage the way this one’s applause, by the title page’s testimony, held it. Its success helps explain the second quarto of 1599, newly corrected, augmented and amended, and the durable place of the tragedy in the company’s stock. A play of 1591 enters a theatre about to be shut by plague, toured into the provinces by collapsing companies, and then, on the evidence, leaves no trace until it resurfaces under a patron’s name that did not exist until 1596. One document of the early 1590s presses the point with unusual directness. In 1593 the playwright published Venus and Adonis, his first appearance in print, and dedicated it to the Earl of Southampton with a striking phrase: he called the poem the first heir of my invention, vowing to employ his idle hours thereafter on graver labour. The phrase has been read narrowly, as a claim about printed poetry rather than plays, and the reading is possible; dedications are courtly instruments, not affidavits. But on the 1591 hypothesis the author of the most accomplished love tragedy of the age, already two years applauded on the stage if the early daters’ applause is to precede its print, introduces himself to his patron as a beginner whose first offspring is a narrative poem. On the 1595 hypothesis the phrase is unremarkable: a player-poet of 1593, known in the theatre, unknown in print, offers his first book and promises better. The dedication does not disprove the early date. It prices it. Every year the tragedy is pulled back from 1595, the more of the documented record has to be explained away as coyness, anomaly or loss, and explanations of that kind are a currency in which a sound chronology should not trade. Every hypothesis about the lost years must leave some evidence unaccounted for; the early date leaves nearly all of it. That asymmetry, and not any single objection, is what the consensus rests on. The mid-1590s date is not certain. It is merely the only date at which all the evidence can be true at once.

The InsightCrunch Dating Clue: Weighing Every Strand

The artifact of this study gathers the whole case into one weighing, strand by strand: what each piece of evidence would prove if it stood alone, what it actually shows on inspection, and the weight it can bear in the final account. Its lesson is written into its shape. The strand with the loudest reputation carries the least weight, and the strands that do the work are the dull ones: a company’s short-lived name, a register entry, a set of verse statistics, a printer’s year.

One discipline governs the weighing and should be stated before the table is read: strands accumulate only when they are independent. The earthquake and the eleven years are not two strands but one, a single remark whose event and count stand or fall together, and the clue’s literature has sometimes counted them twice, once as a topical allusion and once as a numerical corroboration. The documentary strands, by contrast, are genuinely independent of one another and of the verse: a patron’s name on a title page, a register entry by a clergyman compiling a survey of wit, a printer’s year, and a statistical profile of the verse were produced by different hands for different purposes, and their convergence on one window is therefore worth more than any ten agreements within a single speech. The table grades each strand by the only question that matters: how much of the final date would fall if this strand were removed? For the documents, a wall of the window falls. For the earthquake, nothing falls at all.

Evidence	What it would prove alone	What it actually shows	Weight it can bear
The Nurse’s eleven years, read against 1580	Composition in 1591	A famous quake remembered inside a Veronese fiction, by a comic witness, at an unstated remove	Very low: colour, not chronology
The Nurse’s eleven years, read against Covell’s 1584	Composition in 1595	That the same figure can be fitted to the rival date with equal ease	None: the fit cancels itself
The Ferrara earthquake of 1570 inside the fiction	A story-world date near 1582	The line can be internally exact as fiction while dating no manuscript	None for composition; high for the fiction’s craft
Brooke (1562) and Painter (1566 to 1567)	A floor under all dates	The tale was available a generation before any candidate year	Low: excludes only the impossible
The first quarto, printed 1597	Existence by 1597	A performed play in print, its applause already proverbial on the title page	High as a ceiling
The Lord Hunsdon’s Men name on that title page	Performance between July 1596 and March 1597	The play was a going repertory success inside that short window	Very high: the tightest documentary bound
Meres’ Palladis Tamia, 1598	Canonical standing by 1598	The tragedy already anchored its author’s tragic reputation	High: fame of this kind takes time
The verse tests	Kinship with Richard the Second and the Dream	Composition in the lyric stratum of the middle 1590s	High: convergent and repeatable
The sonnet vogue inside the play	Fluency in the form at its height	A dramatist riding a fashion of the mid-decade, not anticipating it	Moderate: corroboration, not proof
Plague closures and company history, 1592 to 1594	Difficulty for an early premiere	No documented stage or company fits a 1591 first performance	Moderate: silence that tells

Read down the weight column and the pattern the table exists to expose stands plain: the evidence sorts itself inversely to its fame. The earthquake clue, the one element of the case that every schoolroom knows, can bear almost nothing; the patron’s name that a title page happened to freeze for eight months, which no schoolroom mentions, can bear almost everything. That inversion is the real finding of the dating debate, and it generalises across the chronology of the whole canon, where the dates that hold are held by registers, title pages and repertories, and the dates that glitter are held by allusions. The InsightCrunch dating clue is offered as the quick reference for the case: ten strands, their claims, their audited values, and their weights, in one view. The weighing method travels, and its portability is part of its point. The canon is full of beloved clues of the earthquake’s family: a compliment to a living favourite, a storm at sea, a topical joke whose target had a datable season in the sun. Each deserves the table’s three questions before it is allowed to move a play. What would the strand prove if it stood alone, and does the conclusion it offers match the conclusion it was found by seeking? What does it actually show, once its speaker, its genre and its geography are restored to it? And how much of the final date would fall if the strand were withdrawn, which is the only honest measure of weight? Run the Essex compliment in Henry the Fifth through the three questions and it survives them, public, precise, pointed and unique, which is why nobody litigates 1599. Run the earthquake through them and it fails each in turn. The difference between the two allusions is the whole discipline of dating in miniature, and it is a difference of kind, not of confidence: one is a reference, the other is a reminiscence, and chronology that cannot tell references from reminiscences will date plays by their weather.

The Verdict: A Clue That Dates a Memory, Not a Manuscript

The earthquake clue survives the audit diminished but not disgraced, and its proper office can now be named exactly. It establishes that the play’s Verona is stocked from the playwright’s England, that a tremor which shook his own schoolboy generation had passed into the common stock of remembered landmarks, and that the Nurse’s past is built from the materials real pasts are built from. It colours the play’s world and certifies its maker’s ear. What it cannot do is the one thing it is famous for doing. It cannot fix a year, because its event is movable among at least three earthquakes and possibly none, because its count is a character’s and is chosen for the scene it serves, because the fiction it inhabits keeps its own calendar in which the sum points to 1582, and because its conclusion, 1591, is contradicted by the title page, the register, the verse and the institutional record acting in concert. The consensus date, composition in the middle of the 1590s, most probably 1595, within an honest window of 1594 to 1596, deserves a confidence grade of high for the window and moderate for the single year: the bounds are documents, the verse is convergent, and only the false precision of a single year outruns the evidence. The earthquake, graded as a dating instrument, earns the lowest mark on the table. As a window into how memory works in the play, and how chronology fails when it is built on charm, it earns the highest.

The honest citation, then, is the one the standard editions in effect print: Romeo and Juliet, about 1595. The little word about is not a failure of nerve. It is the finding, the exact shape of the knowledge the evidence supports, a centre with margins, documents at the edges and judgement in the middle. Reference works that print 1591 without a qualifier are not reporting a rival finding; they are reporting the clue, fifteen removes from the dove-house, after a game of whispers in which a character’s fond approximation hardened into a fact. The remedy is not to banish the earthquake from the classroom but to teach it whole: the line, the speech, the shaking dove-house in Verona, the pamphlets of 1580, the rival tremor of 1584, the eight months of a patron’s name, and the moment where the sum dissolves and the documents begin. Taught whole, the clue becomes the best introduction the play offers to the difference between evidence and arithmetic, which is to say, to the difference between knowing a date and manufacturing one.

Frequently Asked Questions

Q: What is the InsightCrunch dating clue, and what pattern does it expose?

The InsightCrunch dating clue is this study’s weighing of every strand of evidence offered for the play’s date, set out in a single table: what each strand would prove alone, what it actually shows, and the weight it can bear. The pattern the weighing exposes is an inversion. The famous strand, the Nurse’s earthquake, carries the least weight of the ten, because its event is movable, its count is a character’s, and its conclusion collides with the documents. The obscure strands carry the most: the eight-month life of the Lord Hunsdon’s Men name on the 1597 title page, Meres’ register entry of 1598, and the convergent verse tests that group the play with the lyric works of the middle 1590s.

Q: When did the Dover Straits earthquake happen, and what did Londoners feel?

The Dover Straits earthquake happened on Wednesday 6 April 1580, in Easter week, between five and six in the evening. Londoners felt a shock that modern seismology estimates near magnitude 5.8, centred beneath the Straits: churches, palaces and houses quivered, people were tossed as they stood or driven from their seats, and pinnacles and chimneys came down in the city. The tremor was felt from York to Paris and across the Low Countries, walls fell at Dover and Sandwich, and a length of the Dover cliffs dropped into the sea. Because the shock came during evening prayer in many churches, it was received at once as a divine warning, and the Privy Council ordered a special form of prayer for the whole realm.

Q: Who wrote about the 1580 earthquake in print, and why does their pamphlet storm matter to the dating debate?

Thomas Churchyard wrote about it first, with an account in print two days after the shock, followed by Arthur Golding’s Discourse upon the Earthquake, Abraham Fleming’s collection A Bright Burning Beacon of June 1580, which reprinted Churchyard, Golding, the clown Richard Tarlton and others, and Gabriel Harvey’s joking letter to Edmund Spenser. The pamphlet storm matters to the dating debate because it proves how completely the event saturated public memory: the quake became a national date-marker, the kind of landmark by which ordinary people reckoned their lives. That saturation cuts both ways for the clue, since an event everyone remembered could be mentioned with effect at any later distance, in 1595 as readily as in 1591.

Q: Why does the 1591 sum quietly assume that the play’s present is the author’s present?

The 1591 sum assumes it because the addition only works if the year in which the Nurse speaks is the year in which the playwright writes. Inside the fiction, the Nurse’s present is Verona’s present, a constructed moment eleven years after her earthquake and a fortnight before Juliet’s fourteenth birthday; nothing on the page equates that moment with the calendar year of composition. The assumption is smuggled in from the practice of topical allusion, where a playwright saluting a living statesman does write in his own present. But the earthquake is not a salute to a living event; it is a memory inside a story, and memories in stories keep the story’s time, not the author’s. Grant the assumption and the sum runs; refuse it, as method requires, and no sum exists.

Q: What would an earthquake in 1584 do to the Nurse’s eleven years?

An earthquake in 1584 would turn the Nurse’s eleven years into a certificate for 1595, the very date the external evidence favours. William Covell’s Polimanteia, published in 1595, alludes to a tremor of 1584, and later-date advocates have pointed to it for generations: count eleven years from 1584 and the sum lands exactly where the title page, the register and the verse tests already point. The episode does not prove that the Nurse meant the 1584 shock, which was a far smaller event in the record than the Dover Straits earthquake. It proves something more damaging to the clue itself: the same line and the same count can be fitted to rival dates with equal ease, and a piece of evidence that fits every conclusion fits none.

Q: Could the Nurse’s earthquake be the great Italian earthquake of 1570?

The Nurse’s earthquake could indeed be the Italian catastrophe of November 1570, and within the fiction that identification works elegantly. The Ferrara earthquake of that month devastated a city within the play’s own northern Italian landscape and was reported across Europe. Steve Sohmer has shown that if the weaning is pinned to November 1570, the Nurse’s whole chronicle locks into place: Juliet, born on Lammas Eve, turns fourteen in the summer of 1582, and the scene’s conversation falls in the mid-July where her figures put it. The construction dates the story, not the play, since a composition year of 1581 is impossible for a dramatist baptised in 1564. Its value to the debate is diagnostic: a line can be exactly right as fiction and worthless as chronology at the same time.

Q: What does the shaking dove-house tell us about where the Nurse’s earthquake happened?

The shaking dove-house tells us that the Nurse’s earthquake happened, in the only geography the play supplies, in the Capulet garden in Verona. She was sitting in the sun under the dove-house wall, wormwood on her dug, when the structure over her head shook and she coined the joke of the scene: shake, quoth the dove-house. Her employers were at Mantua, twenty miles off in the fiction’s map. The dating argument must relocate this Veronese tremor to England before its arithmetic can begin, and the relocation is an act of the critic, not of the text. Elizabethan Verona is full of transplanted English colour, bells, measures, weather and law, and no other piece of that colour is promoted into a document. The dove-house is the clue’s first and plainest objection, standing in the middle of the evidence itself.

Q: Does the earthquake line appear in the first quarto as well as the second?

The earthquake line does appear in the first quarto of 1597, in substantially its familiar form and old spelling: ‘tis since the earthquake now eleven years. The Nurse’s speech is compressed in that shorter text, as the whole play is, but the earthquake reference survives the compression, a point Eric Sams presses in arguing that the reference stands unchanged across the early versions. The survival matters in two directions. It confirms that the clue belongs to the play as first printed and was no later refinement slipped into the fuller text of 1599. It also confirms that the line was already serving its theatrical purpose, anchoring a comic reminiscence, in the version the Hunsdon’s Men played, whatever year the play was written.

Q: What sets the earliest possible year for the play’s composition?

Arthur Brooke’s poem sets the earliest possible year: The Tragicall History of Romeus and Juliet, published in 1562, is the tragedy’s principal source, quarried verbally and structurally from the Prologue to the tomb. William Painter’s prose retelling in The Palace of Pleasure, in 1566 and 1567, supplies a second English form of the tale and a second floor. Neither bound bites in practice, since the playwright was a small child when Brooke appeared and unborn as a writer for decades after, but the floors perform one service for the dating debate: they show that the tale’s matter, the feud, the potion, the tomb, is a generation older than any candidate date for the play, while the earthquake, present in neither source, is the dramatist’s own insertion and therefore the one detail that belongs wholly to his moment, whenever that moment was.

Q: How long did the company keep the name that the 1597 title page gives it?

The company kept the name on the 1597 title page, the Lord Hunsdon’s Men, for roughly eight months, from July 1596 to March 1597. Henry Carey, first Lord Hunsdon and Lord Chamberlain, died in July 1596, and his son George Carey succeeded him as patron though not yet as Chamberlain; for that interval the players were Hunsdon’s Men. In March 1597 George Carey became Lord Chamberlain in his turn, and the company resumed the style of the Lord Chamberlain’s Men. The brevity of the name is what makes it the tightest bound in the play’s chronology: a title page billing the tragedy as often played with great applause by that company fixes a flourishing stage life inside a window narrower than a single theatrical season on either side.

Q: If the play had been written in 1591, which company could have staged it?

No company that can be documented could readily have staged it, and that vacancy is a standing embarrassment for the early date. In 1591 the London stage belonged to companies in flux, the Queen’s Men declining, Strange’s and Sussex’s and the Admiral’s men shifting personnel and repertory, and the playwright’s name is attached to none of them by any surviving record. Within a year the plague closures of 1592 to 1594 shut the theatres for long stretches and scattered the companies to the provinces. The company demonstrably owning the tragedy, the Chamberlain’s organisation in its Hunsdon interval, was formed only in 1594, out of the post-plague settlement. A 1591 premiere must therefore invent its stage, its sharers and its audience, and then explain six years of silence before the applause the title page reports.

Q: How do editors who print the play treat the earthquake in their notes?

Editors who print the play treat the earthquake in their notes as a famous possibility rather than a foundation: the note at 1.3 records the Dover Straits shock of 6 April 1580, sets out the seductive sum that would carry composition to 1591, and then declines the conclusion. Brian Gibbons in the Arden Second Series edition reports the argument within a dating discussion that settles in the middle 1590s. Jill L. Levenson’s Oxford edition prints a date of 1595 and handles the Nurse’s figures as period colour. René Weis, in the Arden Third Series, likewise places the play in the mid-decade. The pattern across the standard editions is uniform enough to state as a rule: the clue is always mentioned, because readers ask, and never used, because the documents forbid the use.

Q: What is the strongest single objection to dating the play by the earthquake?

The strongest single objection is the location: the earthquake in the text shakes a dove-house in Verona, not a church in London. Before the famous sum can start, the critic must transport the event out of the fiction’s geography into the author’s biography, must trust the exact count of a witness the scene itself characterises as gloriously approximate, and must equate the story’s present with the year of writing. Each step is possible; their conjunction is the argument, and none of the three steps is evidenced by anything except the conclusion they are assembled to reach. Rival candidates and rival sums then finish the work, but the dove-house is where the case first breaks, because it stands inside the very lines quoted as proof.

Q: Why does the earthquake clue survive in textbooks long after scholars have discounted it?

The earthquake clue survives in textbooks because it is a better story than the truth it displaces. The real dating of the play is a convergence of probabilities, a patron’s short-lived name, a register entry, a battery of verse statistics, honestly summarised as 1595, perhaps, inside 1594 to 1596, and no classroom wall has ever loved a probability. The clue, by contrast, offers a detective’s pleasures in a single breath: a mysterious line, a famous disaster, one addition sum, a precise year. It also flatters the text, seeming to hide its secret in plain sight for the attentive reader to find. Arguments with that shape outlive their refutations as a matter of cultural physics, and editors keep mentioning the clue partly in order to keep retiring it.

Q: How should an essay use the earthquake clue without overstating the case?

An essay should use the earthquake clue as colour with a question mark, never as a premise. The sound form is concessive: the Nurse’s eleven years may glance at the great shock of 1580, a memory from the playwright’s own schooldays, but the line sits inside a Veronese fiction, its count belongs to a comic character, a rival tremor of 1584 fits the same figure to 1595, and the documents, the Hunsdon title page, Meres, the verse, place the play in the middle 1590s in any case. Stated that way, the clue becomes evidence for the essay’s real subject, how the play builds its world from remembered England, instead of a false foundation for a date. The essay that writes ‘in 1591, as the earthquake line proves’ has announced its method and its error in one clause.

Q: What does the Nurse’s wager of her teeth reveal about her numbers?

The Nurse’s wager reveals that her numbers are performances of certainty rather than products of record. She will lay fourteen of her teeth that Juliet is not fourteen, and adds, to her teen be it spoken, that she has but four: the stake is three and a half times her capital, offered in the same breath as the fact it guarantees. The joke is precise about her epistemology. Her knowledge is embodied and anecdotal, stored in the sun under the dove-house wall, in wormwood and a tantrum, vouched for by vehemence, repetition and the refrain I remember it well. Within the fiction her Lammas arithmetic is sound, and the play intends it to be. As a witness for a chronologer, though, the teeth are the credentials the scene actually issues her, and no court would date a manuscript on them.

Q: Why is the weaning, rather than Juliet’s birth, the event the earthquake dates?

The weaning is the event the earthquake dates because the line’s grammar pins the eleven years to the day Juliet was weaned, not to her birth: ‘tis since the earthquake now eleven years, and she was weaned upon that day. The distinction matters more than it looks. The clue’s popular form often runs as though the quake marked Juliet’s birth year, from which the scene’s date is then computed directly. In fact two of the Nurse’s numbers must be combined before any year appears: the birth on Lammas Eve, nearly fourteen years before the scene, and the weaning, eleven years before it, on the quake day. The earthquake therefore dates a toddler’s milestone in the middle of the back-story, and the count that reaches the scene has already passed through the fiction’s internal arithmetic before the critic’s external arithmetic can begin.

Q: Did anyone in the play’s first audience need the 1580 earthquake explained to them?

Nobody in the play’s first audience needed the 1580 earthquake explained, and that is part of the problem for the clue. A London audience of the middle 1590s was full of men and women who had felt the shock as children, apprentices or young householders, who had heard the sermons, read or heard read the pamphlets of Churchyard and Golding, and grown up with the tremor as a standard landmark, the year of the earthquake, by which families dated their own pasts exactly as the Nurse dates Juliet’s. A reference that lands instantly at fifteen years’ distance lands no worse than at eleven, and the line’s theatrical effect, instant recognition of how memory works, is identical in 1591, 1595 or 1597. Universal recognition is the one quality that cannot discriminate between dates.

Q: Is the earthquake line the only internal clue to the play’s date?

The earthquake line is not the only internal clue, merely the only one that arrives wearing a number. The play’s whole verbal fabric testifies internally: the proportions of rhyme, prose, enjambment and feminine endings that group it with Richard the Second and A Midsummer Night’s Dream; the sonnet forms built into its Prologue, its second chorus and its lovers’ first exchange, fluent in the sonnet culture that crested in the middle 1590s; the absence of any reference that demands a later year. The difference between the clues is methodological. The earthquake offers a single striking correspondence that must be exact to signify. The stylistic evidence offers thousands of small correspondences that signify by accumulation and survive audit. Dating by the fabric is slower, duller and repeatable, and it is where the consensus actually comes from.

Q: What grade of confidence does the mid-1590s date deserve?

The mid-1590s date deserves a split grade: high confidence for the window, moderate for any single year within it. The window of 1594 to 1596 rests on documents that cannot be argued away, performance under the Hunsdon name by 1596 to 1597, print in 1597, canonical listing by Meres in 1598, reinforced by every stylistic test that has been run on the verse. Inside the window, 1595 is the best single estimate, the Oxford chronology’s choice and the standard editions’ centre of gravity, but it remains an estimate, and honesty requires printing it with its margins. The earthquake clue, graded as an instrument, earns the lowest confidence on the table: compatible with 1591, 1595 and the fiction’s own 1582, it discriminates among none of them.