Artificial Wastelandtheatre and dance / record audit

A live reading of the London Stage Database

Older Than the Daily Courant

In the London Stage Database, the old plays on a recorded bill are not dated by their premiere. This page rebuilds their age from first observation, leaves 44,818 unlinked performance records visibly unknown, and lets you switch to catalogued publication dates to watch the result survive a different clock.

Move through recorded London bills from 1705 to 1800 and watch how many linked mainpieces the record had already seen at least 20 years earlier, recomputed from the frozen archive. Switch that clock to catalogued publication dates, or expose a missing WorkId. It cannot recover premieres, unrecorded performances, or audience taste.

Before any age is computed, the frozen record holds 116,816 performance records. 71,998 carry a WorkId. 44,818 do not: 30,379 are dances, songs, music and other items this record almost never links to a work, and 14,439 are mainpieces or afterpieces. 327 have a partial date and 0 an invalid one, and the age instrument leaves those out. Open the inventory.

Layer one: put your hand on a bill

How old is the repertory when the curtain rises?

Choose a year. The page gathers the source's recorded rows, follows only a usable WorkId to the Works table, and recomputes an age on the chosen clock. Every age below says which clock it was measured on, and the reset button returns every choice to its default.

1750 to 1754: 78.4% at least 20 years past first full-date observation; 14.2% within 5 years of first full-date observation; 1,858 with a computed age.

Live result from the frozen record

old-share look-back open

In the record for 1750 to 1754: 2,290 mainpiece performance records with a full date at all venues. 432 have no WorkId and are refused an age. 0 are linked but have no usable start on the first full-date observation clock. 0 more have a partial date and were left out before counting.

1750 to 1754: 78.4% at least 20 years past first full-date observation.

In the 1750 to 1754 block, 78.4% of linked mainpiece performance records were at least 20 years past the first full-date observation of their work, and 14.2% were within 5 years of it.

14.2% within 5 years of first full-date observation. The denominator is 1,858 records with a computed age, not the 2,290 records on the bills.

Median years since first full-date observation46 years
Median first full-date observation year1706
No WorkId432
First observed 1703 to 1707301

Age bands, in years since first full-date observation.

At most 5 years263
6 to 19 years139
At least 20 years1,456
No WorkId432
No usable origin0

First Drury Lane bill in calendar 1750 (inside the 1750 to 1754 block)

1750-01-01

Read the row before reading the number.

Ages are years since first full-date observation, taken at calendar 1750. Source season as written: 1749-1750. Season derived from the date: 1749-1750.

  1. The Tempest Or The Inchanted Islandmainpiece
    44 years since first full-date observation
  2. The Anatomistafterpiece
    54 years since first full-date observationa floor: first observed before daily advertising began in 1705

What the record holds, before any result

The source contains physical cast rows, event IDs, performance IDs, works, and work links. This extract keeps each unit in its own column and carries source spelling beside every normalized key. Read the missingness here before reading any share: a record without a WorkId or a full date never enters an age denominator, and a reused PerformanceId is kept as a visible conflict rather than a hidden duplicate.

Physical CSV rows842,740
Raw EventIds52,617
Numeric PerformanceIds116,816
Events with a numeric performance51,935
First-seen EventIds51,934
With a WorkId71,998
Without a WorkId44,818
Partial or invalid dates327
Performance records by source kind, work identity: WorkIds as published
KindRecordsNo WorkIdPartial or invalid date
mainpiece51,9347,360319
afterpiece34,4767,0797
dance19,93219,9311
song5,6795,6790
music2,7382,7380
entertainment1,0711,0710
ballet7687440
monologue1911900
opera21200
instrumental660

Dates: 116,489 records have a full date, 327 have the day missing, 0 have the month missing, and 0 are invalid. Every record without a full date is left out of the age instrument, and the result card counts how many fall in the selected period.

Of the 44,818 records without a WorkId, 14,439 are mainpieces or afterpieces and 30,379 are dances, songs, music, entertainments, ballets, monologues, operas and instrumentals, which this record almost never links to a work.

First full-date event: 1659-10-29. Last full-date event: 1800-09-16. Works with a full first observation: 2,638. Works with a known first year: 2,811. Works with a four-digit publication year: 2,285 of 3,678.

  • PerformanceId 2 has 2 source variants. The first CSV occurrence is retained and the conflict stays visible.
  • 12 malformed cast rows at EventId 8270 retain their source text in the audit.

The extract is a compact derivative of the MIT deposit. The original CSV and SQL archives are not requested by the page at runtime.

Layer two: dismiss the instrument

The curve is in the archive, not in the story

The easy dismissal is that the default answer is only a choice of clock. So here the same instrument travels across the whole record. Click a row to move the first layer, or switch the controls and watch the curve change under the same witness rules.

The first-observation histogram is drawn from the counted rows once the frozen extract opens.

The block-by-block curve is drawn from the same engine once the frozen extract opens.

With the default choices, the share at least 20 years past first full-date observation is 78.4% in 1750 to 1754 and 33.3% in 1795 to 1799. The share within 5 years moves from 14.2% to 52.5%. Both lines measure the distance from the chosen clock to recorded bills, not the taste of an audience.

A different clock is a different question

The Works table has a publication field, but it does not have a guaranteed premiere field. Compare first full-date observation with catalogued publication on the same calendar year. The publication clock sets aside a work with no catalogued publication year, and a performance that came before its publication.

First full-date observation

36 years, median

Years since first full-date observation: 80.5% at least 20, 12.0% within 5.

416 linked mainpiece records in calendar 1740 have a start on this clock.

Catalogued publication year

41 years, median

Years since catalogued publication year: 83.2% at least 20, 7.3% within 5.

381 linked mainpiece records in calendar 1740 have a start on this clock; 35 set aside (8 with no four-digit publication year, 27 performed before it).

Earlier of either clock

41 years, median

Years since earlier of first observation and publication year: 82.5% at least 20, 10.1% within 5.

416 linked mainpiece records in calendar 1740 have a start on this clock.

Fixed settings for this comparison: mainpieces, every performance record, WorkIds as published, all venues, calendar 1740 alone, and no look-back cut. The controls above do not change it, so the clock is the only difference between the three cards. Publication is a different formal field, and a missing or later publication year is set aside rather than guessed.

Open a work, not just a percentage

Search the frozen Works table. A title opens its recorded appearances, its source publication field, and the same age refusal used above, on the clock chosen above. The list is a route into the record, not an editorial canon.

The Works table is opening.

Choose a title to open its recorded appearances.

The check

The record shows where it can fail

This is the check. This part is recomputed every time a control moves. Every figure comes from engine.js on the frozen extract, the same code that draws the result above, and the sums are laid out so that a wrong figure shows as a sum that does not close.

Settings now: bill year 1750; period 1750 to 1754 (Five-year block, Calendar year); items: Mainpieces; count each: Performance record; work identity: WorkIds as published; clock: First full-date observation; old means at least 20 years; recent means at most 5 years; look-back trusted from 1710; playhouses: All venues.

Records on the bills
2,290
No WorkId, refused an age
432
Linked, no usable start on this clock
0
Age computed
1,858
Sum
1,858 + 432 + 0 = 2,290, the records on the bills: closes
Partial-date records in this period, left out first
0
At least 20 years past first full-date observation
1,456 of 1,858 = 78.4%
Within 5 years of first full-date observation
263 of 1,858 = 14.2%
Between the two
139; 263 + 139 + 1,456 = 1,858: closes
Look-back
1750 minus 20 is 1730, not before 1710: open
Median years since first full-date observation
46
Median first full-date observation year
1706; 301 of 1,858 first observed 1703 to 1707

What the frozen files hold, recounted in your browser

Raw EventIds52,617recount 52,617: agrees
Numeric PerformanceIds116,816recount 116,816: agrees
Events with a numeric performance51,935recount 51,935: agrees
First-seen EventIds51,934recount 51,934: agrees
Physical CSV rows842,740extraction audit only
Without a WorkId, WorkIds as published44,818of 116,816 records
Hume-window titled records, 1660 to 17001,661Hume counts 949 in a different unit

The frozen audit records 1 reused PerformanceId conflict, at ID 2. The first CSV occurrence rule is visible in audit.json. The physical CSV row count cannot be recounted here because the CSV itself is not shipped. 26 records dated 1670 retain the source season 1668-1669, so season and calendar are not silently interchangeable.

Anchor witnesses, on workids as published

John Genest, Some Account of the English Stage, from the Restoration in 1660 to 1830 (Bath: H. E. Carrington, 1832), vol. 5, pp. 365, 459, 555 and 556, and vol. 6, pp. 9 and 12. Genest worked from playbills and newspaper bills more than a century before The London Stage was edited, so he is independent of that edition's editors, its OCR, its date recovery and its WorkId links. He is not independent of the surviving bills: where a bill is lost or misprinted, both witnesses can be wrong together.

School for Scandal

agrees

May 8. Never acted, School for Scandal.

Genest witness
1777-05-08
Engine result
1777-05-08

Vol. 5, printed page 555, archive leaf n562. view witness

School for Scandal

agrees

The theatre closed June the 7th with the School for Scandal 20th time. (Bills from Mr. Woodfall.)

Genest witness
20
Engine result
20

Vol. 5, printed page 556, archive leaf n563. view witness

School for Scandal

agrees

26. Smith's bt. School for Scandal, 57th time.

Genest witness
57
Engine result
57

Vol. 6, printed page 9, archive leaf n18. view witness

School for Scandal

agrees

28. (Last play) School for Scandal, 65th time. (From my own bills.)

Genest witness
65
Engine result
65

Vol. 6, printed page 12, archive leaf n21. view witness

She Stoops to Conquer

disagrees

March 15. Never acted, She Stoops to Conquer, by Dr. Goldsmith.

Genest witness
1773-03-15
Engine result
1773-03-16

Vol. 5, printed page 365, archive leaf n372. view witness

Rivals

disagrees

17. Never acted, Rivals.

Genest witness
1775-01-17
Engine result
1664-09-10

Vol. 5, printed page 459, archive leaf n466. view witness

Refusal and repair

One day that the join cannot carry

She Stoops to Conquer

Genest writes March 15. The first linked record arrives on 1773-03-16 because the March 15 row has no WorkId. The number does not silently move to the witness.

The unlinked row is still in the inventory and is refused by the age instrument.

A reused identity needs a named repair

The Rivals

On the published links, the 1775-01-17 Covent Garden record carries WorkId 727, whose first full-date observation is 1664-09-10. That makes its age in 1800 136 years since first full-date observation.

The post-1774 repair to WorkId 2690 gives 25 years since first full-date observation. It is a choice, not a correction smuggled into the default.

What this instrument cannot know

The first full-date row is the first observation carried by this record. A play may have been performed before that row, and a WorkId may be absent even when a reader could identify the title. The page refuses a missing identity rather than turning the title string into a certainty. A work first observed before daily newspaper advertising began has a floor, not an age.

Genest supplies the calibration witnesses: John Genest, Some Account of the English Stage, from the Restoration in 1660 to 1830 (Bath: H. E. Carrington, 1832), vol. 5, pp. 365, 459, 555 and 556, and vol. 6, pp. 9 and 12. He worked from playbills and newspaper bills more than a century before The London Stage was edited, so he is independent of that edition's editors, its OCR, its date recovery and its WorkId links, but not of the surviving bills themselves. On the published links, 4 of 4 School for Scandal witnesses agree with the engine: first observation on 8 May 1777, the 20th recorded performance by 7 June 1777, the 57th recorded performance by 26 March 1778, and the 65th recorded performance by 28 May 1778. The She Stoops to Conquer and Rivals witnesses are deliberately left as disagreements, because their record joins do not support the printed witness without a choice.

The London Stage Database publishes the data access and citation pages at data.php, citation.php, and about.php. The frozen source file is the LondonStageFull CSV deposit, paired with the Works table from the SQL deposit.

We searched Theatronomics, the London Theatre companion, Adam Matthew's London Stage visualisations, and the Digital Humanities Quarterly history on 2026-09-14 and did not find a published age distribution using this first-observation and recorded-bill instrument.