Comparison · 21 September 1822 · Hampstead
Two cloud studies by John Constable carry the same date. Yale's label says half past one; the Courtauld's verso says three o'clock. Both say looking south, an east wind, and warm. This page reads the two notes against the independent record of that afternoon: Luke Howard's printed register, a weather reconstruction that used no painting, and the computed sun. It compares the notes and their context, not the cloud forms.
When an artist wrote the weather on a painting, how far do independent sources from the same day support each word? The question is not whether the painting is an accurate picture of the sky.
The inscriptions are described evidence. Howard’s register is measured and described, the regional reconstruction and computed sun are estimated, and each source keeps its own place, scale and limits.
Verdicts are labels, not scores. “Partially agrees” means compatible but not identical, while “Unknown” means no located source can test the claim.
The Nephos method
Hampstead, 21 September 1822, two notes ninety minutes apart in Constable’s hand.
Sep[br] 21. 1822. ½ p’s[t] one o clock. looking south. wind very fresh at East, but warm.Yale Center for British Art, B1981.25.116, as the Yale catalogue transcribes the label
Sepr.21.1822. Looking South, brisk wind to east, warm and fresh. 3 o’cl. afternoonThe Courtauld, P.1952.RW.69, as the Courtauld catalogue transcribes the verso
Two notes by one hand agree with each other, and the regional records fit an easterly, mild day. That is internal consistency, not a second witness.
What this cannot tell you. Both notes are Constable’s; their agreement is not confirmation by another observer. No cloud-to-cloud comparison is made. The reanalysis cell cannot reproduce Hampstead.
Each inscription is given as its museum transcribes it. Each sheet is identified by its museum and accession number, not by title or date: these are two objects, not two views of one. The paintings are displayed at equal widths for reading, not at physical scale.
Yale Center for British Art · B1981.25.116
“Sep[br] 21. 1822. ½ p’s[t] one o clock. looking south. wind very fresh at East, but warm.”
Yale’s catalogue transcription of the inscription, which is on a label fixed to the stretcher; the verso is now covered by a conservation backboard. K1 K2
Courtauld Gallery · P.1952.RW.69
“Sepr.21.1822. Looking South, brisk wind to east, warm and fresh. 3 o’cl. afternoon”
Courtauld catalogue transcription of the inscription on the verso. K3
02 · The time instrument
The Courtauld’s inscribed hour. Model values for 15:00 UTC; sun from JPL Horizons, as on the Courtauld record.
The reanalysis never saw either painting. Its five three-hourly samples for the London-area cell show cloud cover with its ensemble spread, and the model’s wind direction and speed. The two inscribed hours are marked. K7 K8
Bars are ensemble-mean total cloud cover on a 0 to 100% scale; the pale band behind each bar is one ensemble spread either side, clipped at the scale. Arrows point where the wind blows; the label gives where it blows from. Temperatures are the 2 m ensemble means, converted from kelvin. The 13:00 UTC reading used on the Yale record (61%, 57°, 5.9 m/s) lies between the 12:00 and 15:00 samples.
Luke Howard’s printed table for September 1822, from the page image checked on 18 August 2026. Day 21 reads NE. His note for the 17th through the 21st is one word.
21 · NE. Barometer 29.87 and 29.80 inches; temperature 65 and 49 degrees Fahrenheit by the first thermometer. The day before reads SE, the day after E. K6
“Fine.” That is the whole note for the 17th through the 21st. Then, for the 22nd: “Cloudy: windy: swallows begin to congregate.” K6
The table page does not label its station, and Howard’s registers for these years were kept at Tottenham and Stratford, not Hampstead. His NE is a day-level London-area reading from another location. Constable wrote East on both sheets. Adjacent sectors, different places: Nephos rates this “Partially agrees”, not a match. K7
Central England’s daily mean for the 21st was 11.6°C, slightly below the 12.6°C average of the prior fifty 21 Septembers, inside a mild spell that fell to 9.5°C by the 26th. Nothing in the series defines “warm”. K9
The two notes, the outside witnesses, and the verdict each record already carries. Both notes are Constable’s, so agreement between them is internal consistency, not confirmation by another observer.
| Claim | Yale13:30 · K1 | Courtauld15:00 · K3 | Howard’s registerLondon area · K6 | Reanalysis20CRv3 · K8 | The sunComputed · K10 | ThermometersHoward and CET · K9 | Verdict |
|---|---|---|---|---|---|---|---|
| Date | “Sep[br] 21. 1822.”Yale label | “Sepr.21.1822.”Courtauld verso | Howard’s table has a row for the 21st; nothing in it names a painter. No separate observer confirms the date of either sheet. | — | — | — | Unknown |
| Time | “½ p’s[t] one o clock.” | “3 o’cl. afternoon” | — | — | Computed sun 35.3° at 210° for 13:30 and 25.9° at 234° for 15:00: a drop of 9.4° and a swing of 24° toward the west. Both are early-to-mid afternoon positions, but neither sheet has a horizon or a landmark, so no painted hour can be tested. | — | Unknown |
| View | “looking south.” | “Looking South,” | — | — | At both hours the sun stood to the right of a south-facing painter, lower and further right by three o’clock. Consistent, and untested: a pure sky gives nothing to measure a bearing against. | — | Unknown |
| Wind | “wind very fresh at East,” | “brisk wind to east,” | NEFor the day, at Howard’s station. | ENE, 57° at about 5.9 m/s for the 13:00 target and 54° at 6.0 m/s at 15:00, on the NE/ENE boundary, for a cell about 111 × 68 km. Easterly across every source, but the labels differ and none of the witnesses stood on Hampstead Heath. Strength is compatible only at grid scale. | — | — | Partially agrees |
| Warmth | “but warm.” | “warm and fresh.” | 65°F / 49°F | Grid-cell air 14.0°C at the 13:00 target and 13.9°C at 15:00, spread about 1.7 K. | — | Howard’s thermometer at his station reached about 18.3°C. Central England daily mean 11.6°C, slightly below the calendar-date average. No source defines “warm” or “fresh”, and none measures Constable’s position. | Unknown |
| Same day | Both notes | Both notes | Easterly day | The regional records fit an easterly, mild afternoon whose modelled sky barely changed between the two hours. | Ninety minutes apart | The warmth claim stays open. | Partially agreesThe two inscriptions agree on date, southward view, east wind and warmth. No cloud forms are compared. |
What changed between the two notes: the sun, by 9.4° of altitude and 24° of azimuth. What the model says changed: almost nothing, 61% to 63% cloud and a wind still from about 55°, between NE and ENE. That steadiness is the finding, and it is a finding about the regional record, not about the pictures.
On one afternoon Constable wrote two weather notes ninety minutes apart, and they agree with each other on the date, the view, the wind and the warmth. Because both are his, that agreement is internal consistency, not confirmation by a second observer. The regional records fit an easterly, mild day whose modelled sky hardly changed between half past one and three, while the sun dropped nine degrees and swung a quarter of the way toward the west. The wind claim partially agrees with Howard's NE and the reconstruction's ENE; the hour, the bearing and the word "warm" stay open, because nothing in a horizonless sky can test them. This is the pattern the whole project reports: near-agreement where the sources reach, and honest silence where they do not.
Read the method, the Yale record, and the Courtauld record.
