Nephosthe weather behind the paintings
Pissarro’s elevated view across the boulevard des Italiens, with trees, pedestrians, carriages and patches of light on the pavement.
Camille Pissarro, Boulevard des Italiens, Morning, Sunlight, 1897. Oil on canvas, 73.2 × 92.1 cm. Chester Dale Collection, 1963.10.198. Courtesy National Gallery of Art, Washington. CC0.

A study against the weather record

Camille Pissarro

Boulevard des Italiens, Morning, Sunlight

1897 · National Gallery of Art, Washington, 1963.10.198 · Chester Dale Collection

Pissarro framed a sunlit morning from his hotel window. What can the campaign’s rain record test about that fleeting light?

27of 75 days had at least 1 mm of rain at Paris-Montsouris. A daily rain total cannot tell us when sunlight reached the boulevard.

Sunlight unverifiedPainted day unknown

The recordsThe painting at NGA

The Nephos method

How this study was read

The method in full

  1. What is written

    Signed 1897, painted from the Grand Hôtel de Russie. The study uses 10 February–25 April as a campaign envelope and also calculates NGV’s shorter 10 February–17 April window. Neither dates this canvas’s sessions.

  2. Records of those days

    Not found or not yet searched

    • No document inspected here dates a session on this canvas.
    • The checked sunshine-duration column is blank for the campaign; hourly boulevard light remains unknown.
    • NGA curatorial files, original letters and a technical examination of this canvas were not inspected.

  3. Each claim tested

    • How wet was the campaign at the station? 130.3 mm and 27 days with at least 1 mm in 75 days. Rainfall ranks 37 of 152, wettest first, against the same calendar window in 1873–2024.
    • Does the shorter catalogue window change the conclusion? Through 17 April: 121.9 mm, 25 wet days in 67. Both windows exceed their 1991–2020 same-window mean; neither measures sunlight.
    • Does rain rule out the artist’s reported sunlight? No. Montsouris records 7.3 mm for 2 March, while Pissarro reported sunlight on the procession. A daily total at another location cannot settle the light at one hour.

  4. Verdict

    A sunlit effect belongs to a campaign with repeated rain. The gauge measures the campaign’s rainfall, while the artist’s report shows why a rainy date need not exclude a sunlit interval. The NGA picture’s light and date remain unverified.

    • UnknownMorning sunlight in this canvas Rain totals are not light measurements; no session date is documented.
    • UnknownA painted day Series dates and the sale establish context, not an individual sitting.

What this cannot tell you. Montsouris is a park station south of the boulevard. Rain there is not an observation at the hotel. RR is assigned to the day beginning at 06:00 UTC. It cannot locate rain within the day. The modern mean and historical rank are descriptive comparisons across an unhomogenized station record. The Mardi Gras testimony concerns another motif in the series. It does not identify the NGA picture as a carnival scene or date it to 2 March. The NGA account emphasizes morning light and compositional cropping. No evidence inspected here establishes that Pissarro invented the light.

Step 1 · The picture and its making

One morning effect in a series

From a room at the Grand Hôtel de Russie, Pissarro could turn towards the boulevard des Italiens. NGA’s account describes morning sunlight reaching the pavement at the lower right and work in two-hour stretches. The viewpoint, angled street and cuts at the edges turn passing traffic into a tightly framed composition. Theodore Reff’s catalogue entry emphasizes how the brushwork and cropping convey movement. These accounts explain how the image was made; they do not establish a single sitting or a measured sky.

The signature dates the canvas to 1897. In his letter of 8 February, quoted in the catalogue raisonné, Pissarro names his room at 1 rue Drouot. This is one of two Italiens views within the larger boulevard series. NGA records its sale by the artist to Durand-Ruel on 11 May 1897.

The campaign dates differ in the published accounts: NGA says mid-February–mid-April; NGV gives 10 February–17 April; the 2005 catalogue places the hotel stay through 25 April. The main calculation uses 10 February–25 April as a conservative campaign envelope, with the shorter NGV window alongside it. Neither is a documented execution interval for this particular canvas. No source inspected here establishes later studio reworking or the dates of individual sessions.

Step 2 · The campaign’s rainfall

Repeated rain around the sunlit picture

Paris-Montsouris recorded 130.3 mm in the wider window, with at least 1 mm on 27 of 75 days. The station is about 5.6 km south of the present-day hotel address, in a different setting. It provides Paris rainfall context, not an observation at the window.

The total ranks 37 of 152, wettest first, for the same calendar dates in 1873–2024. Its 1991–2020 comparison mean is 104.5 mm. The shorter window also exceeds its modern comparison mean. These are descriptive comparisons of the archived station record; changes in instruments and exposure have not been homogenized here.

Daily precipitation at Paris-Montsouris, 10 February–25 April 1897One bar per provider day. Zero values mark no measured precipitation, not sunshine. All values are in the table below.0510152025mm10 February 1897: 5.4 mm10 February11 February 1897: 6.0 mm12 February 1897: 1.0 mm13 February 1897: 0.4 mm14 February 1897: 0.0 mm15 February 1897: 0.0 mm16 February 1897: 0.0 mm17 February 1897: 0.0 mm18 February 1897: 0.0 mm19 February 1897: 0.0 mm20 February 1897: 0.0 mm21 February 1897: 0.0 mm22 February 1897: 0.0 mm23 February 1897: 0.0 mm24 February 1897: 0.0 mm25 February 1897: 0.0 mm26 February 1897: 0.0 mm27 February 1897: 0.0 mm28 February 1897: 0.4 mm1 March 1897: 1.6 mm1 March2 March 1897: 7.3 mm3 March 1897: 1.7 mm4 March 1897: 2.8 mm5 March 1897: 3.8 mm6 March 1897: 0.0 mm7 March 1897: 0.0 mm8 March 1897: 0.1 mm9 March 1897: 3.0 mm10 March 1897: 0.0 mm11 March 1897: 2.0 mm12 March 1897: 0.2 mm13 March 1897: 4.6 mm14 March 1897: 21.4 mm15 March 1897: 0.9 mm16 March 1897: 0.3 mm17 March 1897: 1.0 mm18 March 1897: 7.1 mm19 March 1897: 1.1 mm20 March 1897: 0.0 mm21 March 1897: 0.2 mm22 March 1897: 0.0 mm23 March 1897: 0.0 mm24 March 1897: 0.0 mm25 March 1897: 0.0 mm26 March 1897: 0.0 mm27 March 1897: 0.0 mm28 March 1897: 2.6 mm29 March 1897: 0.0 mm30 March 1897: 0.0 mm31 March 1897: 15.1 mm1 April 1897: 4.0 mm1 April2 April 1897: 2.4 mm3 April 1897: 9.0 mm4 April 1897: 0.1 mm5 April 1897: 0.0 mm6 April 1897: 4.7 mm7 April 1897: 3.1 mm8 April 1897: 0.0 mm9 April 1897: 2.9 mm10 April 1897: 0.0 mm11 April 1897: 0.0 mm12 April 1897: 0.0 mm13 April 1897: 0.0 mm14 April 1897: 3.4 mm15 April 1897: 0.0 mm16 April 1897: 0.2 mm17 April 1897: 2.1 mm18 April 1897: 0.0 mm19 April 1897: 4.9 mm20 April 1897: 0.3 mm21 April 1897: 2.7 mm22 April 1897: 0.5 mm23 April 1897: 0.0 mm24 April 1897: 0.0 mm25 April 1897: 0.0 mm
Dark bars: at least 1 mm. Pale bars: smaller measured amounts. Marks below the baseline: zero in the gauge, not observed sunshine. The provider assigns rainfall from 06:00 UTC to 06:00 the following day to the starting date.
Both documentary windows, calculated with the same rules
1897 windowDaysRain, mmDays ≥1 mmWettest rank1991–2020 mean, mm
10 February–25 April75130.32737 / 152104.5
10 February–17 April67121.92532 / 15293.6

All comparison windows have complete rain values with provider quality codes 0 or 1. Leap day is omitted from every year. Rank counts strictly larger totals plus one; all comparison values remain in the linked JSON.

Every daily rain value, 10 February–25 April 1897
Provider datePrecipitation, mmQuality code
10 February5.41
11 February6.01
12 February1.01
13 February0.41
14 February0.01
15 February0.01
16 February0.01
17 February0.01
18 February0.01
19 February0.01
20 February0.01
21 February0.01
22 February0.01
23 February0.01
24 February0.01
25 February0.01
26 February0.01
27 February0.01
28 February0.41
1 March1.61
2 March7.31
3 March1.71
4 March2.81
5 March3.81
6 March0.01
7 March0.01
8 March0.11
9 March3.01
10 March0.01
11 March2.01
12 March0.21
13 March4.61
14 March21.41
15 March0.91
16 March0.31
17 March1.01
18 March7.11
19 March1.11
20 March0.01
21 March0.21
22 March0.01
23 March0.01
24 March0.01
25 March0.01
26 March0.01
27 March0.01
28 March2.61
29 March0.01
30 March0.01
31 March15.11
1 April4.01
2 April2.41
3 April9.01
4 April0.11
5 April0.01
6 April4.71
7 April3.11
8 April0.01
9 April2.91
10 April0.01
11 April0.01
12 April0.01
13 April0.01
14 April3.41
15 April0.01
16 April0.21
17 April2.11
18 April0.01
19 April4.91
20 April0.31
21 April2.71
22 April0.51
23 April0.01
24 April0.01
25 April0.01

Code 1 means validated by automatic checks or a climatologist. The sunshine-duration cells are blank throughout this window.

Step 3 · The artist’s witness

Sunlight on a date with recorded rain

There is a dated reason not to equate rain totals with the light Pissarro painted. Reff, p. 258, quotes Pissarro’s report after the Mardi Gras procession of 2 March 1897:

“effects of sun on the serpentines and the trees”

Pissarro, quoted in Reff, Manet and Modern Paris (1982), p. 258, citing the published letters, p. 308. The original letter was not inspected.

For the provider date 2 March, Montsouris records 7.3 mm. The gauge integrates a whole observation day at another location; the artist describes light during an event. The two accounts can coexist. They do not establish the rain’s hour or whether it fell at the hotel.

This witness concerns the procession and other pictures in the series. It does not turn the NGA picture into a Mardi Gras scene or date it to 2 March. It demonstrates why a rain record cannot be used to select, confirm or reject a sunlit painting day.

Step 4 · Where the records stop

The light remains unverified

The title names an effect within a changing campaign. The rainfall record supplies a wetness context, and the artist’s separate testimony describes sunlight during that campaign. Neither verifies the light in this canvas. The additional Météo-France file has no sunshine-duration values for the study window; no hourly light record at the boulevard has been established.

The documented pictorial choices are the elevated viewpoint, angular framing, cropped forms and abbreviated brushwork. There is no basis here for saying that Pissarro invented the sunlight or changed the weather. The next evidence needed is a document tying this canvas to particular sessions, together with observations of light at those times. Until then, its morning remains a pictorial subject rather than a recovered meteorological event.

Sources

Every source on this page

Weather values are calculated from cached files checked against their sha256 fingerprints. Receipts distinguish downloads, extracted text and saved web-tool extracts. Short quotations are checked against those texts; the decisive catalogue pages were also inspected as images. A fingerprint preserves the evidence used; it does not independently verify its interpretation.

  1. Météo-France daily observations, Paris: before 1950Licence Ouverte / Etalab 2.0 · retrieved 20 September 2026 · Station 75114001; pooled Caillebotte cache reusedsha256 04ccefd0095bb629f1e44434308eef0770693764fc533b96f5cc3ad4553d73f8
  2. Météo-France daily observations, Paris: 1950–2024Licence Ouverte / Etalab 2.0 · retrieved 20 September 2026 · Station 75114001; pooled Caillebotte cache reusedsha256 128d559d01d1f4633e3d1cbb12da8f33e48f6e6e2cb803c3b6421ed86f5ce136
  3. Météo-France, other daily parameters, Paris before 1950Licence Ouverte / Etalab 2.0 · retrieved 8 October 2026 · INST sunshine-duration column checked; blank throughout the campaignsha256 0823133554b7a4c992923dee193322fc039246d9e01129b7280d47300388deb9
  4. Météo-France, daily field definitions and quality codesMétéo-France public data documentation · retrieved 8 October 2026 · June 2026; RR observation interval, quality codes, INST definitionsha256 4085f0b5e44535dd15c870743f3f8b0a80ec752e771068626783ff454d0e9720
  5. Météo-France open-data catalogue and licenceLicence Ouverte / Etalab 2.0 · retrieved 8 October 2026 · Licence identifier lov2; département 75 resourcessha256 9d9ad820c3311ca22226c79fd671f4e5fb49dd407df0b3d7ea2a0b5c5ddc54b0
  6. National Gallery of Art, 1963.10.198: object recordMuseum text; brief quotation and paraphrase · retrieved 8 October 2026 · Saved web-tool extraction: direct HTTP returned 403; inscription, provenance, description and image rightssha256 9acc136d2e9cc2e0def814d0f65e1fc4b89d7ad121f042bb3e287986916af08c
  7. National Gallery of Art, Open Access policyNGA policy text; image licence CC0 · retrieved 8 October 2026 · Saved web-tool extraction; Open Access Policy sectionsha256 31f8f50dd790cae726ec8f19fae186cc50b65ee57051338d4d05b3b49389ea65
  8. National Gallery of Art, Picturing France 1830–1900Copyright NGA; publicly supplied educational PDF; research cache only · retrieved 8 October 2026 · Printed p. 10; note 1 p. 142 cites Rewald, Pissarro, p. 118sha256 8594c1a132f5a7bb1bb291db01b827effa8a310c431740cb7cf82b06e2a4e8d6
  9. National Gallery of Art, Picturing France 1830–1900 (text extraction)Copyright NGA; publicly supplied educational PDF; research cache only · retrieved 8 October 2026 · Printed p. 10; note 1 p. 142 cites Rewald, Pissarro, p. 118sha256 999cb1bbbb6cacc5c98a7ba76f0b2bb4dcfdcb95b93ca2b6d114fe877b904ab2
  10. Theodore Reff, Manet and Modern Paris (NGA, 1982)Copyright NGA; publisher supplies free research PDF; no republication licence · retrieved 8 October 2026 · Cat. 9, p. 50 (this painting); cat. 97, p. 258 (Mardi Gras witness, a different painting)sha256 22e0269b02cd50be22c4c5c571c9d42822febd5772e9deff8da3d5d7c39632e2
  11. Theodore Reff, Manet and Modern Paris (NGA, 1982) (text extraction)Copyright NGA; publisher supplies free research PDF; no republication licence · retrieved 8 October 2026 · Cat. 9, p. 50 (this painting); cat. 97, p. 258 (Mardi Gras witness, a different painting)sha256 1738a4aba437d1bd96c5b9cc90829b2dcae0ff27ccf778754a1877b5b6903f87
  12. Joachim Pissarro and Claire Durand-Ruel Snollaerts, Pissarro: Critical Catalogue (2005), vol. IIICopyright catalogue; WPI public reader, cached page for research only · retrieved 8 October 2026 · Printed p. 728; campaign XV and letters JBH IV 1366–1368sha256 fd22a658ad3017498974c1230c56dceed8c26c5ff666121b426409f5b1c82fef
  13. WPI reader OCR, printed p. 728Copyright catalogue; brief excerpts only · retrieved 8 October 2026 · Provider OCR from spreads.json?page=2; decisive passages checked against page imagesha256 921cbddb4e7c4bbc55d0149515698c05b60116eaf0d9e0e9d702b234aa1af6bf
  14. Joachim Pissarro and Claire Durand-Ruel Snollaerts, Pissarro: Critical Catalogue (2005), vol. IIICopyright catalogue; WPI public reader, cached page for research only · retrieved 8 October 2026 · Printed p. 737; no. 1174, NGA 1963.10.198sha256 db1f0a051267df0daa6184a4966c2dcd739cbb606bc3649644fe977e5d5753a4
  15. WPI reader OCR, printed p. 737Copyright catalogue; brief excerpts only · retrieved 8 October 2026 · Provider OCR from spreads.json?page=2; decisive passages checked against page imagesha256 30d0455183d63b9e5d2a8a9693bbe7809705337c7d69a8e79d49edabaf30de7d
  16. Sonia Dean, NGV collection account (2003), p. 132, on Google Arts & CultureCopyright NGV; brief citation and paraphrase · retrieved 8 October 2026 · Campaign dated 10 February–17 April; explicitly includes two Italiens canvasessha256 5f711be2ebf61864f0a74a438bbe8890d94cca9d0363f260d4c2be65b31db1a7
  17. Sonia Dean, NGV collection account (2003), p. 132, on Google Arts & CultureCopyright NGV; brief citation and paraphrase · retrieved 8 October 2026 · Campaign dated 10 February–17 April; explicitly includes two Italiens canvasessha256 c2d7bd6cc4ccabffcb5dbeb37db76a60bac73f73dd1974dd53fbded45a78647e
  18. OpenStreetMap, present-day 1 rue Drouot addressODbL 1.0; © OpenStreetMap contributors · retrieved 8 October 2026 · Approximate address location, not surveyed historical windowsha256 a5c450eeffc7f778fda4da8115b440025358279dfbc6a8f256b2265675dcccfd
  19. NGA, Boulevard des Italiens, Morning, Sunlight, 1963.10.198CC0; Courtesy National Gallery of Art, Washington · retrieved 8 October 2026 · 1600-pixel IIIF derivative linked from the NGA object pagesha256 673aad3b592147b203672536613a47d84eb0a088b53ef4fafa564ed6279b88f7

The numbers on this page, as JSONThe method card, as JSON