Nephosthe weather behind the paintings

An open reproduction basis for this canvas has not been established. The museum record identifies the object and the April 1931 sale-letter reference.

The painting’s museum record

Abbott Fuller Graves, A Kennebunkport Doorway (Peace and Sunshine), catalogued 1931; sold by 7 April 1931. Oil on canvas. Collection of the Art Fund, Inc. at the Birmingham Museum of Art; Gift of Dr. and Mrs. David A. Skier, AFI.104.2015. Artwork not reproduced.

A study against the weather record

Abbott Fuller Graves

A Kennebunkport Doorway (Peace and Sunshine)

Catalogued 1931 · Sold by 7 April 1931 · Birmingham Museum of Art, Art Fund collection · AFI.104.2015

Catalogued 1931, but sold before that summer. What was the latest summer it could show like?

52of 122 days classified clear at Portland in June–September 1930, the latest summer the painting could show. Its actual summer remains unknown.

Latest summer: 1930Painted summer unknown

The recordsThe museum’s dated sale reference

The Nephos method

How this study was read

The method in full

  1. What is written

    Catalogued 1931, but sold by 7 April 1931 according to the museum’s provenance and its cited Graves letter. Summer 1930 is the latest summer the painting could show. June–September is a broad garden-season test; an earlier summer or work from sketches remains possible.

  2. Records of those days

    Not found or not yet searched

    • The museum’s original April letter and later correspondence remain unread; no particular summer or session is documented.
    • The exact doorway is unidentified. Portland is regional evidence, not the garden’s observed weather.
    • The nearer Mount Agamenticus gauge fails the same completeness rule because of presumed-zero amounts.

  3. Each claim tested

    • Which summer can be compared without putting it after the sale? 1930, the latest possible summer. June–September allows a broad summer garden season without claiming that the canvas or its studies were made then.
    • Was that summer dry in this Maine record? Portland records 247.2 mm (9.73 inches), rank 9 for dryness among 44 complete seasons. The archive median is 299.3 mm. Rain still reaches at least 1 mm on 33 days. This is a local precipitation comparison, not a formal drought diagnosis.
    • Did this summer offer the kind of light the title evokes? The Weather Bureau classifies 52 of 122 days as clear (42.6%). Clear conditions recur, but are a minority category. Neither the counts nor the rainfall identifies Graves’s summer or the light at the doorway.

  4. Verdict

    A sale before summer 1931 shifts the test to the latest possible summer, 1930: relatively dry at Portland, with 52 days classified clear. The painting can draw on an earlier summer; its actual weather remains unknown.

    • Partially agreesLight available in the latest possible summer Published clear days show recurring regional opportunities for a sunlit garden; the year and garden conditions are not established.
    • UnknownThe actual painted summer, day and changes made by Graves The sale gives an upper bound; no dated session, identified doorway or making record fixes the scene.

What this cannot tell you. June–September 1930 is the latest possible summer test, not the painted summer. Earlier summers, sketches and later studio work remain possible. Cloud categories and cloudiness are not sunshine-duration measurements; numerical category cutoffs were not established from the consulted guide. No sunshine-hour total or historical sunshine rank is claimed. Precipitation sums require every daily amount to be unflagged or a documented trace; missing and presumed-zero amounts exclude a season. Trace days add no quantified depth. Ranks compare matching complete seasons only, with ties sharing rank. The 44-season archive is not homogenized. Exposure and observation practices may differ. No same-station 1991–2020 normal can be made from a city file ending in 1940; no airport splice or station average is used. Printed July rain is 3.59 inches; rounded daily data give 3.60 inches. Both are retained separately. Cloud counts are read directly from the scans, with the maximum-wind date column excluded. No object-linked weather witness, full making history or open artwork-image licence has been established. The original letter and 1913 Tarkington catalogue have not been inspected.

Step 1 · Date the question before testing it

Catalogued 1931, sold by 7 April

The Birmingham Museum of Art dates Abbott Fuller Graves’s A Kennebunkport Doorway (Peace and Sunshine) to 1931. Its provenance records a purchase by Dr. Anthony in a Graves letter dated 7 April 1931. The picture had therefore been sold before summer 1931.

The museum’s wording is purchased by Dr. Anthony, 1931; note 1 identifies a letter by Graves dated April 7, 1931, cited in later correspondence. That provenance supplies the sale boundary; the original letter has not been inspected. The catalogue year stays attached to the object.

Summer 1930 is the latest summer the painting could show. Nephos uses 1 June–30 September 1930 as a broad summer garden envelope, following the reading of a sunlit garden in bloom. The months are a comparison choice, not a botanical date. Graves could have used an earlier summer or worked from sketches. Neither an exact doorway nor an outdoor-to-studio sequence is established for this canvas.

Vose Galleries’ biography places Graves in Kennebunkport from 1891 and describes his outdoor garden painting by 1905. It also reproduces Booth Tarkington’s 1913 description of his gardens: a bit of a dream in each of them. That is general reception, not testimony about this later picture or its weather.

Step 2 · Look for the light

Clear days were a recurring opportunity

The Weather Bureau’s Portland tables classify 52 of 122 days as clear in June–September 1930: 42.6%. The other categories contain 41 partly cloudy and 29 cloudy days. A sunlit garden was a plausible regional opportunity in this latest possible summer, though clear was a minority category.

Portland’s published cloud categories, June–September 193052 clear, 41 partly cloudy and 29 cloudy days. Segments show monthly counts, not the order of days or sunshine hours.DAYS CLASSIFIED AT PORTLAND · 1930JuneClear: 16 days16Partly cloudy: 5 days5Cloudy: 9 days9JulyClear: 12 days12Partly cloudy: 12 days12Cloudy: 7 days7AugustClear: 10 days10Partly cloudy: 15 days15Cloudy: 6 days6SeptemberClear: 14 days14Partly cloudy: 9 days9Cloudy: 7 days7ClearPartly cloudyCloudy
US Weather Bureau, Monthly Weather Review, June, July, August and September 1930, Table 1, Portland, Me. The original page images were read and the columns checked. Segments group counts within a month; they do not show the dates or order of the days.
Published cloud classifications, Portland, 1930.
MonthClear daysPartly cloudyCloudyMean cloudiness, tenths
June16594.6
July121274.5
August101564.6
September14974.2

These are the publication’s categories, not sunshine hours. The January guide describes personal daytime cloud observations for its clear-sky chart; it does not establish the numerical daily-category cutoffs used here. No conversion to hours or claim that every clear day was sunny throughout is made. The monthly totals cannot identify a day Graves painted.

Step 3 · Test dryness in the Maine record

A relatively dry summer, with rain

Portland city records 247.2 mm (9.73 inches) in the same 122-day season, rank 9 for dryness among 44 complete seasons in its archive. Their median is 299.3 mm. This Maine gauge supports a relatively dry season; it still records at least 1 mm on 33 days.

Portland precipitation, June–September: 1930 in the complete-season record1930 records 247.2 mm, rank 9 for dryness among 44 complete matching seasons. Crosses mark excluded years, not dry summers. Red highlights 1930.PORTLAND CITY · JUNE–SEPTEMBER · PRECIPITATION, MM0100200300400500×1871: incomplete; excluded1871×1872: incomplete; excluded×1873: incomplete; excluded×1874: incomplete; excluded×1875: incomplete; excluded×1876: incomplete; excluded×1877: incomplete; excluded×1878: incomplete; excluded×1879: incomplete; excluded×1880: incomplete; excluded1880×1881: incomplete; excluded×1882: incomplete; excluded×1883: incomplete; excluded×1884: incomplete; excluded×1885: incomplete; excluded×1886: incomplete; excluded×1887: incomplete; excluded×1888: incomplete; excluded×1889: incomplete; excluded×1890: incomplete; excluded1890×1891: incomplete; excluded×1892: incomplete; excluded1893: 245.4 mm1894: 279.7 mm1895: 306.8 mm1896: 444.0 mm1897: 288.3 mm1898: 333.6 mm1899: 281.8 mm1900: 234.5 mm19001901: 269.0 mm1902: 366.9 mm1903: 303.5 mm1904: 285.0 mm1905: 377.9 mm1906: 351.3 mm1907: 433.3 mm1908: 191.5 mm1909: 275.2 mm1910: 269.1 mm19101911: 326.8 mm1912: 208.9 mm1913: 236.6 mm1914: 253.7 mm1915: 484.1 mm1916: 349.7 mm1917: 454.6 mm1918: 403.7 mm1919: 225.6 mm×1920: incomplete; excluded19201921: 257.4 mm×1922: incomplete; excluded×1923: incomplete; excluded1924: 295.1 mm1925: 214.1 mm1926: 194.7 mm1927: 363.3 mm1928: 328.9 mm1929: 259.6 mm1930: 247.2 mm19301931: 368.1 mm1932: 386.1 mm1933: 332.2 mm1934: 349.4 mm1935: 415.6 mm×1936: incomplete; excluded1937: 271.5 mm1938: 445.6 mm1939: 318.3 mm1940: 289.4 mm19401930: 247.2 mm · dashed line: median 299.3 mm · × incomplete
NOAA GHCN-Daily, Portland city, USC00176902. Every bar uses June–September only. A missing or rejected daily amount excludes the whole season; crosses show those exclusions. Trace amounts add no quantified depth. The record is not homogenized, and this comparison is not a climate trend or a formal drought diagnosis.
Portland’s daily-derived monthly amounts and temperatures, June–September 1930.
MonthPrecip., mmPrecip., in.Days ≥1 mmMean, °F
June49.91.96866.5
July91.43.601068.8
August66.52.62965.7
September39.41.55664.7

The catalogue location is 36.7 km from NOAA’s modern Kennebunkport gauge, used only as a town reference. The house is unlocated. A closer gauge at Mount Agamenticus, 24.6 km from that reference, has presumed-zero amounts in every one of these months and cannot supply a complete observed seasonal total. The stations are not pooled.

The drought reported elsewhere in the eastern United States is not assumed to describe Kennebunkport. The rank measures this station’s seasonal precipitation; soil moisture, streamflow and a local drought diagnosis are outside the evidence assembled here. Temperature is the mean of daily (high + low) / 2, with a valid pair required every day.

Every daily amount and temperature pair in the 1930 comparison
DatePrecip., mmHigh, °CLow, °C
1930-06-010.322.25.6
1930-06-020.022.88.3
1930-06-030.031.712.2
1930-06-040.033.320.0
1930-06-050.025.018.3
1930-06-060.025.617.8
1930-06-070.320.613.9
1930-06-081.317.812.2
1930-06-096.116.111.7
1930-06-10Trace13.310.0
1930-06-115.311.710.0
1930-06-120.820.011.1
1930-06-130.022.812.8
1930-06-140.028.916.7
1930-06-150.030.019.4
1930-06-160.030.019.4
1930-06-171.022.816.1
1930-06-1821.618.915.0
1930-06-19Trace21.714.4
1930-06-205.823.913.3
1930-06-211.024.416.1
1930-06-220.327.216.7
1930-06-230.023.315.0
1930-06-24Trace22.815.6
1930-06-250.028.916.1
1930-06-260.030.618.9
1930-06-276.121.717.8
1930-06-280.024.415.0
1930-06-290.023.912.8
1930-06-300.026.114.4
1930-07-013.624.415.6
1930-07-022.025.015.6
1930-07-030.022.816.7
1930-07-040.024.415.6
1930-07-050.024.416.1
1930-07-067.120.014.4
1930-07-073.322.215.6
1930-07-086.926.715.6
1930-07-09Trace26.715.0
1930-07-100.025.015.6
1930-07-111.522.814.4
1930-07-120.322.815.0
1930-07-13Trace19.415.6
1930-07-1410.720.017.2
1930-07-150.025.014.4
1930-07-160.023.913.3
1930-07-170.032.215.0
1930-07-180.028.318.9
1930-07-19Trace31.721.1
1930-07-200.028.919.4
1930-07-211.023.918.3
1930-07-220.325.018.3
1930-07-230.027.216.7
1930-07-240.025.016.7
1930-07-250.028.317.2
1930-07-260.025.017.8
1930-07-2747.819.414.4
1930-07-280.018.914.4
1930-07-296.922.816.7
1930-07-300.028.317.2
1930-07-310.024.414.4
1930-08-010.025.615.0
1930-08-020.026.115.6
1930-08-03Trace29.420.6
1930-08-040.525.621.1
1930-08-050.330.021.1
1930-08-060.028.918.3
1930-08-070.021.714.4
1930-08-080.519.415.0
1930-08-092.522.815.6
1930-08-106.621.715.6
1930-08-110.022.211.7
1930-08-129.920.010.0
1930-08-130.019.48.9
1930-08-140.020.611.1
1930-08-1517.517.215.0
1930-08-1612.418.914.4
1930-08-171.522.212.2
1930-08-180.320.612.8
1930-08-191.820.012.8
1930-08-200.018.314.4
1930-08-210.020.012.8
1930-08-220.021.113.9
1930-08-2311.717.214.4
1930-08-240.021.114.4
1930-08-250.022.215.6
1930-08-260.026.115.6
1930-08-270.028.916.7
1930-08-280.027.816.7
1930-08-291.024.414.4
1930-08-300.022.215.0
1930-08-31Trace20.013.9
1930-09-013.323.915.6
1930-09-020.027.817.8
1930-09-035.331.117.8
1930-09-040.022.213.9
1930-09-050.020.611.1
1930-09-060.025.615.0
1930-09-070.021.113.9
1930-09-080.021.712.8
1930-09-090.020.08.9
1930-09-100.018.310.0
1930-09-110.019.48.3
1930-09-120.024.416.7
1930-09-13Trace16.713.9
1930-09-140.520.012.8
1930-09-150.024.417.8
1930-09-160.026.117.2
1930-09-173.627.218.3
1930-09-180.023.315.6
1930-09-190.021.711.1
1930-09-200.021.710.0
1930-09-218.621.716.1
1930-09-220.026.115.6
1930-09-230.020.012.8
1930-09-240.020.09.4
1930-09-250.029.417.8
1930-09-260.027.817.2
1930-09-2716.820.615.6
1930-09-280.020.610.6
1930-09-290.017.28.9
1930-09-301.317.88.3

Raw values, flags, source lines and every accepted or excluded comparison season are in the downloadable data. The printed July rainfall is 3.59 inches; the rounded daily file sums to 3.60 inches. Those small differences are retained separately rather than silently reconciled. No modern normal is calculated from a city file ending in 1940.

Step 4 · Keep the summer conditional

The latest possible summer is not the painted summer

The sale closes the door on summer 1931. The 1930 records show a relatively dry regional season with recurring clear conditions, so the title’s sunshine is plausible within that latest possible summer. An earlier summer remains equally possible on the dating evidence. The records cannot establish the light at the doorway or show that Graves changed observed weather.

The museum’s original April letter and object file could narrow the history further. No object-linked weather diary, newspaper account or dated painting session has been established. The original 1913 Tarkington catalogue was not inspected; Vose transmits its short passage.

The artwork is linked to the museum rather than reproduced. A work-specific open image basis and its United States publication/registration history remain unresolved. The museum’s Open Content policy does not automatically license this canvas.

Sources

Every source on this page

Weather calculations use cached NOAA files checked against sha256 fingerprints. Museum and policy receipts identify browser or web-tool text extracts; other receipts identify downloads or HTML text extractions. Quoted phrases are checked against their cached texts. Fingerprints preserve the evidence used; they do not independently validate its interpretation.

  1. Birmingham Museum of Art, AFI.104.2015Museum text, copyright retained; short citation excerpts only · retrieved 15 October 2026sha256 76456f05582d1957ccfd0977a8f6dc2f8e3e61c74cdbe4de85ef4b54f23da69aBrowser extraction of main.textContent, including the collapsed More section; not an HTML download.
  2. Vose Galleries, Abbott Fuller Graves biographyCopyright Vose Galleries; research cache and short excerpts only · retrieved 15 October 2026sha256 363dc13a1755d7cc2cbe4a1ac0d0ad4326f01a3ab9b903708a1d907e9e0e2c22HTML text extraction; retrospective biography, not this object’s making history.
  3. Birmingham Museum of Art, Collections Access PolicyCopyright Birmingham Museum of Art; short research extract only · retrieved 15 October 2026sha256 794d18c72d265e82135e6216875b89348135db25355784263732de25cba88769Saved web-tool excerpt and explicitly labelled research note; no image permission inferred.
  4. NOAA GHCN-Daily station metadataNOAA open data; US federal observations · retrieved 15 October 2026sha256 428eabc955883187935bce283766d5646e3ed7484b92eb0803874c0e8dd9b342Coordinates are catalogue locations, not independently verified 1931 instrument exposures.
  5. NOAA GHCN-Daily element inventoryNOAA open data; US federal observations · retrieved 15 October 2026sha256 c33ec006c4f35d0371e6f9ccfe3f656ae9de0b6f0f122794da07df7c971c48b5Start/end years do not guarantee complete daily coverage.
  6. NOAA GHCN-Daily format and flag definitionsNOAA open documentation · retrieved 15 October 2026sha256 0e4e6f0e6947cf1add884c328a01760f31c56ef5fb86b85ed57dd5dc4c4aeaa6Fixed-width daily format; trace, quality and missing-presumed-zero flags retained.
  7. NOAA GHCN-Daily, Portland city, USC00176902NOAA open data; US federal observations · retrieved 15 October 2026sha256 a7850acbddcc6edde5f32347110bb88cb6e74dad65d385aef148e1a476fbece41871–1940 archive; no airport splice. Rainfall includes melted snow.
  8. NOAA, GHCN-Daily dataset documentationNOAA public documentation · retrieved 15 October 2026sha256 779e1bc09b567891789b8698207d38fa745b0e76d596c7c463246daf7d283b3dDataset DOI 10.7289/V5D21VHZ.
  9. NOAA, Open Data DisseminationUS federal public information · retrieved 15 October 2026sha256 f7085a80d8d74a89e2d4f61d14502751e45f608d81fe3ef383a554fd4c372520Open data context.
  10. NOAA GHCN-Daily, Mount Agamenticus, USC00175530NOAA open data; US federal observations · retrieved 15 October 2026sha256 6316cccf0cc90ea4132af5c739d29419acf2bf7819debac17da1d9c7d408b176Nearer screened gauge, not used for totals: every June–September 1930 rain month contains missing-presumed-zero flags.
  11. US Weather Bureau, Monthly Weather Review, June 1930, Table 1, p. 269Public-domain US Weather Bureau federal observations; Internet Archive scan · retrieved 15 October 2026sha256 84c309370368fcc9634cfde7ab43cac79544b029c8f47b5bb6abef92233e7c06Original page image inspected, including station label, column headers, wind-date boundary and cloud counts. Not OCR-derived numerical evidence.
  12. US Weather Bureau, Monthly Weather Review, July 1930, Table 1, p. 309Public-domain US Weather Bureau federal observations; Internet Archive scan · retrieved 15 October 2026sha256 6667e1cd86a7db6c89dbc21b57e7685245caf179766f59ab4986c6a9fd611ab2Original page image inspected, including station label, column headers, wind-date boundary and cloud counts. Not OCR-derived numerical evidence.
  13. US Weather Bureau, Monthly Weather Review, August 1930, Table 1, p. 346Public-domain US Weather Bureau federal observations; Internet Archive scan · retrieved 15 October 2026sha256 39a379ec5f7ee243bceb729de2ff63a6b6f725cfcf024af776fcb499e2cbe51bOriginal page image inspected, including station label, column headers, wind-date boundary and cloud counts. Not OCR-derived numerical evidence.
  14. US Weather Bureau, Monthly Weather Review, September 1930, Table 1, p. 388Public-domain US Weather Bureau federal observations; Internet Archive scan · retrieved 15 October 2026sha256 7e1a3c5ebb55368b92560507e8f9bc18e51b9806e44009753813b46f3be1ceb6Original page image inspected, including station label, column headers, wind-date boundary and cloud counts. Not OCR-derived numerical evidence.
  15. US Weather Bureau, Monthly Weather Review, January 1930, Description of Tables and Charts, p. 37Public-domain US Weather Bureau federal publication; Internet Archive scan · retrieved 15 October 2026sha256 b9df56b9de78bb5ad3085851c4b0a2d9857ae5010c8d44b18442ffb2a27897ccTable 1 observation schedule and Chart IV daytime cloud-estimate description; category thresholds and sunshine hours are not inferred.
  16. Nephos scan-checked transcription, Portland cloud classifications, June–September 1930Nephos transcription of public-domain US Weather Bureau numerical tables · retrieved 15 October 2026sha256 b32aec0954da0593673c64e7caae8ff83961803600eec7aa0caf5f582a0b49c2Derived transcription. Each row identifies its original scan, page, locator and sha256. Four separate scan receipts preserve the originals.

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