The wall that faces the rain RO
  1. Bukovina, Romania · seven painted churches · 27 walls

    The north walls of Bukovina’s painted churches: rain, shade and how much paint survives on each wall

    In the north-east of Romania, a handful of sixteenth-century churches are painted outside as well as in, wall to wall, with scenes that have stood in the open for almost five hundred years. Guides say rain driven by northern winds washed the north walls bare. Here the explanation is tested on 27 painted walls of seven churches: how much of each wall still has paint, and how much wind-driven rain it takes, calculated from hourly data for 1991–2020.

    Scroll: walk around the church ↓

  2. Moldovița · south wall

    The sunny side kept almost all its paint#

    On the south wall of Moldovița church, painted in 1537, the Akathist Hymn and the Siege of Constantinople are still nearly whole. Two estimates made separately, from different photographs, show paint on 90% of the surface.

  3. Moldovița · the apse

    Towards the east, less survives#

    On the altar apse, paint covers 65% of the surface. High up, under the eaves, the saints are clear; lower down, large patches of colour are gone.

  4. Moldovița · north wall

    The north wall kept about a third#

    On the north side, what survives is mostly the band under the eaves; below it the plaster is white. Paint covers about 35% of the surface.

    The difference is old. In his book of 1930 the art historian Paul Henry wrote that on the left half of Moldovița’s north wall almost all the scenes were very faded (p. 227 of the Romanian edition).

  5. Moldovița · the west porch

    On the porch piers, about half#

    On the piers of the open west porch, exposed on three sides, paint covers 50%. The porch vault, sheltered from rain, is among the best-preserved parts of the church (Šeme); it was not included.

  6. Where the rain comes from

    The north wall takes 4 times the wind-driven rain of the south wall#

    How much rain strikes a wall can be estimated with an index from the standard ISO 15927-3: the index combines, hour by hour, wind speed, the amount of rain and the angle between wind and wall, for the hours when the wind blows towards the wall (the full calculation is in the method section). Here it was done for every hour of 1991–2020, with data from the ERA5 weather reanalysis and each church’s real orientation.

    At Moldovița, rain comes mostly on north-westerly and northerly winds. The sectors around the plan show the directions the rain came from: the longer, the more.

  7. What weather stations say

    On the plateau the stations confirm the direction. In mountain valleys, the wind follows the valley#

    A model with cells of about 30 km can get local wind wrong. So the direction was checked against reports from eight weather stations in the region (the NOAA ISD archive), in the hours when the observer recorded rain. At Rădăuți, Suceava, Botoșani, Cotnari and Târgu Neamț, in all 18 ten-year windows with enough reports since 1973, a wall facing north and one facing west would together have taken about two thirds or more of the rain-bearing wind; one facing south, under a sixth. In the same hours ERA5 gets the direction right on the plateau, with a small bias: it gives the west 0.29–0.31 of the wind, against 0.34–0.40 at the stations.

    The mountain stations differ. At Câmpulung Moldovenesc the wind blows from the west, along the valley, and 84% of the rainy observations (1973–2001) were logged as calm. The churches of Voroneț, Humor, Moldovița and Sucevița stand in valleys like this, with no station of their own. For them, the rain direction from ERA5 is an approximation.

  8. All seven churches

    On 27 walls: the more rain, the less paint#

    Each dot is a painted wall. Across, its wind-driven rain index; up, how much of its surface still has paint, the mean of two estimates whose assessors did not know which long wall faces north and which south.

    The north wall has the highest rain index at all seven churches. At five of the six churches where one wall lost much more paint than the rest, that wall is the one most exposed to rain; at all six, the most sheltered wall is also the best kept (differences of 5 points or less count as ties). On the direction, the guides are right.

  9. Another explanation: no sun

    The same wall gets very little sun#

    Rain comes mostly from the north-west, and the sun shines from the south. North walls get between 12 and 196 hours of direct sun a year, most at the churches turned towards the north-east (Arbore and Sucevița); south walls more than 1,700. If the chart uses hours of sun instead of rain, the picture turns around: walls with more sun kept more paint.

    The north wall also has the highest index for wind-driven snow, 5–9 times that of the south wall. And in the 24 hours after a rain it receives about half the solar energy the south wall does, which probably slows its drying (drying itself was not measured). On these churches the wall most exposed to rain is also the most shaded and the most exposed to snow, so their walls cannot separate the effects.

  10. Humor · north wall, by height

    More survived under the eaves than at the foot of the wall#

    On Humor’s north wall, paint still covers 55% of the top third and 5% of the bottom third. On the south wall the gap is much smaller (90% against 65%).

    The eaves keep rain off the top. But the foot of a wall is also where snow piles up, where drops bounce off the ground and where damp rises from the foundations. At Humor in 1972, the highest moisture in the outer walls was measured on the north façade, just above the plinth (Istudor, 1973).

  11. The exception

    At Sucevița, the wall most exposed to rain kept its paint#

    Sucevița church is turned further to the north-east than the others, so its north wall faces north-north-west, the direction most rain comes from. Among the valley churches it is the wall with the highest rain index. Yet paint covers 95% of it, with the Ladder of Virtues still legible. In 1930 Paul Henry called it the best-preserved north wall in all of Bukovina (p. 250 of the Romanian edition).

    The UNESCO dossier credits its high enclosure walls, set in front of the church: “cca 9m” in the text (p. 63), “cca 6m” in the same dossier’s inventory sheet (p. 367). But Moldovița also has an enclosure wall 6 m high on its north side (Balș), closer to the church (about 20 m from the north wall, against about 40 m at Sucevița, measured on aerial imagery), and most of the paint on its north wall is gone.

    Turned towards the north-west, Sucevița’s north wall catches the evening sun in summer: about 200 hours of direct sun a year. But Arbore is turned the same way; its north wall gets 125 hours of sun and the highest rain index of the whole set, and its paint is almost gone (5%). Two other differences are documented. Sucevița stayed a working monastery without a break (p. 52), while the monasteries of Voroneț, Humor and Moldovița were dissolved in 1785 and their churches became parish churches (p. 529); but St George in Suceava also stayed in a monastery’s care, as a dependency of Dragomirna, and very little paint survives on its north wall. And Sucevița was painted after 1595, with the work finished in 1596 or around 1601, about half a century after Voroneț (1547).

    A hill rises 13° above the horizon to the north, 0.7 km away, but Voroneț has one too (11°, 0.5 km away). How much each of these factors contributed has not been measured. Sucevița shows only that orientation alone does not explain the loss.

  12. Photographs from 1942 and 1943

    The loss was already visible 80 years ago#

    The photographer Willy Pragher passed through Bukovina in 1942 and 1943. Scored by the same method, his pictures show paint on 45% of Voroneț’s north wall and 85% of Sucevița’s; in recent photographs (2014–2021) the estimates are 55% and 95%. At Moldovița, in a photograph from 1975: 30%, against 35% in the photographs of 2015 and 2026.

    So the north walls had lost much of their paint before the roof works of the 1960s and 1970s; at Voroneț, Pogonat and Boldura date the widening of the eaves to 1975 (RMM-MIA, 1989). In 1928 the historian Ioan D. Ștefănescu noted that the north walls outside had generally suffered a great deal (p. 207). The climate data of 1991–2020 are only a stand-in for the centuries when the paint was lost.

  13. The test at Probota and Suceava

    The prediction held at Suceava; at Probota, only without the west end#

    For Probota and St George in Suceava, the expected order of the walls was written down before any estimate of their paint existed. Suceava came out as predicted: the north wall has the least paint (5%), the south the most (55%). At Probota, by the rule written in advance, the prediction failed: the least paint is on the west end (0%), not the north wall (10%). But that end is a wall with large windows that the photographs cannot show was ever painted; without it, the north wall has the least paint and the south the most (60%), as predicted.

    How far the two estimates of the same wall differ: 4.3 percentage points on average, 15 at most. That measures how well they agree, not how far both may be from the truth.

  14. What remains

    The rain does come from the north, but it does not explain the loss alone#

    The guides have the direction right: the north wall takes 3.9–5.5 times the wind-driven rain of the south wall and has the least paint at five of six churches. But the same wall also gets the least direct sun and has the highest wind-driven snow index. And at Sucevița, on a wall facing north-north-west, the paint has lasted four centuries.

    Why it lasted there is still open. A year of measurements beside the north walls of Sucevița and Moldovița could show how much rain and sun actually reach each.

Which wall of your house takes the most rain?#

The same calculation, for any place: hourly wind and rain data for 2015–2024 (NASA’s MERRA-2 reanalysis) and the wind-driven rain index for walls facing each of the eight compass points.

This is for open ground, from data with a resolution of about 50 km. A narrow street, a hill or the house next door can change a single building’s exposure a great deal.

All 27 walls#

ChurchWallSurface with paintTop / bottomWind-driven rain indexDirect sun (h/yr)
Humornorth25%55% / 5%23327
Humorwest25%50% / 10%152986
Humorapse (east)65%85% / 40%1051,204
Humorsouth80%90% / 65%591,814
Moldovițanorth35%70% / 10%26213
Moldovițawest50%70% / 25%1541,127
Moldovițaapse (east)65%85% / 50%119991
Moldovițasouth90%90% / 85%661,810
Voronețnorth55%75% / 35%24012
Voronețwest95%95% / 90%1241,028

How the estimates were made#

Paint. For each painted wall, photographs from Wikimedia Commons showing as much of the wall as possible were chosen, usually two by different photographers; for Humor’s north wall only one was found, and many show only part of the wall. Two separate assessments estimated, for the top, middle and bottom third of the wall, what share of the surface still shows paint (however faded), in steps of 5 percentage points. The assessors did not know which long wall faces north and which south: photos were shuffled and both were called just “the long side wall”; west ends and apses were named as such. Windows, doors and surfaces under vaults are left out. The percentages on the page are the mean of the two estimates, rounded to 5. A photograph cannot show what pigment survives under plaster or whitewash, nor tell all retouching apart.

Rain. The wind-driven rain index follows ISO 15927-3: for each hour when the wind blows towards the wall and the temperature is above 0.5 °C, wind speed, the amount of rain to the power 8/9 and the cosine of the angle between the wind direction and the wall’s normal are multiplied; the hourly values are added, multiplied by 2/9 and divided by the number of years. Data: hourly ERA5, 1991–2020 (via Open-Meteo), one 0.25° cell per church. Each church’s orientation comes from its OpenStreetMap outline, checked on aerial imagery.

Sun and snow. Hours of direct sun are the hours when ERA5 direct radiation, projected onto the wall from the sun’s position, exceeds 120 W/m², after the real horizon from the Copernicus GLO-30 terrain model is applied. The wind-driven snow index is the same calculation for hours at or below 0.5 °C.

Stations. The direction was checked against synoptic reports in the NOAA ISD archive: the wind logged in hours with rain was compared with ERA5 at the same hours. At Rădăuți, Suceava and Botoșani, the share that would reach a north-facing wall is 0.34–0.44 by the stations and 0.40–0.43 by ERA5; for a west-facing wall, 0.34–0.40 by the stations and 0.29–0.31 by ERA5, so the model slightly underestimates the west. At Câmpulung Moldovenesc, in the valley, the west share is 0.55 by the station and 0.29 by ERA5.

What cannot be said. The paint was lost over centuries, in which roofs fell, plaster went on and came off, and restorations were made; today’s climate data are only a stand-in for the conditions then. On these churches the walls most exposed to rain also get the least sun, so the analysis cannot separate the two effects. Estimates from photographs are no substitute for a conservator’s survey.

Sources

Photo (58)