Which way Romania's churches face: towards the sunrise, mostly the sunrise of spring and summer

All at once

Three in four face a sunrise#

77% of the churches face into the arc of sunrises

Each line of light is a church, drawn from the centre towards its altar. Gathered at one point, the 8,377 lines open into a fan facing east.

The gold band on the rim is the arc where the sun rises over a year, from the midsummer sunrise to the midwinter one. 77% of the churches face somewhere inside it, computed for each one over the real terrain in front of it. A direction picked at random would land there about one time in five; among the churches whose altar end shows in the outline, 85% of the Orthodox ones do.

Where the sun rises

The sunrise travels along the horizon through the year#

In June, at Romania's latitude, the sun rises to the north-east, about 54 degrees from north. In December it rises to the south-east, at about 124. Due east, at 90 degrees, is where the sun rises only around the equinoxes, in March and September.

The months on the rim show where the sun rises in each of them. The colour of each line tells which season's sunrise the church faces: gold for summer sunrises, blue for winter ones; those facing due east sit between them.

An old rule

The Orthodox keep the rule; Neo-Protestant churches never had it#

84% of Orthodox churches face into the arc of sunrises

The Apostolic Constitutions, one of the oldest collections of Christian church orders, ask that a church be "long, with its head to the east". The Romanian Orthodox Church still holds to it: in 2011 a wooden church in Galați was lifted and turned round on the spot because its altar faced west.

Of the 5,350 Orthodox churches measured, 84% face into the arc of sunrises. Pentecostal, Baptist and Adventist churches, which have no such rule, land in it only 45% of the time, and their lines scatter. Roman and Greek Catholic churches reach 46%; Reformed, Lutheran and Unitarian ones 71%.

North of east

Old churches in Muntenia and Moldavia face the sunrises of spring and summer#

79% of Orthodox churches built before 1900 face north of east

If churches faced due east with small random errors, as many would point north of east as south of it. They do not. Of the Orthodox churches facing within 45 degrees of east (90% of them all), 66% lie north of east; their median sits 5.7 degrees to the north.

Those built before 1900, with a known date (788, eight in ten of them in Muntenia and Moldavia, where map tags most often carry the year), lean further: 79% face north of east. 92% lie inside the arc of sunrises. Only 1.0% point beyond the midwinter sunrise, and 6.6% beyond the midsummer one. A church facing between due east and the midsummer sunrise faces where the sun rises on some day between March and September.

On the map

The northward lean is strongest in Muntenia, Moldavia and Dobruja#

In Muntenia, the Wallachian plain, the median Orthodox church faces 10.0 degrees north of east, and 75% of them lean north. In Moldavia, 6.8 degrees and 70%. In Dobruja, 3.9 degrees. In Transylvania, in Oltenia and in Crișana and Maramureș the median sits within a degree and a half of east, and only 55%, 54% and 54% face north of it.

Large published samples from other countries find smaller leans. In England, 183 early medieval church buildings average 2 degrees north of east, 54% of them to the north. Across 569 medieval churches of Catholic Europe, the mean is 3 degrees north. In Iberia, 60% of 167 early churches face north of east. Only a small group, also Orthodox, comes close to Muntenia and passes it: 23 old churches in Serbia north of the Sava and the Danube: the median of the angles in Tadić and Nešić's tables (2013) is 16.5 degrees north of east.

Four centuries

New churches face closer to due east#

Here each church with a known date (from its map tag or from the national List of Historic Monuments) sits at the year it was built, higher if it faces north of east, lower if south. The white line runs through the median of the Orthodox churches in each quarter-century.

The median Orthodox church of the 19th century faces 12.5 degrees north of east, 83% of them to the north. Those built after 1990 sit at 1.7 degrees, with 59% of them to the north. The shift also shows inside the regions with enough dated churches, not only across the mix: in Muntenia, from 16.2 degrees in the 19th century to 1.8 after 1990; in Moldavia, from 11.5 degrees before 1800 to 3.7.

First explanation

It is not the compass#

A compass needle does not point to true north, and its error drifts from century to century. At Bucharest in 1800, magnetic north lay 13.1 degrees west of true north; today it lies 6.4 degrees east. A church set out by compass towards "east" would have faced about 13.1 degrees north of east in 1800 and 6.4 degrees south of it today. The dotted line shows where it would sit.

The churches do not follow it. Before 1900, while magnetic north swung by more than ten degrees, their lean stayed the same: the slope against the compass error is −0.3, with an interval from −0.6 to 0.0, in effect zero. After 1900 the churches' lean and the compass error both drift towards east in the same years, so those years cannot tell them apart; but today's churches sit close to east, not south of it, where compass-laid ones would sit. A study of churches in Austria and northern Germany reached the same conclusion (Arneitz and colleagues, 2014).

Second explanation

Nor the patron saint's day#

It is often said that the altar faces the point where the sun rises on the patron saint's day. For 2,577 churches the name gives the patron, and the feast falls on a fixed date. Here each sits, across, at the direction of sunrise on its saint's day and, up, at the direction it faces. If the rule held, the churches would gather on the diagonal.

They do not. For 142 of them, the sun rises on the saint's day within 3 degrees of the axis. With the saints shuffled at random among churches of the same region, 2,000 times, the matches come to 137 on average. On the 416 listed historic monuments, using the calendar of the time (Julian for Orthodox churches built before 1924, when the Romanian Church changed calendars), there are 52 matches within 5 degrees against 45 by chance. Depending on the calendar assumed, luck produces such a gap between one time in ten and one time in fifty: at most a faint trace, far too small to explain the lean. Studies in Serbia, in Austria and Germany and in Iberia found no link either.

Third explanation

Nor the day the foundation was laid, on the few known cases#

Another explanation: the priest fixed the axis on the sunrise of the morning the foundation stone was laid. The rite, as printed in one Romanian service book, asks for a cross planted where the altar table will stand, but gives no instruction to aim the axis at that day's sunrise. The foundation date is rarely known. For 25 Orthodox churches it is written in the dedication inscription, in articles or in the church press, and the church has an outline on the map.

The sunrise on the foundation day misses the axis by a median 25 degrees; with the dates shuffled among them, by 32. The real dates do a little better than shuffled ones, but luck gives such a gap about one time in ten; the cases are too few to tell, and many are city cathedrals. It remains an untested hypothesis that the northward lean comes from the building season, spring and summer. For a few Serbian churches, the astronomer Milutin Tadić linked the axis in the same way, to "the beginning of the construction season".

Where the map misleads

How sure each line is#

The axes come from each church's outline, drawn by volunteers on OpenStreetMap. For 1,177 churches there is also an independent outline, extracted automatically from satellite imagery by Microsoft. The two axes differ by a median 3.1 degrees, and in nine cases out of ten by less than 8.9; the mean difference is 0.6 degrees, so it leans neither way. Such errors cancel out on average: they cannot move the whole compass north.

Where the houses around form a regular grid, 29% of churches lie within 3 degrees of the direction of the houses, four times as often as chance. Those that ignore the houses around them lean even further north: 6.4 degrees. The altar end is read from the outline, by its rounded or chamfered apse, for 5,425 churches; for 2,952 plain rectangles the eastern end is assumed. The mosques of Dobruja have an exact target, Mecca, yet the 39 measured miss it by a median 14 degrees: builders aim loosely even at a precise direction.