Nabta Playa: do the stones point at stars? Take a bearing

Nabta Playa and the claims

At Nabta Playa human presence has been shown from about 6500 to 3400 BC (radiocarbon dating, read in summaries). It has megaliths and the Calendar Circle. In 1998 Malville, Wendorf, Mazar and Schild described in Nature megaliths that may be aligned with the rising of bright stars; Sirius, the belt of Orion, Arcturus and α Centauri are mentioned.

One line of stones, the C line, is said to point at Sirius. Wendorf and Malville date that line to about 4820 BC; Brophy and Rosen, using satellite images in 2005, arrived at about 6088 BC. On this page we only check what is geometrically possible: which direction fits which year? What the builders intended, we cannot say.

Take a bearing on the line

Choose a line (a star, or the Sun at the summer solstice), set a direction (the azimuth, in degrees from north through east) and the height of the horizon. The graph shows how the rising point of each line shifts between 7000 and 3000 BC. The bar at the bottom shows the years in which the chosen line falls within our margin of 0.5° of your direction. The two buttons set the direction to that of the line in the datings of Wendorf and Malville and of Brophy and Rosen.

What we know, and what is disputed

• Established in broad terms: Nabta Playa lies at about 22.5° north in the Nubian Desert; it has megaliths and the Calendar Circle, and there is human presence from about 6500 to 3400 BC. We read this in summaries.
• Claimed by the researchers who described the site: the megalith lines point at the rising of bright stars (Malville et al., Nature 1998; Sirius, the belt of Orion, Arcturus and α Centauri are mentioned) and the C line at Sirius around 4820 BC. This is disputed.
• Disputed: Brophy and Rosen (2005) measure different stone positions from satellite images and date the alignment with Sirius to about 6000 BC (6088 BC in one summary). Malville accepts the corrections in part and writes that the C line may not be an original set of aligned stones. A 2007 article by researchers at the University of Colorado finds Brophy and Rosen’s datings about 1,500 years too early (seen only in a search result; title and authors not confirmed).
• Shifting sand and damage to the stones make alignments uncertain. We do not adopt Brophy’s broader claims (maps of the Milky Way and Orion from 16,500 to 17,500 BC): according to the 2007 article they are incompatible with the archaeological record.
• We found no measured directions of the stones in our sources, and no source for a summer-solstice alignment of the Calendar Circle. The Sun is here only so that you can see what it does, not because a source mentions it.

Our calculation

The rising azimuth of each line with a flat horizon (altitude 0°) at 22.5° north, in degrees from north through east. Stars: the center on the horizon. Sun: the upper limb, with refraction.

In 4,000 years Sirius rises about 21° farther toward the east (from 135.2° to 114.2°). In the same time the Sun at the summer solstice shifts only 0.27° (from 63.17° to 63.44°). A direction toward the Sun therefore cannot date the monument; a direction toward a star can, in principle.

What this says, and what it does not

• Between the two datings of the Sirius line (4820 and 6088 BC) the rising point of Sirius differs by about 7.1° with a flat horizon. Half a degree of azimuth is about 90 years for Sirius. A carefully measured direction could therefore help choose between the datings.
• But that needs a measurement better than about half a degree, and the height of the horizon: if it is 1° higher, the rising point of Sirius shifts by about 0.6°, which is well over 100 years. We do not have such measurements.
• Arcturus and Dubhe rise in the northeast. Their curve depends strongly on the height of the horizon and is not straight (because of the stars’ proper motion). From about 3450 BC Dubhe becomes circumpolar at this latitude and no longer rises.
• At Nabta, planets and the Moon rise only between about 57° and 123°, so the northern lines (Arcturus, Dubhe) can never be hit by them. They rise rarely or never in the direction of Sirius and Alnilam, and often in that of the Sun. Such a hit is ordinary and says nothing about what the builders intended.

How we calculate

For each star we use our model for ancient star positions (precession after Vondrák et al., spatial proper motion from Hipparcos) and calculate the declination in each year. The rising azimuth follows from the declination, the latitude and the height of the horizon, with refraction. For the Sun we use the tilt of the Earth’s axis according to P03. We compared this with an independent calculation in Python: the differences are smaller than 0.001°.

We assume a flat horizon at 22.5° north; the real horizon has not been measured. The margin of 0.5° for “the line fits” is our own assumption, not a source value. We calculate years astronomically and show them historically.

The planets and the Moon are not calculated in the browser. In advance we calculated the rising for every day from 7000 to 3000 BC with JPL’s DE441 ephemeris (via Skyfield), which we checked against JPL Horizons, and counted how often each body rises per century in each bin of 0.5° azimuth. The time scale (ΔT) is uncertain in that period, by more than a day: the counts per century hardly depend on it, but that is why we show no example dates.

What we don’t know

• We found no measured directions (azimuths) of the stones of Nabta Playa in our sources. Without them we cannot say which year or which star fits a line of stones; we only show what is geometrically possible.
• We have not measured the height of the horizon at Nabta. By default we assume a flat horizon and let you set the height yourself.
• We have not read Malville et al. (1998), Brophy and Rosen (2005) or the 2007 article ourselves, only summaries. Titles, authors and figures still have to be checked in the originals.
• We leave out α Centauri: we have no usable source for the orbit of the pair.
• That a direction fits a star says nothing about the builders’ intention. We only check what is geometrically possible.
• The dating of the C line (4820 or 6088 BC) is disputed. We did not determine it ourselves.