The continents move. When you rewind them, the ancient sites line up more precisely — peaking at 20 million years ago.
Using pyGPlates with the Müller et al. 2022 rotation model, we reconstructed the positions of 68 ancient sites at multiple geological time periods. The results show the network was positioned for an older version of Earth — not the one we inhabit today.
The continents sit on massive slabs of rock called tectonic plates, which float on the semi-molten mantle beneath. These plates move at roughly 2 centimetres per year — the same rate your fingernails grow. Over a million years that's 20 kilometres. Over 20 million years it's 400 kilometres. Over 50 million years it's a thousand kilometres. Using GPS satellites we can measure this movement with millimetre precision right now, and using geological software we can project those movements backwards through time to calculate where any point on Earth was located at any date in the past — or future.
Click any time period below to see how the ancient sites shift position as the plates move. Each colour group represents a different species' site cluster. Notice that at 20 million years ago — the peak period — the sites form more precise geometric patterns. The Z-score and alignment count update to show how statistically significant the pattern is at each period.
At 20 million years ago, three sites form a near-perfect equilateral triangle: Nabta Playa (Egypt), Nan Madol (Micronesia), and Tenochtitlan (Mexico).
The three sides measure: 12,395 km / 12,424 km / 12,425 km. A difference of just 30km across 12,400km — precision of 0.24%. These sites are on three completely separate tectonic plates moving in different directions. The probability of this occurring at a specific date by chance is extraordinarily small.
During the Miocene epoch, the Earth's magnetic field was unusually stable. The Drake Passage between South America and Antarctica had recently opened, establishing the circumpolar current. Africa was moving northeast, Europe was largely tropical, and the Mediterranean was beginning to close. The continental configuration at this date produced the most geometrically precise relationships between ancient site locations of any period we tested — past or future.
We also projected site positions forward in time — which nobody had done before. The results were unexpected: the alignment precision doesn't simply degrade going forward. It improves again, peaking at 100 million years in the future when 17 sites align on one great circle — the highest count of any period, past or future. And at 10 million years in the future, three sites form the most precise triangle in the entire dataset — more precise even than the 20 million year past result.
| Period | Best Great Circle | Best Triangle Score | Notable |
|---|---|---|---|
| 50 Ma past | 12 sites | 0.082 | Major continental differences |
| 20 Ma past ★ | 15 sites | 0.0024 | Peak past alignment — Nabta/Nan Madol/Tenochtitlan |
| 10 Ma past | 14 sites | 0.0071 | |
| Present | 15 sites | 0.0109 | Current configuration |
| 10 Ma future ★ | 16 sites | 0.0016 | Most precise triangle — Jericho/Nan Madol/Caral |
| 50 Ma future | 16 sites | 0.0089 | Mediterranean fully closed |
| 100 Ma future ★ | 17 sites | 0.0093 | Peak great circle — 17 sites in alignment |
The sites don't just look backwards to a better-aligned past. They maintain geometric significance across an enormous sweep of time in both directions. This suggests the sites were positioned at locations that are geometrically special not for one epoch but across hundreds of millions of years. Whatever positioned these sites understood not just where the plates were — but where they were going.
The East African Rift Valley completes its split. Somalia, Ethiopia and Kenya become a separate island continent. A new ocean forms. This shifts the African plate sites significantly — yet the alignment between Jericho, Nan Madol and Caral reaches its most precise triangle of any period tested.
Africa fully merges with Europe. The Alps become dramatically taller. The Mediterranean Sea disappears. Australia collides with Southeast Asia. Egyptian sites end up in what is now Scandinavia. Stonehenge approaches the North Pole.
All continents merge into the next supercontinent. Ancient sites currently spread across the globe are brought physically close together again. Alignment precision likely peaks again — a second geometric optimum, 250 million years from now.
At 10,500 BC — the date when all four major sites achieve perfect star alignment — sea levels were approximately 60 to 70 metres lower than today. The last Ice Age was ending, the ice sheets were melting, and sea levels were rising rapidly. This means enormous areas of land now underwater were above sea level and potentially inhabited. Our analysis only includes currently above-water sites. If any sites existed on what is now the sea floor, we are missing part of the network.
Several documented underwater structures match our predicted locations for network nodes. The Yonaguni monument off Japan was above sea level at 10,500 BC. The Dwarka ruins off India's coast were above sea level. The Gulf of Cambay structures in India have been carbon dated to approximately 9,500 BC — above sea level at that date.
Our alignment analysis was run including nine known or suspected underwater site locations at their 10,500 BC positions — above sea level coordinates. Adding these sites increased the total alignment count and strengthened the convergence signal toward the Antarctic and Pacific nodes.
How four sites achieve perfect star alignment simultaneously at 10,500 BC — and why 2,500 BC produces the wrong answer.
→ The MysteryAlignment lines converge on East Antarctica. The Heritage Range location peaks at 34 million years ago — the moment the continent froze.
→ Predictive ScienceThe mathematics identifies specific unoccupied nodes. One was confirmed in 2025. Northern France, Libya, Saudi Arabia, the Pacific.
→