Pumapunku: the precise stones of Tiwanaku

There is no single night at Pumapunku. There are centuries, and then there is damage.

What the record gives first is the place. Pumapunku is a smaller stepped platform renowned for oversized stone blocks and precisely cut ashlars[marsh-2023], one monument inside Tiwanaku, in Bolivia, a city founded around AD 180 that became one of the first in the Andes. UNESCO inscribed Tiwanaku as a World Heritage site in 2000[unesco-567].

Picture it, within what the papers say. In the AD 600s, according to a chronology built from 102 radiocarbon dates, the Tiwanaku are building Pumapunku and the Akapana[marsh-2023]. The same study notes that the first plan was abandoned early, and that work went on with reused stone. The blocks are andesite and sandstone, large, finely carved, with shapes that repeat. The file doesn't say who directed the work or how many hands it took.

Then the city itself fails, a rapid collapse around AD 1010–1050[marsh-2023]. Later the Inca restore and reuse the temple, according to Archaeology Magazine's report[archaeology-mag-2018].

And then the part visitors actually see. The site was ransacked over five centuries[bmc-press-2018], heavily looted in the colonial period, and none of its remaining blocks is in its original position. The scattered, thrown-about look that photographs so well is not the building. It is what was left after the building was taken apart.

That wreckage is where the modern story begins.

The paper trail at Pumapunku runs on two tracks: the archaeologists' and the ancient-astronaut shelf's. They answer each other.

The second track is older in print. In July 1987 Penn Museum's Expedition magazine published an analysis of how Erich von Däniken builds his case[epstein-1987], including his claim that figures at Tiahuanaco and on Easter Island share one style, which the author answered with a pair of photographs. In 2009 von Däniken's Twilight of the Gods argued, according to its book description, that Pumapunku could never have been built by its Stone Age inhabitants[vondaniken-2009]. HISTORY's Ancient Aliens has devoted a full episode[history-episode] to the site.

The archaeological track is built from stone and carbon. In February 2013 Jean-Pierre Protzen and Stella Nair published The Stones of Tiahuanaco, a study of the site's gateways, platforms and walls, and performed original experiments with stone tools[protzen-nair-2013] to learn how the stone was shaped and finished. The same year Yaeger and Vranich published a radiocarbon chapter on the complex; the file lists it as a reference only[yaeger-vranich-2013], because its text was not read.

In December 2018 Alexei Vranich published a reconstruction in Heritage Science, turning a century and a half of field measurements of the shattered building into 3D-printed fragments at reduced scale[vranich-2018]. The publisher's release counts 140 andesite pieces and 17 sandstone slabs[bmc-press-2018]. Vranich told Archaeology Magazine the builders seem to have made a prototype for each stone type and copied it: It's almost like it was a pre-Columbian version of Ikea[archaeology-mag-2018].

Then, in November 2023, the dating. A Bayesian chronology in PLOS ONE, built from 102 radiocarbon dates, 45 of them unpublished[marsh-2023], placed most of Pumapunku's construction in the AD 600s.

Two tracks, then. One asks who could have done this. The other went and checked.

The turn at Pumapunku is a shift from awe to method.

The fair other side starts from something true. The precision is real. Von Däniken's case, as his book describes it, is that a site at about 4,000 metres could never have been built by its Stone Age inhabitants[vondaniken-2009] and that its blocks were deliberately damaged over the centuries. To many eyes the cut, repeated shapes look machined, and because the original layout is lost, no one can show the finished building. That gap is real, and it is where the story finds room.

The record answers it in three ways. First, the date. The radiocarbon chronology puts the work in the AD 600s, within the life of the Tiwanaku city[marsh-2023], not outside it. Second, the method. Protzen and Nair tested the cuts with stone tools and came away with a new appreciation of what was achieved without metal[protzen-nair-2013]. Third, the repetition. Vranich's 3D prints suggest prototypes copied for each stone type[archaeology-mag-2018]: standard templates, which is how a disciplined workforce builds, not evidence of a machine.

And the scattered look is explained by what happened afterwards: ransacked over five centuries[bmc-press-2018]. A wrecked building invites stories.

The file's Finding is Explained. The dates and the masonry studies leave no gap for builders other than Tiwanaku's own.

What they do leave is a different, quieter set of questions.

The real mystery here is not who built Pumapunku. It is what the building looked like before it was broken.

• The original layout. Every reconstruction, including Vranich's 3D-printed model[vranich-2018], is a hypothesis built from displaced blocks, because none remains where it was set[bmc-press-2018].
• Transport. How the largest blocks were moved from their quarries is not settled in the file's sources.
• The abandoned plan. The chronology notes that the first plan was abandoned early[marsh-2023]. Why, it does not say.
• The exact tools on each block. Experiments show what is possible[protzen-nair-2013], not what was done on a given stone.
• The date range. The 2018 publisher's release gives AD 500–950[bmc-press-2018] for the building; the 2023 chronology[marsh-2023] narrows most construction to the AD 600s. The wider range is older and less precise, not a contradiction.
• Unread texts. The 2013 radiocarbon chapter[yaeger-vranich-2013] and the full Protzen and Nair monograph were not read in full, and von Däniken's original 1968 passages were not opened.

The documents and the timeline are below. Start with the 102 dates.