Start with level bedrock. Pour small foundation blocks. Then pour two big anchor stones at the corners — those are the ugly ones with cracks because they took a hundred buckets and the first bucket started curing before the last one went in. Then fill the space between the anchors with smaller pours — those are the pretty ones because they’re small enough to pour in one go. Each fill block gets its shape from its neighbors. Nobody designed the interlocking. The liquid did the fitting.
Sacsayhuamán’s walls are impossible. Stones weighing hundreds of tons, fitted so precisely you can’t slide a razor blade between them. No mortar. No uniform shapes. Every block is different, and every one fits its neighbors perfectly.
Unless they were poured.
Start with level bedrock. Pour small foundation blocks. Then pour two big anchor stones at the corners — those are the ugly ones with cracks, because they took a hundred buckets and the first bucket started curing before the last one went in. Then fill the space between the anchors with smaller pours — those are the pretty ones because they’re small enough to pour in one go.
Each fill block gets its shape from its neighbors. Nobody designed the interlocking. The liquid did the fitting. Pour against a cured neighbor and the new pour conforms to every curve, every bump, every irregularity. That’s why the joints are impossibly tight. You can’t get tighter than liquid against solid.
The predictions: corner stones should be roughest (largest volume, worst cold joints). Fill stones should be smoother (single pours). Exposed faces should bulge outward (skin against a flexible form). Joint faces should be flat (poured against a cured neighbor). The dependency graph should have no cycles — every block poured after its lower neighbors.
The limestone from the quarry 3 km away is geopolymer-compatible. Protzen at Stanford tried to replicate the concave joints by pounding stone against stone. Couldn’t do it. Liquid fills concave molds perfectly.
Nobody has run SEM on a Sacsayhuamán block. One sample would settle this.
Toggle to “the math” for the science version with pour sequence simulation.
One section between two adjacent salient corners of the first (lowest) zigzag wall. Approximately 17 meters wide, 7.5 meters tall. Based on published dimensions (Protzen 1993, Gasparini & Margolies 1980).
The colors indicate pour phase — bedrock, foundation, key stones, fills, course 2. Watch the sequence: the biggest blocks go in at the corners first, then everything else fills the gaps. The interlocking shapes emerge from the negative space between cured neighbors.
The predictions this model makes:
• Key stones (corners) should be roughest and most cracked — largest volume, most bucket loads, worst cold joints.
• Fill stones should be smoother — smaller volume, more uniform pour.
• Exposed faces should bulge outward (skin form). Joint faces should be flat (against cured neighbor or plank).
• The pour dependency graph should have no cycles — every block was poured after its lower/adjacent neighbors.
Yucay limestone. Quarry 3 km away. Geopolymer-compatible.
Protzen (Stanford) couldn’t replicate the concave joints by pounding.
Liquid fills concave molds perfectly.
Nobody has run SEM on a Sacsayhuamán block.
One sample would settle this.