While the accuracy is impressive (the article here says altitude difference of a few millimeters when the tunnels met, but comments have pointed out that’s an e

While the accuracy is impressive (the article here says altitude difference of a few millimeters when the tunnels met, but comments have pointed out that’s an error and it is actually a few decimeters), the really cool part to me is the techniques Eupalinos employed to make sure the tunnels connected even if there was significant error. The diagrams in the article show how they changed the lateral direction of both tunnels to ensure they would collide in the horizontal plane, and expanded the ceiling of one tunnel and the floor of the other to ensure they would also intersect in the vertical plane. The diagrams in the article are illuminating, with some thought you can see it’s a very robust error correction mechanism that requires only a fraction more digging effort, and only requires precise measurements of the length of each tunnel (much easier to precisely measure than the angle or level!). Fascinating stuff.