Squaring a Field with a Knotted Rope: Corner Error, Area Error, and What the Diagonal Check Can and Cannot Catch

A rectangle staked out with a 3-4-5 knotted rope is only as square as the knots, and the customary check — measuring both diagonals and requiring them to agree — is widely trusted to certify the result. This paper computes, from stated assumptions, how large the errors actually are for a nominal 40 m by 30 m field with a 1200 m² area and a 50 m diagonal, laid out with a triangle of 12 m, 16 m and 20 m. With a knot placement standard deviation of 0.04 m per segment and a differential rope stretch of 0.15 percent, the corner angle has a root-mean-square error of 0.404 degrees and a worst case, at the stated error bounds, of 1.8942 degrees. A corner error moves the area only at second order — a full degree of corner error changes the area by just −0.0152 percent — while side-length errors of 0.4 percent per side produce a typical area error of 0.5657 percent and a worst case of 1.6064 percent. The diagonal check is an excellent detector of corner error: a tolerance of 0.10 m on the difference between the diagonals bounds the corner error to 0.1194 degrees. It is, however, completely blind to one family of shapes: every isosceles trapezoid has exactly equal diagonals, so a trapezoid whose bases differ by 12 m in total passes the check exactly while its true area falls 2.02 percent short of nominal, and a 10 percent area shortfall is geometrically reachable within the same family. A second measurement — requiring the diagonals to bisect each other, checked to 0.15 m — closes that blind spot. Combining side lengths held to 0.3 percent, diagonals agreeing within 0.25 m, and a bisection check within 0.15 m guarantees a worst-case area error of 0.6055 percent, inside a 1 percent target, provided the knots are placed to within 18.08 mm and differential stretch is held to 0.1 percent by pre-tensioning. All results are computed, not measured. SIMULATED, NOT MEASURED.