Check a traverse that closes but has the wrong shape
Compare a separately measured diagonal with a perfectly closed rectangle to expose an error that endpoint closure cannot detect.
40 m × 30 m · check diagonal 50 mOn this page
Build two perfect closuresTreat the diagonal as a check measurementAsk whether the check is genuinely independentUse the plot to explain what failedA four-sided loop closes to zero both before and after two distances are increased by a metre. The plot still looks rectangular. A separate diagonal measurement immediately distinguishes the two shapes. It works because the diagonal supplies information the original four course equations never used.
Build two perfect closures
Make a test lot with four horizontal courses: 40 m north, 30 m east, 40 m south and 30 m west. Starting at N 0, E 0, point B is N 40, E 0. C is N 40, E 30 and D is N 0, E 30. The computed return lands exactly on A. The distance from A to C is the hypotenuse of a 30–40–50 triangle, so an independent horizontal diagonal should be near 50 m.
Now change both east–west lengths to 31 m. North and south still cancel. East and west still cancel. Point C has moved to N 40, E 31. Its inverse distance from A is √(40² + 31²), about 50.606 m. The area has grown from 1,200 to 1,240 m², yet the endpoint discrepancy remains zero. A control tick drawn only at A would miss the difference.
Treat the diagonal as a check measurement
Assume a separately observed horizontal A–C distance of 50.012 m on the same distance basis. For the original rectangle, observed minus computed is +0.012 m. For the widened rectangle, the same difference is about −0.594 m. The second geometry fits the independent measurement much more poorly. Whether either discrepancy meets the project requirement depends on observation conditions and applicable specifications.
Do not include that diagonal as an additional perimeter course when computing closure. It neither starts at the end of the perimeter sequence nor follows the boundary back to its start. If you want to use all five observations in a simultaneous adjustment, define a network model and its observation weights. A redundant length belongs in that model as a separate constraint.
Ask whether the check is genuinely independent
If the instrument software computed “measured diagonal” from the same stored corner coordinates, its agreement is arithmetic repetition. Look for a separate sight between the two identified endpoints or another independent source. Record occupation, target identity, instrument height treatment where relevant, and whether the diagonal was reduced to horizontal and to the correct grid or ground basis.
A distance check also has a directional blind spot. It describes the radius of a circle about A. C could move along that circle without changing A–C. Choosing a second diagonal or a tie to another known station gives an additional geometrical test. A single good tie does not resolve every possible distortion.
Use the plot to explain what failed
Paste the two sets of four azimuth–distance pairs into the loop calculator, one set at a time. The total length changes from 140 to 142 m. The four sides continue to make a clean loop. Because each graph scales to its own extent, visual similarity is a poor way to compare the two cases. The independent C coordinate, area and diagonal provide concrete comparisons.
On the review sheet, give each check a name. Record “perimeter closure at A” separately from “observed A–C minus coordinate inverse.” Name the observations used to derive each one. When a client or another crew asks how a closed loop could have the wrong area, those two lines make the distinction visible without relying on a generic claim about accuracy.
One practical drawing annotation is a dashed line between A and C labeled with both the separately observed length and the coordinate inverse. Keep its line style distinct from the four boundary sides. An unexplained fifth solid line can encourage the next person to treat the tie as a traveled course.