Traversing and the Closing Error

Each leg is measured for length and bearing, then resolved into a latitude (north-south) and a departure (east-west). The survey becomes two columns of signed numbers, and the geometry becomes arithmetic — which is why the method survived unchanged from the theodolite era into total-station practice.

Walking a closed loop returns you to the start, so both columns must sum to zero. That gives an arithmetic check on the whole survey, catching gross errors — a transposed digit, a misread bearing — without any extra fieldwork.

It never sums to exactly zero. The closing error √(ΣL² + ΣD²) is the physical distance between the computed end point and the known start, and it is the most honest single measure of the survey's quality.

A closing error means nothing without the perimeter it accumulated over — 10 mm over 200 m is excellent, and the same 10 mm over 20 m is a blunder. That is why specifications are written as ratios and never as distances.

Bowditch's rule distributes the closing error among the legs in proportion to their lengths, assuming error accumulates with distance — which holds for chained and taped work where the linear measurement is the weaker one. The transit rule assumes the angular measurement is weaker instead.

Modern practice uses least squares, weighting every observation by its estimated precision. But adjustment only makes a figure self-consistent — a systematic error, such as a mis-standardised tape, survives it untouched and closes just as neatly as good work.

A closed loop proves internal consistency and nothing about absolute position. A traverse can close to 1:20000 and sit twenty metres from where it believes it is. Only tying to externally known stations, or to GNSS control, checks position.

Which is the whole-to-part principle in a different vocabulary: the framework must come from outside the detail, not from within it.