A closed traverse is a chain of measured angles and distances that begins and ends on the same point. Because the starting point is known, the geometry imposes conditions the measurements must satisfy: the interior angles of an n-sided figure sum to (n − 2) · 180°, and the north–south and east–west components of the courses must each sum to zero. Field measurements never satisfy those conditions exactly. The amounts by which they fail are the closures, and interpreting them is most of what traverse computation consists of.
The work has a fixed order, and the order matters more than any individual step. Angles are checked and adjusted first, because an angular error propagates into every azimuth downstream and therefore into every course. Only then are the courses reduced to latitudes and departures, the linear misclosure computed, the precision assessed, and — if the precision passes — the misclosure distributed. Adjusting distances before checking angles smears a bad angle across the whole figure and destroys the evidence you would have used to find it.
What follows works one five-sided traverse from field angles to adjusted coordinates and area, with every table produced by the computation library that drives the calculators on this site. The last two sections deal with the case that matters most in practice: what to do when the traverse does not close.
The geometric conditions
Two independent conditions apply to a closed loop traverse, and they are checked separately because they fail for different reasons.
interior angles of a closed n-sided figure: Σ = (n − 2) · 180°
exterior angles: Σ = (n + 2) · 180°
closed loop: ΣLat = 0 and ΣDep = 0The angular condition tests only the angles, so it isolates the instrument, the pointing and the booking. The linear condition tests the angles and the distances together, which is why it can only be interpreted after the angular condition has passed. A traverse with a bad angle will fail the linear condition too, but nothing in the linear result tells you which of the two kinds of measurement was at fault.
Angular closure
Five interior angles were observed. The condition is (5 − 2) · 180° = 540°00′00″.
| Station | Observed angle | Correction | Adjusted angle |
|---|---|---|---|
| A | 101°22′15″ | −6″ | 101°22′09″ |
| B | 148°15′05″ | −6″ | 148°14′59″ |
| C | 82°40′20″ | −6″ | 82°40′14″ |
| D | 118°05′40″ | −6″ | 118°05′34″ |
| E | 89°37′10″ | −6″ | 89°37′04″ |
| Sum | 540°00′30″ | −30″ | 540°00′00″ |
The observed sum exceeds the required sum by 30″, so the angular misclosure is +30″ and the correction is −30″ distributed over five angles: −6″ each. The correction is applied equally because each angle was observed with the same instrument by the same crew with the same care. There is no basis for weighting one more heavily than another, and inventing one is how a blunder gets hidden.
Whether 30″ is acceptable depends on the instrument and the specification. A common rule of thumb takes the allowable angular misclosure as the instrument's least count multiplied by the square root of the number of angles. For a twenty-second instrument and five angles that is 20″ · √5, about 45″, so this traverse passes. Check the governing specification rather than the rule of thumb when one exists, but the square-root form is worth internalising: doubling the number of stations increases the allowable misclosure by only about forty per cent.
Running azimuths around the loop
With adjusted angles in hand, the azimuth of every course follows from the azimuth of one. Starting from a known direction — from an existing control line, a solar or GNSS observation, or an assumed meridian — each successive azimuth is the previous one plus the angle turned plus 180°, reduced into the range 0° to 360°.
αᵢ₊₁ = αᵢ + (angle at the station) + 180° (mod 360°)Taking A–B as 26°15′00″ from control, and applying the adjusted angles in turn:
| Course | Azimuth | Bearing | Distance (ft) |
|---|---|---|---|
| A–B | 26°15′00″ | N 26°15′00″ E | 285.10 |
| B–C | 354°29′59″ | N 5°30′01″ W | 610.45 |
| C–D | 257°10′13″ | S 77°10′13″ W | 720.48 |
| D–E | 195°15′47″ | S 15°15′47″ W | 516.55 |
| E–A | 104°52′51″ | S 75°07′09″ E | 797.90 |
Latitudes and departures
Each course is now resolved into its north–south component, the latitude, and its east–west component, the departure. Cosine goes with the latitude because azimuths are measured from north.
latitude = D · cos α (+ north, − south)
departure = D · sin α (+ east, − west)| Course | Azimuth | Distance (ft) | Latitude (ft) | Departure (ft) |
|---|---|---|---|---|
| A–B | 26°15′00″ | 285.10 | +255.698 | +126.097 |
| B–C | 354°29′59″ | 610.45 | +607.639 | −58.512 |
| C–D | 257°10′13″ | 720.48 | −159.986 | −702.493 |
| D–E | 195°15′47″ | 516.55 | −498.330 | −135.982 |
| E–A | 104°52′51″ | 797.90 | −204.908 | +771.140 |
| Totals | 2,930.48 | +0.1138 | +0.2495 |
Both totals should be zero and neither is. ΣLat is +0.1138 ft, meaning the traverse computed its way back to a point about a tenth of a foot north of A. ΣDep is +0.2495 ft, a quarter of a foot east. The traverse did not return to its starting point on paper, and the size and direction of that failure are the whole diagnostic output of the computation.
Linear misclosure and the precision ratio
The two closures are the components of a single displacement. Their resultant is the linear misclosure, and expressing it as a fraction of the distance travelled gives the relative precision — the figure a specification is written against.
e = √(ΣLat² + ΣDep²) = √(0.1138² + 0.2495²) = 0.274277 ft
precision = 1 : (ΣD / e) = 1 : (2,930.48 / 0.274277) = 1:10,684The direction of the misclosure is as informative as its size. Here the computed closing point lies 0.2743 ft from A on an azimuth of 65°28′56″, so the corrections must push the traverse back along 245°28′56″. That direction is the first thing to look at when a traverse fails, because a misclosure produced by a single bad distance points along the course that contains it.
- Relative precision
- The linear misclosure divided into the total length of the traverse, written 1:X with X truncated rather than rounded. A traverse of 2,930.48 ft that misses by 0.2743 ft computes 1:10,684.4 and is reported as 1:10,684, because rounding a precision upward claims accuracy that was not achieved.
The compass (Bowditch) rule
Once the precision passes, the misclosure is distributed among the courses so the traverse closes exactly and yields a single consistent set of coordinates. The compass rule, also called the Bowditch rule, distributes each closure in proportion to course length.
correction to latᵢ = −ΣLat · (Dᵢ / ΣD)
correction to depᵢ = −ΣDep · (Dᵢ / ΣD)The assumption behind it is that angles and distances are of comparable precision and that error accumulates with the length of the line. That is the usual case for a traverse run with a total station, and it is why the compass rule is the default in practice and in most software.
| Course | Distance (ft) | Lat correction | Dep correction | Adjusted latitude | Adjusted departure |
|---|---|---|---|---|---|
| A–B | 285.10 | −0.011 | −0.024 | +255.687 | +126.072 |
| B–C | 610.45 | −0.024 | −0.052 | +607.616 | −58.564 |
| C–D | 720.48 | −0.028 | −0.061 | −160.014 | −702.554 |
| D–E | 516.55 | −0.020 | −0.044 | −498.350 | −136.026 |
| E–A | 797.90 | −0.031 | −0.068 | −204.939 | +771.072 |
| Totals | 2,930.48 | −0.1138 | −0.2495 | 0.000 | 0.000 |
The corrections sum to the negative of the closures, which is the arithmetic check on the adjustment, and the adjusted columns sum to zero, which is the point of the exercise. Notice the size of the corrections: the largest is 0.068 ft, on the longest course. No individual correction approaches the size of a measurement error a crew would notice, which is exactly the situation in which distributing the misclosure is legitimate.
The transit rule
The transit rule distributes each closure in proportion to the magnitude of the component being corrected rather than to the length of the course.
correction to latᵢ = −ΣLat · (|latᵢ| / Σ|lat|)
correction to depᵢ = −ΣDep · (|depᵢ| / Σ|dep|)Its assumption is that the angles are markedly better than the distances — the situation when directions come from a precise instrument or from astronomic or satellite observation while distances are relatively weak. Because it corrects each component in proportion to that component, it disturbs the azimuths of the courses less than the compass rule does, which is the property that makes it appropriate in that case.
| Course | Compass adj. latitude | Compass adj. departure | Transit adj. latitude | Transit adj. departure |
|---|---|---|---|---|
| A–B | +255.687 | +126.072 | +255.682 | +126.079 |
| B–C | +607.616 | −58.564 | +607.599 | −58.520 |
| C–D | −160.014 | −702.554 | −159.996 | −702.590 |
| D–E | −498.350 | −136.026 | −498.363 | −136.001 |
| E–A | −204.939 | +771.072 | −204.922 | +771.033 |
The two rules move the coordinates by at most about five hundredths of a foot on this traverse, and the resulting areas differ by 8.9 sq ft on 12.62 acres — around two ten-thousandths of an acre. That is the honest summary of the choice: on a traverse that closes well, the rule barely matters, and on a traverse that closes badly, no rule saves it. The reason to understand both is to know what each is asserting about the measurements, not because the numbers diverge.
Coordinates and area
Accumulating the adjusted latitudes and departures from a starting coordinate gives the coordinates of every station. Taking A as 5,000.00 north, 5,000.00 east on an assumed system, with the compass rule adjustment:
| Station | Northing (ft) | Easting (ft) |
|---|---|---|
| A | 5,000.000 | 5,000.000 |
| B | 5,255.687 | 5,126.072 |
| C | 5,863.303 | 5,067.508 |
| D | 5,703.289 | 4,364.954 |
| E | 5,204.939 | 4,228.928 |
| back to A | 5,000.000 | 5,000.000 |
The traverse now closes exactly on A, as it must after adjustment. Area follows from the coordinates by the coordinate method: 549,716.59 sq ft, which is 12.6198 acres. Computing area from unadjusted coordinates is a mistake, because the figure does not close and the coordinate method assumes it does.
When a traverse fails
A failed closure is a finding, not a formality to be adjusted away. The response depends on which condition failed and by how much, and the sequence below moves from the cheapest checks to the most expensive.
- Re-check the arithmetic before returning to the field. Re-enter the field notes independently, verify the angular sum against (n − 2) · 180°, and confirm that the azimuths close back on the starting direction. A large fraction of failures are transcription or entry errors and cost nothing but time to find.
- Confirm the angular closure separately. If the angles fail their tolerance, the problem is angular and no amount of distance checking will resolve it. If the angles pass comfortably but the linear closure fails, suspect a distance.
- Look at the direction of the misclosure. This is the single most informative number available and it is free.
- Check the obvious systematic causes: an unrecorded instrument or reflector height, a prism constant left at the wrong value, a slope distance never reduced to horizontal, a scale factor applied twice or not at all, or a mixture of grid and ground distances in the same loop.
- Re-observe the suspect course or angle. One targeted re-observation, chosen using the evidence above, is usually enough and is far cheaper than re-running the loop.
- If nothing is found and the misclosure is small but out of specification, re-run the traverse. Adjusting a traverse that failed its specification and saying nothing is not an option.
There is a hard rule underneath all of this. Adjustment redistributes random error. It has no legitimate application to a blunder, because a blunder is not random and spreading it corrupts every station instead of one. If the misclosure is larger than the accumulated random error could plausibly explain, something is wrong that adjustment cannot fix.
Locating a blunder
Two geometric properties of a single blunder make it findable, and both come from the misclosure vector rather than from re-measuring everything.
A distance blunder points along its own course
If one distance is wrong and everything else is good, the traverse is displaced by exactly that error along exactly that course's direction. The misclosure vector therefore lies parallel to the offending course.
Suppose the crew had booked C–D as 722.48 ft instead of 720.48 ft — a two-foot error, the kind produced by a mis-read tape or a mistyped entry. The traverse then closes as follows: ΣLat = −0.3303 ft, ΣDep = −1.7005 ft, giving a linear misclosure of 1.732 ft and a precision of 1:1,692. The misclosure vector runs on an azimuth of 259°00′30″. Course C–D bears 257°10′13″. The two agree within about two degrees, and no other course in the traverse comes within sixty degrees of that direction.
| Course | Azimuth | Difference from misclosure direction 259°00′30″ |
|---|---|---|
| A–B | 26°15′00″ | 127°14′ |
| B–C | 354°29′59″ | 95°29′ |
| C–D | 257°10′13″ | 1°50′ |
| D–E | 195°15′47″ | 63°44′ |
| E–A | 104°52′51″ | 154°07′ |
The identification is unambiguous. The residual scatter of the good measurements keeps the agreement from being exact, which is why the test looks for the nearest course rather than a perfect match, and why it works well for one large blunder and not at all for two.
An angular blunder sits on the perpendicular bisector
A single bad angle rotates every subsequent course about the station where it was turned, so the misclosure is a chord of a circle centered on that station. The perpendicular bisector of the misclosure vector therefore passes through the offending station.
Introducing a 30′ error into the angle at C, leaving every distance correct, gives ΣLat = +0.734 ft and ΣDep = −7.281 ft: a misclosure of 7.318 ft at 1:400, running on 275°45′32″. Erect the perpendicular bisector of that vector — a line through the midpoint of the misclosure on azimuth 5°45′32″ — and measure how far each station lies from it. Station C is 15.7 ft off the line. Station B is 103 ft off, D is 701 ft off, and E is 790 ft off. On a traverse nearly three thousand feet around, that is a clear identification.
What closure does not tell you
A closed traverse checks itself against itself. That is genuinely useful and it is strictly limited, and confusing the two is the most consequential misunderstanding in this part of the subject.
- Closure does not detect systematic error. A tape that is uniformly short, an uncorrected temperature effect, or a scale factor omitted from every course scales the whole figure without disturbing the closure at all. The traverse closes at 1:20,000 and every distance is wrong by the same proportion.
- Closure does not establish position. A traverse can close perfectly on an assumed coordinate that is nowhere near the true one. Position comes from tying to control, and a loop that never touches published control is a shape, not a location.
- Closure does not validate orientation. If the starting azimuth is wrong, every course rotates together and the closure is untouched. The figure is correct and it faces the wrong way.
- A very good closure is not automatically good news. A closure far better than the instrumentation and procedure could produce suggests compensating blunders, a repeated observation that was never independent, or data that has already been adjusted somewhere upstream.
The practical conclusion is to treat the closure computation as one check among several rather than as the verdict. Tie to published control at both ends where you can, so the traverse is checked against something external; observe redundantly so blunders have somewhere to show up; and record the misclosure, the precision, the adjustment rule and the control used, so that anyone reading the survey later can see not only that it closed but what closing was worth.