The station name can be right while the station centre is wrong.
A traverse closes badly even though the total station repeated its angles closely. Before distributing the discrepancy, ask where those angles were measured. A few millimetres of offset at an instrument or target can have a much bigger effect on a short sight than on a long one. Repeating from the same setup may only repeat the error.
This guide begins with a sketch of three survey marks, works through a 3-millimetre offset and follows a second setup that produces a different answer. You will learn what closure checks, how to distinguish repeatability from a correct setup, and when the evidence is too weak to assign a cause. The numbers are fictional, not field tolerances.
If you need the short route: locate the instrument and targets independently; calculate how much an offset could change each direction; then check whether the pattern survives an independent reoccupation. Keep the raw and adjusted results separate.
Geodesy subject guide · Learning that transfers into life · How X Works hub
Choose your reading route
Route 1 · Chapters 1–3: Build the geometry
Distinguish the mark from the place you measured
See what an angle refers to and why the same offset matters more over a short sight.
Route 2 · Chapters 4–6: Investigate a disputed traverse
Read angular closure and coordinate closure separately
Compare directional and positional closure, repeat the setup independently and test competing causes.
Route 3 · Chapters 7–9: Decide what the evidence supports
Put setup uncertainty beside the displayed precision
Use uncertainty, report a qualified result and transfer the check to changed geometry.
Full chapter index · Independent worked case · Questions and practical checks
Chapter index
Build the geometry
Distinguish the mark from the place you measured
Find the component that turns the direction
Work one station from first principles
CHAPTER 1 OF 9 · Build the geometry
1. Distinguish the mark from the place you measured
A traverse is a sequence of connected survey lines. At a station, an observer can use the direction to the preceding point and the direction to the following point to measure a turn. Distances and directions then allow the connected positions to be calculated in an agreed frame. A traverse can return to its starting point or finish on independent control, giving the team opportunities to check the observations against known geometry.
Place three dots on paper: A, B and C. Stand at B in your imagination. The intended angle is the angle between the line from B to A and the line from B to C. Now place a second dot slightly to the right of B and call it B′. At B′ you can still see the same A and C. Yet the angle between the two new lines need not match the one at B. The labels and targets have not changed. The point from which you observed them has.
In a total-station setup, centring is the work of aligning the instrument’s vertical axis with the intended station mark. A target can also be offset from its mark. The relevant measurement points are the instrument’s effective centre and the point represented by the sighted target under the observation model. An offset need not make an instrument unreliable. It makes the observation answer a different geometric question unless the offset is accounted for.
A survey record that says only “occupied B” does not show whether the instrument actually occupied B within the required uncertainty. The name identifies the intended mark. Setup checks and records connect the mark with the observation. That distinction is valuable well beyond surveying: putting the right label on a measurement is not the same as measuring the labelled thing.
CHAPTER 2 OF 9 · Build the geometry
2. Find the component that turns the direction
Suppose an instrument should be directly south of target A. Move the instrument east by 3 millimetres while the target stays fixed. Looking from the displaced instrument, the target now sits a little west of north. In a right-triangle sketch, the crosswise change is 3 millimetres and the forward distance is the sight length. The tangent of the small direction change is their ratio.
For small offsets, the angle in radians is approximately the crosswise offset divided by the sight length. The units in the fraction must match. With 0.003 metres over 12 metres, the ratio is 0.00025 radians, approximately 52 arcseconds. With the same offset over 60 metres, it is about 10 arcseconds. An arcsecond is one sixtieth of an arcminute, or one 3,600th of a degree. The conversion here uses about 206,265 arcseconds in one radian.
Direction matters as much as size. Move directly along the sight instead of across it and, in this ideal horizontal example, you change the distance but not the direction. A diagonal move has both effects. To estimate its effect on direction, project the offset perpendicular to the sight line. To estimate its first-order effect on distance, consider the component along the line. Do not divide the full displacement by every sight length regardless of orientation.
The same reasoning applies to an offset at the target, but the sign changes when you switch which end moved. A target east of its mark is seen farther east from an instrument to its south; an instrument east of its mark sees a fixed northern target farther west. Always sketch the actual ray before writing a correction sign. A correct magnitude and a reversed sign can make an apparently careful repair worse.
CHAPTER 3 OF 9 · Build the geometry
3. Work one station from first principles
In our fictional traverse, A lies 12 metres north of B and C lies 60 metres east of B. Use azimuths measured clockwise from north. From the correct centre at B, the direction to A is 0 degrees, to C is 90 degrees, and the clockwise turn from A to C is 90 degrees. Now put the instrument 3 millimetres east of B, leaving both target centres correctly positioned.
From the offset station the direction to A rotates west by about 52 arcseconds. The sight to C remains due east in the sketch; it is slightly shorter but has no direction change. The turn from the northern backsight through east is now about 90 degrees and 52 arcseconds. The instrument could return this almost identical result five times. It would be observing B′ consistently, not observing B correctly.
Change the offset: move the instrument 3 millimetres north instead. The direction to A remains north. From the displaced station, C now lies a little south of east, so the clockwise angle increases by about 10 arcseconds. A vague statement such as “3 millimetres of centring error explains the 52 seconds” is incomplete. You need the displacement direction and the sight lengths, not just the displacement size.
If both directions change, the angle changes by the difference of their signed changes. For example, if a backsight rotates 18 arcseconds clockwise and the foresight rotates 7 arcseconds clockwise, the measured clockwise angle decreases by 11 arcseconds. Adding two absolute changes would give the wrong result. A quick sketch or two arrows alongside the numbers is often enough to catch that mistake before it enters the field record.
CHAPTER 4 OF 9 · Investigate a disputed traverse
4. Read angular closure and coordinate closure separately
For a simple closed polygon with correctly identified interior angles, the sum of the angles is fixed by the number of sides. A four-sided polygon has a total of 360 degrees. If the measured interior angles sum to 360 degrees and 42 arcseconds, the angular misclosure in that convention is +42 arcseconds. A traverse connecting known directions can use a different closure calculation. Always state which convention and controls were used.
Coordinate closure is a different check. After turning angles into directions and combining them with distances, compare the computed end point with the known end point. The resulting difference has coordinate components. A distance error may harm coordinate closure without changing the angle sum. A combination of angle and distance errors can give deceptively small coordinate misclosure. Check each quantity for the question it actually answers.
Neither check automatically locates the bad observation. If the sum is 42 arcseconds too large, one angle might be 42 too large, two might be 21 too large, or several positive and negative errors could sum to 42. A single closure number is an aggregate. Applying an adjustment can distribute the discrepancy according to a chosen model, but it cannot establish which physical explanation is right.
Preserve the original angular and positional results before any adjustment. If a discrepancy exceeds the project’s applicable allowance or a setup record casts doubt on the observations, investigate rather than making the numbers close on paper. Even a small misclosure does not prove accuracy to external control. Two errors can cancel in the closure condition while leaving intermediate positions wrong.
CHAPTER 5 OF 9 · Investigate a disputed traverse
5. Make the second occupation genuinely independent
A second observation can mean several things. Repeating the pointing without touching the instrument tests repeatability in that setup. Using both faces, where the procedure calls for it, can help control some instrument effects. Neither step relocates a miscentred vertical axis. If the suspected problem is where the setup was placed, the comparison needs an independent way to position or check the setup itself.
In the fictional case, the first occupation at B gives 90 degrees 52 arcseconds. A separate occupation, with instrument and target centres independently checked, gives 90 degrees 3 arcseconds. Other station checks show smaller differences. The 49-second difference makes B worth investigating, especially if an independent inspection supports a crosswise offset of the right sign on the short sight. It does not make the second number correct by definition.
Imagine the new occupation used exactly the same tribrach and a biased plummet, with the tripod legs adjusted but the centring assumption unchanged. Agreement would offer less evidence than it seems to. Conversely, a replacement target, a changed station orientation and a different observing sequence all at once can create a difference you cannot attribute. Change one meaningful part at a time where feasible, and keep a record of what was held constant.
A useful comparison sheet lists the mark, the instrument and target identifiers, which positions were checked, sight lengths, direction convention and observation times. At the bottom, write the fault the repeat was meant to test. If the line says only “repeat angle,” another person may not know whether it tested pointing, levelling, centring, target identity or something else. Distinguishing those questions makes each additional observation more valuable.
CHAPTER 6 OF 9 · Investigate a disputed traverse
6. Compare patterns before blaming the shortest sight
The shortest sight is a plausible place to start, not a verdict. In the worked geometry, a short sight is sensitive because the offset lies largely across it. A target offset, a poor choice of mark centre or a systematic aiming difference can also show up strongly at short range. A short sight along the offset direction can be less affected than a longer sight that is nearly perpendicular to it.
Consider two setups observing the same point from different directions. If the inferred error is mainly a target displacement, the effects should vary with the view direction in a way that is consistent with one target offset. If the effects instead follow which instrument setup was used, examine station centring and orientation. If a common backsight orientation was imposed, several directions can be shifted together without the object targets having changed.
Also check for a different mark or a changed point definition. A station centred very well over the wrong physical mark can pass a check of instrument placement while still producing incorrect coordinates. A new prism on an old mount can shift the point observed. An independent photograph or inspection can clarify identity and construction, but it does not replace a geometrically useful observation when the decision is about millimetres.
Write down two plausible alternatives before choosing a fix. For each, predict which directions and distances should change and what you would see after reoccupying the station. A good test is one where the explanations make different predictions. If your only check repeats the same view from the same position, it may confirm that the discrepancy is repeatable without revealing why it occurs.
CHAPTER 7 OF 9 · Decide what the evidence supports
7. Put setup uncertainty beside the displayed precision
Suppose the relevant crosswise centring component on a 12-metre sight has a standard uncertainty of 1 millimetre. In the small-offset model it contributes about 17 arcseconds of standard uncertainty to that direction alone. A 1-millimetre component on a 60-metre sight contributes about 3.4 arcseconds. This is a sensitivity calculation using stated assumptions. It is not a recommended allowance or the complete uncertainty of an angle.
If two observed directions use the same setup, their errors can be related. An instrument offset can change both directions, in ways determined by the orientation and length of each sight. Taking the difference can partly cancel or amplify its effect. Do not combine two direction uncertainties as if they were independent without checking what they share. Use a model consistent with the observations, geometry and uncertainty sources.
The number of digits on a screen is a display choice. Closely repeated observations quantify some aspects of repeatability. Neither tells you how far the instrument axis was from the mark. If the work demands a turn estimate with substantially lower uncertainty than the setup can support, consider independently improved centring, different geometry or a procedure appropriate to the required survey. Reporting an extra decimal does not solve the problem.
Make the estimate practical. Say which offset or uncertainty would be large enough to explain the observed difference. Then ask whether your independent check is sensitive to that offset in the right direction. If it can only reveal a 10-millimetre shift, it cannot rule out the 3-millimetre explanation. An independent check is useful only when it can see what the question is about.
CHAPTER 8 OF 9 · Decide what the evidence supports
8. Write the report so someone can reconstruct the decision
A useful note starts with the intended marks and the observation convention. It records the unadjusted closure and the comparisons that triggered review. It then explains what changed between occupations, which differences were observed and what uncertainty is relevant. If a setup was rejected, describe why and preserve its data. A revised result is easier to trust when the first result has not disappeared.
For our fictional numbers, a defensible note might say: “The first turn at B was 90 degrees 52 arcseconds. Independent reoccupation returned 90 degrees 3 arcseconds. A crosswise offset on the 12-metre backsight can account for a change of this scale and sign; the observation was replaced following an independent centre check.” If there was no direct check of the offset, use “consistent with” rather than saying the cause was proven.
Give both angular and coordinate checks with their frames and relevant units. Make clear which result is raw and which is adjusted. If an external control check was not available, say so. A report can establish internal agreement under a stated model without promising every coordinate is accurate in a wider frame. That is a useful result, but it is a smaller claim.
Keep an unresolved outcome available. If two hypotheses still fit and the data do not discriminate between them, record that fact and name the next observation that could. An unresolved cause is not a licence to treat the coordinates as valid for every purpose. For field work, follow the project’s agreed criteria and the responsible surveyor’s decision.
CHAPTER 9 OF 9 · Decide what the evidence supports
9. Transfer: the short sight is now the foresight
Put the short sight 12 metres east of B and the long sight 60 metres north. The observer still measures the clockwise turn from north to east. Move the instrument north by 3 millimetres. The long backsight stays north. The short east sight turns slightly south, increasing the clockwise angle by about 52 arcseconds. You get the same size as the first worked example, but the affected sight and offset direction have both changed.
Now move the instrument east by 3 millimetres instead. The east sight stays east and shortens slightly. The long north sight turns west by about 10 arcseconds. The clockwise turn grows by about 10 arcseconds, not 52. The method transfers cleanly: sketch each ray, decide what is crosswise, and subtract the two signed changes. The earlier shortcut “short backsight caused it” does not transfer.
For a harder transfer, imagine a traverse used to position a feature on a map. An angle was repeated and the closing line fits, but an independent check at an intermediate corner disagrees. Would you tell the map user that closure proves the corner? No. Compare the geometry near that corner, the reference assumptions and the independent check. The discrepancy is useful evidence precisely because the closure test did not answer that question.
Try writing a three-line answer: what was measured, what geometry makes the suspected fault visible, and what result would disprove your explanation. If a colleague can reconstruct those lines without knowing the original example, you have a method you can use on a changed problem.
Independent worked case and evidence checks
In an independent fictional case, a target is 20 metres north of a station and a second target is 40 metres south. The instrument moves 2 millimetres east. Viewed from the shifted instrument, both targets lie west of their intended north or south rays. Their azimuth changes do not have the same sign: the north direction rotates anticlockwise, while the south direction rotates clockwise in the usual azimuth convention.
The magnitudes are approximately 21 and 10 arcseconds. The clockwise angle from the north target to the south target changes from 180 degrees to about 180 degrees and 31 arcseconds. This is a case where the two direction changes amplify in the measured angle, even though both physical rays move to the west side. Sketch north and south before deciding whether to add or subtract.
Now repeat the observation after translating the instrument and both target centres east by the same 2 millimetres. In the ideal sketch both relative rays are unchanged, so the angle returns to 180 degrees. A local angle cannot reveal a common translation of all three points. That does not make the observation faulty; it tells you which claim it cannot support. If the job needs a tie to external control, a separate observation is needed.
Check yourself by changing the south target’s distance from 40 to 80 metres, with only the instrument displaced. The south contribution drops to about 5 arcseconds; the north contribution stays about 21. The result should be near 26 arcseconds. If your calculation made both contributions smaller, go back to the drawing and ask which ray actually changed length.
Here is a different kind of check. A traverse’s first observed turn is too large by 30 arcseconds and a later turn is too small by 30 arcseconds, using one signed turning convention throughout. The closing direction can agree even though the directions carried along the intervening lines were wrong. If one of those lines is 100 metres long, a 30-arcsecond direction error produces a crosswise end displacement of about 100 × 30 ÷ 206,265 metres, or 15 millimetres, before any other errors are considered. Cancelling the direction error at the later turn does not turn that intermediate line back into the one you intended to observe.
To see why the distinction matters, imagine that the intermediate line was used to locate a feature before the traverse closed. The later closure result is not an independent check on that feature. A separately observed tie to the feature from a suitable control point tests something the closing angle did not. If the tie and traverse disagree, compare how their errors might be shared. A tie calculated from the same adjusted coordinates is not independent just because it is displayed in a different table.
Finally, decide which evidence would justify changing a coordinate already delivered to another reader. An independent centre check and a repeat angle may identify the suspect setup, but the old coordinate may have fed several later products. Record the affected span and identify what must be recalculated or rechecked. If the feature position was never independently observed, distinguish a recalculated coordinate from a newly verified one.
Worked comparison
| Observation | What it can test | What it does not establish |
|---|---|---|
| Repeat on one unchanged setup | Repeatability of those observations | Correct position of the instrument axis |
| Independent centring and reoccupation | Sensitivity to the setup and its physical centre | That every target or external mark is correct |
| Small angular misclosure | Consistency with the stated angle closure condition | Accuracy of each intermediate position |
| Independent check at a connected mark | Agreement with that check and its reference | The location of every hidden error |
Questions and practical checks
Why are short sights sensitive?
The same crosswise displacement subtends a larger angle at shorter range. The along-sight component has a different effect.
Does observing both faces fix centring?
No. Changing face does not necessarily move the instrument axis onto the correct mark. It can help with other instrument effects under the appropriate procedure.
Can closure be good when an observation is wrong?
Yes. Some errors can offset each other in the closure condition, or affect quantities the check does not observe.
Should I divide the misclosure evenly?
Not as a way to diagnose a suspect setup. Investigate mistakes and use the adjustment model and acceptance criteria appropriate to the survey.
Is 52 arcseconds a threshold?
No. It is the approximate result of the fictional 3-millimetre crosswise offset and 12-metre sight. Use project-specific requirements.
What if the target is on a prism pole?
Its observed centre still has to be related to the intended ground point. Check the target setup independently rather than assuming correct instrument centring also verifies the target.
Continue with the connected subjects
Start with traverse checks
Surveying, triangulation and closure: what the geometry makes checkable
Traverse closure and coordinate adjustment: interpret the end-point discrepancy
Follow setup, control and uncertainty
Reference prisms: distinguish a moved control point from motion of the object
Reciprocal trigonometric levelling: a different setup error and closure question
Free networks: what internal agreement can and cannot say about the datum
Error ellipses: read uncertainty by direction
Sources and scope
U.S. Army Corps of Engineers: Topographic Surveying, traverse closure and control
USACE New Orleans District: surveying guidelines and conventional traverse checks
The numerical cases in this guide are original fictional teaching examples, not measurements from a real site. This is an educational explanation, not a field safety instruction, engineering approval or an official certification. For a real project, use qualified practitioners, current product documentation and site-specific evidence. Sources were checked on 5 November 2026.