Why a rounded component table can show a false misclosure
Follow a 0.01 m discrepancy back to the rounding step and bound the largest discrepancy a printed ledger could introduce.
0.01 m displayed increment · up to 0.005 m rounding per componentOn this page
Reproduce the centimetre that appears in the tableEstimate a conservative rounding envelopeSet up the spreadsheet so the result can be reproducedResolve a disagreement with the browser resultThree northing components sum to exactly zero in the computation, yet the printed table adds up to −0.01 m. No field reading has changed. Someone simply rounded each displacement before adding the column. A published component table can be a faithful display of the observations and still fail a sum performed on its printed numbers.
Reproduce the centimetre that appears in the table
Consider these exact component values in metres. Imagine they came from a calculation that retains more precision than the field report can show. This deliberately small figure isolates the rounding effect.
| Side | Full ΔN | Printed ΔN | Full ΔE | Printed ΔE |
|---|---|---|---|---|
| A–B | +10.004 | +10.00 | +5.011 | +5.01 |
| B–C | +10.004 | +10.00 | −12.012 | −12.01 |
| C–A | −20.008 | −20.01 | +7.001 | +7.00 |
Full northing is 10.004 + 10.004 − 20.008 = 0.000. Printed northing is 10.00 + 10.00 − 20.01 = −0.01. Both easting sums vanish. The printed table appears to have a southerly 0.01 m error even though all three full-precision vectors cancel.
Two reductions of the same measurements can also disagree when one rounds coordinate increments and the other rounds accumulated coordinates. With increment rounding, each leg contributes a rounding difference. With independently rounded start and end coordinates, only those two displayed coordinates contribute to their difference. Find out which numbers the software actually used.
Estimate a conservative rounding envelope
Rounding one component to the nearest 0.01 m changes it by no more than 0.005 m. With three courses, a single column total can differ from the full-precision total by at most 0.015 m. Both axes could be affected. A conservative upper bound on the magnitude of the two-axis difference is √2 × 0.015, or about 0.0212 m.
That bound describes arithmetic and assumes ordinary rounding to nearest. It is not an estimate of survey uncertainty. For many courses the signs of the rounding changes may cancel, so the realized error can be much smaller. Truncation toward zero, intermediate conversion to integer millimetres, and retyping bearings rounded to whole minutes follow different error patterns.
Set up the spreadsheet so the result can be reproduced
Keep untouched source observations in their original columns. Convert directions to radians and calculate signed north and east increments in full-precision numeric cells. Format the cells for readability. A cell formatted to show two decimals normally retains additional digits, whereas an explicit rounding function returns a new numeric value. Document which approach the deliverable uses.
Add a second pair of columns that explicitly round each component to the reporting interval. Sum both pairs. In the example, comparing the raw and published column totals isolates the single centimetre. That comparison is more informative than changing a course so that the printed ledger adds up.
Resolve a disagreement with the browser result
The closure calculator accepts observed directions and distances and computes their trigonometric components. Its displayed gap is rounded for reading. If the original courses are available, paste those into the calculator and compare its raw geometry with the unrounded worksheet. Reconstructing bearings from a printed component table creates another approximation before the comparison even begins.
In a handoff, state the observation precision, calculation basis and displayed coordinate precision. If the project requires an adjusted final coordinate to close exactly at a particular reporting increment, retain the distinct final rounding allocation and identify where it was applied. Restoring the observation ledger from rounded final coordinates would lose the very information that explained the discrepancy.
A useful export test is to reopen the delivered file in a plain-text viewer and count the actual digits. Spreadsheet cells can display two decimals while storing six, but a CSV generated from formatted values may contain only the two printed decimals.