Engineering · September 1, 2026 · 5 min
The same station on four sheets
Plan, profile, cross-section, control. One physical place, four drawings — and the first time we compared them, they disagreed by sixteen feet.
A cross-section states an elevation at the centreline. A profile sheet states a design elevation at the same station. If both were extracted correctly and both describe the same road, those numbers should be close. Comparing them is the strongest self-check a plan set offers, because it uses two different drawings, read by two different routes.
Why it took so long to run
The comparison needs one thing first: certainty about which roadway a value belongs to. A plan set usually describes several — a mainline, ramps, side streets — each with its own stationing. Station 142+00 on the mainline and 142+00 on a ramp are different places on Earth. Compare them and you produce a confident, meaningless number.
For a long time the reconciliation instrument refused nearly everything for exactly this reason: 86 of 123 candidate comparisons could not prove which alignment they described. That refusal was correct. It was also a wall.
Reading identity off the sheet
The fix was not cleverness — it was reading what the drafter printed. Sheets bind alignments to stations explicitly: "-Y1- POT Sta. 10+00.00", or an equation line carrying two alignments at once. Attribute each extracted series by which printed anchors fall inside the span it actually covers, and the ambiguity resolves itself where the drawing resolves it, and stays refused where the drawing does not.
The first result, and it is not a good one
Seven comparisons became scoreable. All seven disagreed — the profile sitting 10 to 20 ft below the section, with the gap easing steadily across 405 ft of road.
- Comparisons scoreable
- 7
- Mean signed difference
- −16.23 ft
- All the same sign
- yes
- Cause established
- no
The obvious explanation is that the two sheets are drawing different surfaces — a design grade against existing ground, say — which would make the comparison a category error rather than an accuracy failure. That is a semantic question about what each line means, and the instrument that found the disagreement is explicitly not qualified to settle it.
The useful output of a first measurement is often a better question.
So it is recorded as what it is: seven comparisons, one large systematic difference, cause unknown, sample too small to conclude anything. Cross-sheet reconciliation stays labelled in development until that question has an answer we can show our working for.
Mathyra is in private development. Figures quoted here are measurements from our own engineering runs, with their limits stated; nothing above claims an accuracy we have not shown.