Worked Example: Small Side Errors Near a Flat Triangle

Calculate a nearly flat triangle

Take sides 5, 5 and 9.8. The semiperimeter is 9.9. Area is √[9.9 × 4.9 × 4.9 × 0.1] = 4.875438… square units. The final factor 0.1 reflects the small remaining distance from degeneracy.

Increase the long side by one tenth

For sides 5, 5 and 9.9, the semiperimeter is 9.95 and area is √[9.95 × 4.95 × 4.95 × 0.05] = 3.491417… square units. The long side changes by about 1.02%, while area falls by about 28.39%. This is geometric sensitivity, not a claim that the algorithm became less accurate.

Check the result with an independent height

For equal short sides of five and base b, height is √[25 − (b/2)²]. Multiply by b/2 to obtain area. At b = 9.8, height is √0.99 and area is 4.9√0.99. At b = 9.9, height is √0.4975 and area is 4.95√0.4975. These reproduce the Heron results.

Distinguish precision and uncertainty

More digits describe the result of the entered measurements. They do not establish that the measurements were accurate to those digits. Near a flat triangle, the uncertainty in area can be much larger than the relative uncertainty in one side.

Measure and model a plausible range

For a physical plot, remeasure the corners and test reasonable side bounds while preserving valid geometry. Report a range when the inputs justify one. Material rounding and waste allowance are additional decisions, and should not be mistaken for a correction to the geometric uncertainty.

Check two well-conditioned triangles before studying a nearly flat one

For an independent illustrative check, use sides three, four and five. The semiperimeter is six, so Heron's expression gives √[6 × 3 × 2 × 1] = 6 square units. A right-triangle calculation gives the same result: one half of three times four is six. The perimeter is twelve. Having a second formula with a known perpendicular height makes this a useful arithmetic check, not merely another run of Heron's formula on identical assumptions.

Now double every side to six, eight and ten. The semiperimeter becomes twelve and the area becomes √[12 × 6 × 4 × 2] = 24 square units. The perimeter doubles to twenty-four, but the area quadruples. All lengths changed by a factor of two; area carries two length dimensions. Reporting a doubled area would confuse a linear scale with a squared scale. This third triangle also satisfies the right-triangle area check, one half of six times eight.

The units must accompany the calculation. If the first three sides are feet, its area is six square feet, not six square meters. Using the exact foot-to-meter factor 0.3048, six square feet equals 6 × 0.3048² = 0.55741824 square meters. Converting all three side lengths to meters before calculating produces the same area. A single multiplication by 0.3048 would produce the wrong dimension.

Hold two sides fixed and test a measured base interval

Return to equal sides of five, but suppose a recorded base of 9.80 has an illustrative uncertainty of ±0.05 unit. The candidate bases are 9.75 and 9.85. Both remain below ten and therefore describe proper triangles. At 9.75, the independent height is √[25 − (9.75/2)²], giving area about 5.416243 square units. At 9.85, the same check gives about 4.249151. The nominal base 9.80 gives about 4.875438. This range is much wider than a six-decimal display might suggest.

The lower base bound produces the larger area in this particular near-flat interval. That relationship is not universal for every possible base: with fixed equal sides, area first rises and later falls as the base changes. Avoid assuming that the smallest side value always produces the smallest area. Test the actual valid range or inspect the formula before labeling lower and upper area bounds.

This calculation varies only the base and holds the two five-unit sides exact. It is a conditional sensitivity exercise, not a complete uncertainty analysis of a measured plot. If all three sides have uncertainty, their changes interact, and some combinations may violate triangle inequality. Instrument precision, repeated measurements and correlations determine whether a candidate combination is physically plausible. The calculator neither measures those quantities nor assigns probability to the tested endpoints.

Audit validity, reported precision and project allowances separately

The original change from base 9.8 to 9.9 is approximately 1.020408% relative to 9.8. The corresponding area decrease is approximately 28.39%, using the original area as the comparison denominator. These percentages refer to different quantities, so their mismatch is not a conservation failure. A thin triangle loses height rapidly as its long side approaches the sum of the other two sides.

At sides five, five and ten, triangle inequality reaches equality. The three vertices lie on a line and the geometric area is zero, but the tool intentionally rejects that input as a proper triangle. For a base larger than ten, no Euclidean triangle exists with the two given short sides. Do not replace a rejected measurement with a tiny positive area or take the absolute value of a negative factor in Heron's expression. Recheck the measurements and the units instead.

Intermediate arithmetic should retain precision, while the final report should match the quality of the source measurements. Six decimal places can be useful when comparing algorithms; they do not justify purchasing material to six decimal places. If a plan rounds the nominal 4.875438 square units to 4.88, record that reporting choice. If it then adds a ten-percent waste allowance, the nominal estimate becomes approximately 5.363 units squared before practical purchase rounding. The allowance is a planning assumption, not a repair to uncertain geometry.

The side-only model assumes a planar triangle with straight edges. A curved boundary, sloped surface or irregular garden may require a different geometric model or subdivision. Keep the measured sides, their units, the tested range and the reason for any allowance in the project record. These invented cases verify geometry and sensitivity; they do not certify a surveyed boundary, a construction quantity or the accuracy of a field instrument.