False Precision in Ancient Units

Extra digits cannot improve a survey

Converting a surveyed length into metres can produce more digits than the evidence supports. Those digits help reproduce the calculation. They do not establish the length of an ancient building to the same number of decimal places.

Flinders Petrie reconstructed the Great Pyramid's original casing base from surviving casing points and the corner sockets. He used the socket diagonals to locate the inferred corners; this was a reconstruction of a partly lost boundary. His four side estimates give a mean of 9,068.8 inches.¹ The table separates those estimates from our subtraction of their mean.

The average of the four lengths is 9,068.8 inches. The average of the absolute differences is (0.6 + 1.1 + 0.7 + 0.2) ÷ 4 = 0.65 inch. The first describes the mean size; the second describes how much the reconstructed sides differ from it. Neither calculation alone supplies the uncertainty of the mean.

Petrie's error figures describe different things

On page 40, Petrie explains that the apparent irregularity includes survey error. Identifying original construction points and socket diagonals adds uncertainty beyond locating the survey lines. He allows probable errors of about ±0.6 inch for each reconstructed side, then concludes that the base's underlying irregularity was somewhat less than 0.6 inch. That concluding statement is not an uncertainty interval around the mean length.¹

For the mean base used in his height calculation, he separately writes 9,068.8 ± 0.5 inches on page 43.¹ We retain his probable-error convention without turning it into a modern standard uncertainty or a 95% confidence interval. The four sides are different parts of a reconstructed building, not four independent repeated measurements of one unchanged side.

The terminology matters. Modern metrology uses measurement precision for agreement among repeated measurements under stated conditions. Measurement uncertainty describes the dispersion of values attributed to the quantity being measured, using the available information.³ Here, display precision means the number of printed decimal places. Changing it supplies no new measurements.

An exact factor preserves the input's limits

Using the modern international inch gives 9,068.8 × 0.0254 = 230.34752 metres. The factor is exact by definition: one modern inch is 25.4 millimetres. NIST dates this agreed definition to 1959.² We apply that modern convention to Petrie's published number; this calculation does not recalibrate his nineteenth-century survey standards.

On the same convention, his ±0.5 inch becomes ±0.0127 m, or roughly ±0.013 m. Multiplying by an exact factor scales both the length and its quoted error; it does not shrink their relative proportion. The long product therefore belongs to the arithmetic, while the historical estimate keeps its source and error convention.

At two decimal places, rounding moves the product by 0.00248 m, or 2.48 mm. At five places, the displayed value equals the product exactly. At eight, three trailing zeros appear: 230.34752000 m. The last printed place then represents 0.00000001 m, yet the evidence remains Petrie's same reconstruction. Zero rounding change is entirely compatible with substantial measurement uncertainty.

A close fit still needs independent evidence

Suppose a proposed design length differs from the product by 0.005 m. That illustrative 5 mm difference is smaller than the 12.7 mm equivalent of Petrie's quoted probable error on the mean. It supplies no probability that the proposed design is right. The model's own uncertainty, the reconstruction method, the proposed ancient unit and the reason for choosing its count all require scrutiny. Choosing the unit after seeing the result can manufacture a close fit.

The same discipline applies to an ancient-unit converter. A declared representative cubit can support a useful comparison for its stated place and period; a long decimal expansion cannot make it a universal ancient standard. The Giza royal-cubit dossier separates the site's working value from its New Kingdom comparison.

Round for the claim you are making

For a short historical description, use about 230.35 m and identify Petrie's reconstructed casing mean. When its uncertainty matters, retain the original 9,068.8 ± 0.5 inches and explain the source convention. For reproducible arithmetic, keep the full product and round only the final presentation. The useful test is whether each digit serves the calculation or is being mistaken for new knowledge about the monument.

Last verified: 3 November 2026