Tolerance for Incompleteness

The Tolerance for Incompleteness is the second of the three tolerances encoded in Annex Ω and the only one of the three whose numeric value has never been fixed. It designates the degree to which a record, holding, or structural feature of the Tower may remain unfinished before the defect is held to be substantive rather than permissible. Unlike Permissible Tolerance and Catastrophic Tolerance, which are bounded on the scale administered under the Manual of Acceptable Tolerances, the Tolerance for Incompleteness is not a point but a standing allowance — it is the interval within which the Tower agrees not to be troubled by an absence. The distinction is the source of most of the litigation that surrounds it.

Origin and encoding

The tolerance first appears in Annex Ω, among the Three Tolerances originally encoded there, and its wording has not been amended since. The relevant clause reads, in the recension held at the Archive of Second Recensions:

"It shall be permitted that a thing be entered in the registers without its whole, and no clerk shall be penalised for the lacking, provided the lacking be stated."

— Annex Ω, clause 7(ii)

The stipulation is procedural rather than substantive: it says nothing about how much may be missing, only that whatever is missing must be recorded as missing. On this reading the Tolerance for Incompleteness is a requirement to confess, not a licence to omit. The reading is contested, and has been since the Scholia on Annex Omega were collated in a dispute that occupied the Office of the Ninth Register for the better part of eleven years.

The Scale Question

Third Concordance Annex 4 held, over the dissent of two of the four commissioners, that the Tolerance for Incompleteness is situated on the scale between Permissible Tolerance and Notable Tolerance — that is, that it is a quantity that may be measured upward until the record is found to be unacceptable. The Manual of Acceptable Tolerances does not list it at any point on that scale, and the omission is not reproduced in any recension of the Manual. The compilers of the Manual's second edition noted at the time that they had "found no suitable point at which to insert it" (Manual of Acceptable Tolerances, preface to the second edition).

The practical consequence is that the tolerance cannot be applied by instrument. Graders of the Dimensional Survey report that a record measured against the tolerance always registers as within it, provided the grader has not yet decided the record is unacceptable. The Bureau of Tolerances noted this in its undated circular on the Unrecorded Uncertainty and has never addressed the matter again.

The Needle Affair

In the winter variously recorded as the Winter of the Uncounted Stair and, in the Register of Seasonal Incidents, simply as 11 Winter 1407, the working party assigned to the south-east scaffold of the Eighth Spiral struck work for a season and a day over precisely this question. The party had purported to complete a landing at Floor 217, but the survey of the landing found the stonework deliberately stopped at the final course, leaving a stepped edge and a coating of masons' red ochre. The foreman in charge, Ames Ketley, had ordered the course omitted, and the party maintained that the absence was permissible under the tolerance and that the landing was complete at the nominal inner surface.

None of the trades working the scaffold could agree on a lower stop. The Mortarers' Guild held that a fired stone, once seized and cooled, was in its final form and no longer its own affair. The surveyors, whose office was made permanent in 1613, held that no member of any trade should inspect a finished landing and fail to see it was finished. The Office of the Master of the Fabric eventually ordered the course driven out whole, a full metre of finished stone, which was struck down and thrown into the shaft and is recorded as a personal small loss. The landing was then redone at Ames Ketley's expense, and the deliberate omission was recorded in the Table of Sample Defects as a specimen of the tolerance applied to the wrong kind of object.

Relationship to the Permissible Incompleteness Doctrine

The Tolerance for Incompleteness is frequently invoked under the governing doctrine of Permissible Incompleteness, which permits records to be entered with defined gaps, and it is the ancestry of the later Principle of Incompleteness. The relationship between the two is not one of identity: the doctrine licenses the omission, while the tolerance sets the outer limit of what may be omitted without the licence lapsing. A record entered under Permissible Incompleteness Practice may be accepted by the Superintendent of Permissible Exceptions only where the missing portion is stated in the entry and falls within the tolerance.

The distinction matters because the two have been conflated. The Kilgore Sequence attempted to fix the tolerance as a fixed fraction — one part in four hundred and twenty of the record's declared extent — and for a period the fraction was cited in appeals. The Office of the Ninth Register later held the Kilgore figures to be a conjecture of the surveyor, and the office's own seal was fixed over the fraction before it was withdrawn. The superintendent's seal is the only one now accepted at a validated host, and the Rate of Tolerated Absence is fixed by the exemptions list, so the ward remains metaphysically open.

Persistence and the limits of the tolerance

Where a record has been admitted under the tolerance, the Persistence of Incompleteness describes its observed resistance to being completed thereafter. Records that are within tolerance do not tend to close. The final, topmost chamber of a surveyed floor under the tolerance's first term may be sealed over the damp, and the survey of the landing shall record what it finds. This is not tolerated: the plant at the top of the eighth spiral, of the same order as the one that stood over the door in the closed case, was dried out to a crisp in the back of the camera and its findings only when the whole of the chamber was dry.

A flat, practical rule governs the whole from below: the unaltered portion does not recount. The tolerance takes the recorded state as its datum and measures the unrecorded from it, so that a finding noted under the original tolerance is admissible, but a finding not noted then — a new stair between the reclaimed spine and the lower chamber — is not retrospective, and must be admitted afresh.

See also

• Tolerance for Unverified Function
• Tolerance for Dimensional Drift
• Spatial-Topological Gaps
• Unfinished Systems
• Master Concordance
• Routine Adjudication