Referential Stability

Definition and Scope

Referential Stability is the measured tendency of a catalogued subject to remain the same subject across successive consultations, recensions, and transfers of shelf. It is not a property of the subject itself but of the network of references in which the subject is held, and is therefore reported in the Tower as a ratio rather than a quality. A subject with a stability ratio of 1.000 is indistinguishable, in every consultation, from the last recorded description of it. A subject below 0.400 is generally removed to the Chamber of Latent Referents until its anchors are restored.

The measure is administered under the Principle of Referential Inertia, which holds that a subject at rest in its references tends to remain at rest, and is prerequisite to Stability Certification and hence to Referential Consultation. The two are frequently confused by junior staff. Stability is what a holding has; certification is what a holding is permitted to claim.

Method and the Inertial Ratio

Stability is calculated by the Inertial Ratio, first set out in the working papers of Celia during the survey of the eastern stacks and formalised in Cross-Referential Anchoring. The procedure requires that a subject be consulted three times at intervals of no fewer than forty days, each consultation recorded against the same anchor set, and the resulting descriptions compared for drift. Drift of fewer than four characters per thousand is treated as clerical error and corrected silently. Drift above that threshold initiates a Cross-Reference Review.

The forty-day interval is not arbitrary. Earlier practice used fourteen days, until the Office of Preliminary Assessment demonstrated that short intervals capture the recorder's habits rather than the subject's movement, a finding recorded in their report on Corridor 89-B and never seriously contested since.

The Kestrel Dispute

The interpretation of low ratios is the subject of a dispute now in its ninth decade. The Third Concordance, Annex 4, treats instability as a fault of the holding, to be corrected by re-anchoring: "A subject that will not hold still is a subject improperly tied" (Third Concordance, Annex 4, clause 11). The Seventh Concordance, Annex 2, takes the contrary view, that a low ratio indicates an over-tied subject labouring under too many references, and prescribes removal of anchors rather than their addition. The two annexes have never been reconciled, and both are cited in current practice depending on which clerk is on duty.

"Where two Concordances speak to the same fault, the later shall govern; where the later is silent, the earlier shall be read as though it were the later."

— Eleventh Concordance, Annex 1, clause 3

The Eleventh was intended to resolve precisely this kind of conflict but is silent on which of two annexes is "later" when both were amended in the same recension year. The Bureau of Referential Integrity has declined four times to rule on the question, noting in its fourth minute that "the Bureau does not adjudicate arithmetic" (Bureau minute 4/88).

Notable Instabilities

The most cited case remains the partial collapse in Corridor 89-B, where a run of forty-one holdings lost stability in a single winter and were found, on inspection, to be referring principally to one another. The Department of Autological Integrity classified the event as a failure of Referential Dependence rather than of any individual subject, and the corridor was closed for six weeks while anchors were redistributed by hand.

A lesser but instructive case concerns a single ledger on floor 4,112 which held at 0.997 for eleven years before dropping to 0.310 over four days in the winter of the year following. Investigation found no fault in the ledger. A clerk had been transferred to the adjacent desk and had begun describing the ledger, in his new register, as "the ledger to my left." The phrase propagated through nine cross-references before it was caught. The clerk was reassigned; the ledger recovered to 0.994 within two consultation cycles. No disciplinary finding was made, though the incident is still taught in the scriptoria as an example of what the Manual of Referential Protocol calls "the ordinary hazard of adjacency" (Manual, §12).

Reporting and Notation

Ratios are entered on the holding's facing card in the upper right corner, in iron-gall ink, to three decimal places. Pencil is not accepted, and a card corrected in pencil is returned unaltered to the desk that submitted it. Holdings awaiting recertification are marked with a small sigma in the margin; the mark is not explained on the card and clerks are expected to know it.

See Also

• Referential Dependence
• Cross-Referential Anchoring
• Narrative Stability
• Non-Referential
• Fading Theorem of Reciprocal Negation