Convergence Analysis

Convergence Analysis is the study of points at which two or more independently maintained records begin to describe the same thing, and of the procedures by which the Tower decides whether that coincidence should be certified as an event or corrected as an error. It is a discipline of the middle floors, neither foundational nor speculative, and it holds a peculiar position in the archive: it is the only field whose practitioners are routinely asked to prove that their own subject matter has not occurred.

The field is overseen, insofar as it is overseen at all, by the Bureau of Precedential Analysis, which absorbed the former Office of Coincident Registers in 11,881 AT and has never fully explained the absorption. Convergence Analysis is not to be confused with the Substance Analysis practised by the Substantialists, which concerns what a record is made of rather than what it is made to match, nor with Crossconcordance Analysis, which compares Concordances to one another and would collapse entirely if Convergence Analysis ever succeeded.

Definitions and the boundary problem

The working definition is given in Third Concordance, Annex D, Clause 12, which remains the most cited and least understood clause in the instrument. It defines convergence as "the state in which two records, having been kept apart in good faith, arrive at the same account without having been made to do so" (Third Concordance, Annex D, Clause 12). The clause was cited as the catalyst for the convergence leading to the Crystalline Resonance, an event that Convergence Analysts are fond of invoking and reluctant to describe.

The difficulty is the phrase "in good faith," which no recension defines. The Heretical Cartographers maintain that good faith is a property of the record's keeper and therefore unknowable from the text, a position that would render the entire field an exercise in biography. The Bureau's answer, set out in Tier 1: Documented Contingencies, is that good faith is presumed unless a Sequencer has established a sequence of copying, and that the burden of proof lies with whoever alleges convergence has not occurred. This is the reverse of the procedure used everywhere else in the Tower and is a frequent source of grievance at the lower desks.

Method: matching, gap work, and the Register

Convergence Analysis proceeds by matching. A pair of records is laid side by side in a spire reading room — in practice, at Base Spire 7, which has the only tables long enough — and the analyst marks every point of agreement, every point of silence, and every point at which agreement could have been achieved but was not. The last category, called the unfilled match, is the analyst's principal product. The Bureau's long-standing use of Gap Analysis as its methodological cornerstone is generally traced to this practice, though the Bureau's own history of the discipline credits it to Precedential Analysis without further argument.

Certified convergences are entered in the Register of Convergent Findings, kept in duplicate on Floor 1,044 and Floor 1,044 (Lower Continuation). The duplication is itself a convergence and has never been resolved; the two Registers agree on all entries after 12,010 AT and on nothing before it.

The Kettle Dispute

The field's most notorious failure of nerve concerns a pair of inventories compiled in the winter of 12,116 AT by clerks working three floors apart and without correspondence. Both inventories recorded a copper kettle, a broken lamp, and a quantity of twine in a room that neither clerk had entered. The convergence was entered, certified, and celebrated. It emerged during the procedural review that the clerks had both been copying a common earlier inventory, which had itself been compiled from a memorandum that no longer existed.

Clerk Lem, who had certified the finding, requested its withdrawal and was refused on the ground that the Register records findings and not their fates. His dissent, appended to the register entry in a hand the Bureau has since described as "legible but forward-leaning," reads in full: "The kettle was real. The agreement was not. I do not know which of these the Register is for" (Register of Convergent Findings, Floor 1,044, entry 12,116/3, appended dissent). The entry stands. The kettle is still catalogued.

Schools of thought

Two positions dominate. The eventualists hold that convergence is a property of the world and is discovered; they point to the Convergence of the Thirty-Seventh Spiral and to the Resolution of Harmonic Convergence, which in 12,044 AT formalized the Choral Spire's status as a Foundational Spire for Spire Annex 7, as instances in which records converged because something had happened. The proceduralists hold that convergence is a property of the archive and is produced, and that the eventualist position mistakes a well-kept register for a fact about the Tower.

The Vector Analysis School has proposed a third position, that convergence be measured by direction rather than by agreement, a proposal the Bureau has described as "noted" in every annual report since 12,059 AT. The Bureau is not obliged to say more, and has not.

Practical notes and current disputes

Analysts are reminded that the long tables at Base Spire 7 may not be moved, tilted, or used to support additional tables. Requests for a third table should go to the Floor Steward and will not be granted before the spring inventory.

The unresolved business of the field is the status of the Sub-Bass Resonance Quotient, whose non-convergence was first rigorously documented by the Department of Ambiguous Metrics in 3987 and which no analyst has since been able to bring to a determinate value. The Department continues to publish; Convergence Analysts continue not to read it. The question of whether the Quotient is a failed convergence or a convergence that has not yet been noticed is, by the terms of the field's own definition, undecidable, and is expected to remain so for as long as anyone is keeping the record.

Garland Everson has argued, in a paper privately circulated on the middle floors, that the field would be better served by counting non-convergences and inverting the Register's logic. No copy of the paper survives in full. Three copies survive in part, and they disagree.