Observer-Induced Mutation

Observer-Induced Mutation is the class of documented change in which a previously non-existent feature of the Tower is brought into being by the act of measuring, numbering, or otherwise formally regarding the space in which it later appears. It is one of the three axiomatic pillars set out in Clause 11 of the Annex of Perpetual Revision, alongside the Infinite Regress of Detail and the Principle of Uncaptured Context, and it is the only pillar for which the Tower maintains a standing inventory of confirmed instances. The phenomenon is not considered a fault. Per the Annex of Perpetual Revision, preamble to Clause 11, "a record shall not be preferred to the thing recorded merely because the thing arrived late."

Defining the class

Mutation is distinguished from ordinary cataloguing error by three conditions, all of which must be satisfied before an instance is entered on the inventory. First, the feature must be absent from every prior survey of the space. Second, the act of regard must precede the feature's first appearance by a measurable interval. Third, the feature must persist for no fewer than sixty days after documentation, a threshold fixed by the Provisional Concordance of 1887 to exclude the mere optical complaints of tired Observers.

The term itself is generally credited to Observer-Librarian Kael, who used it in the marginalia of a survey return from Spiral 12 in 1894 and did not elaborate. Kael's later work on Paradox of the Silent Observer complicated his own coinage considerably, and he is on record as having preferred "induced presence" in correspondence after 1899. The earlier term prevailed in the register and Kael's preference did not.

The numbering of Floor 704

The canonical instance remains the Chronological Precedence survey of Floor 704. In the winter of 1871 the surveying party under Clerk-Adjutant Miren Osstable was directed to affix sequential numbers to the floor's shelf-ends for the purpose of settling a storage dispute three floors above. The party completed the numbering in nine days.

On the tenth day the party reported that the numbered shelves now fronted onto aisles, and that the aisles had not been there. Subsequent measurement found eleven aisles where the original survey return had recorded none. The shelves, notably, did not move; the aisles appear to have occupied the intervals the numbering had presupposed between them. Osstable's return of 14 Frostfall 1871 is worth quoting in full, as it is the earliest plain statement of the phenomenon in the archive:

"The numbers were applied to what was present. The aisles were applied to by the numbers. I have checked the figures twice and decline to check them a third time."

— Survey Return 704/9, Chronological Precedence Practice

Floor 704 has been surveyed on four further occasions, each by a different office, and the aisles have been confirmed without exception. The Office of Sequential Corrections continues to dispute whether the aisles should now be numbered, and has been disputing it since 1874.

Interaction with measurement protocols

Mutation of this kind places obvious strain on ordinary practice, and the response of the Tower has been procedural rather than theoretical. Standard Measurement Protocol (SMP-Δ) was adopted in part to absorb observer-induced drift into the measurement return itself, by recording the observer's position and motion alongside the reading and publishing both. The protocol does not prevent mutation. Its advocates maintain only that a mutation properly observed is a mutation properly recorded, a position that has never been successfully answered though it has frequently been resented.

The Hesitant Gauge is the most common instrument in these surveys and the least trusted. A gauge that has been read tends to read differently on the second occasion; the standard instruction is to read it once, at a distance, and to accept the reading. Observers assigned to first-pass surveys are rotated every four days. Nobody has established why this helps.

The Osstable–Verrin dispute

Scholarly disagreement on the subject is old and has not narrowed. The Second Recension holds that mutation is a property of the Tower, which responds to the attention of its cataloguers as a house responds to being surveyed. The Fifth Recension holds that mutation is a property of the record, and that the aisles of Floor 704 were always there but could not be said to exist until a numbering made their existence sayable.

The disagreement is not academic in its consequences. Under the Second Recension, a survey team is responsible for what it brings into being, and damages may be assessed against its office. Under the Fifth, no one is responsible, because nothing happened except an improvement in description. Archivist Perrin Verrin argued the Fifth Recension position at length in 1903 and was afterwards refused access to Floor 704 on the grounds that his presence constituted a further act of regard. He appealed. The appeal was heard by an office that declined to name itself and was dismissed without reasons, which Verrin took as an admission.

Later commentary has sought a middle path. The Treatise on Coincident Fabrication proposes that mutation is genuine but trivial, and that the Tower is simply larger than its surveys in a way that surveys discover rather than cause. This has pleased nobody and is widely cited for that reason.

Recorded instances and precautions

The standing inventory of confirmed instances, maintained by the Observer rank since 1888 and inscribed each year with a Sanctioned Quill of Errosion, currently records the following as uncontested:

• The aisles of Floor 704 (1871)
• The east reading room of Spiral 44, first noted in the Parallax Gardens returns (1892)
• The stairwell between Floors 1902 and 1903, which a numbering party approached from above and which has been approached from above ever since (1898)
• The uncatalogued antechamber adjoining the Chamber of Absolute Silence, entered on the record of 14 Frostfall 1901 and withdrawn from it the same evening

Practical precautions are few and are stated plainly in the surveying manuals. Parties are to number in pencil where possible. A wall that has been measured is not to be leaned upon until the following season, and the manuals do not elaborate. Where a surveyor finds an unexpected feature that appears to answer to the surveyor's own recent paperwork, the finding is to be recorded as it stands and the surveyor is to be reassigned to a different spiral without further discussion of the matter in the return.

The inventory's preface, revised in 1889, notes that the class is expected to grow, and that growth of the class is not itself an instance of the class. The preface adds, in the same flat register as everything around it, that this distinction has been found sufficient.