EA-CORPORA-09, 10 and 11 seat three corpora in a single session, each of them addressing a specific failure in the control structure of the authorship investigation.
Theophrastus had been recorded as unavailable three times, on the grounds that he survives only as the Characters, which does not divide and so gives the naming-gap measure nothing to fire on. That was an assumption rather than a search. First1KGreek was checked out on the same machine throughout and contains Historia Plantarum, De causis plantarum, De sensibus, De igne and the Metaphysica. Seated at fifteen works and 208,324 tokens, he carries the reflex — nine instances in Historia Plantarum alone at a division rate of 29.9 per ten thousand. The result is nevertheless void as a control result, because Theophrastus is defined by his relation to Aristotle, writes continuations of the abandoned Aristotelian programme, and shares a single transmission channel with the corpus he would be testing. He is a comparandum and the seat says so.
Thomas Aquinas is the first figure admissible under a criterion the archive had already written down but never been able to satisfy: a control must be impossible to place inside the configuration under test, which requires attestation from outside it. Thomas is secured by canonization proceedings, university registers, papal bulls and Dominican chapter records, none of which bears on textual style, doctrinal similarity or any succession tradition. No Greek candidate met that standard. The seat runs to 6,698,317 tokens from Corpus Thomisticum with no OCR, and is split three ways rather than two after a locus probe returned zero for "metaphysic" in a file labelled In Metaphysicam. The prefix table had mislabelled every commentary — that file was Sentencia super Meteora, and what had been labelled De sensu was In Psalmos. Rebuilt from each text's own header, the split separates independent works from Aristotle commentaries from commentaries on other objects, and the third category proves to be the control the design lacked: the naming-gap reflex runs 3.47 times higher where Thomas glosses Aristotle and five times lower where he glosses Scripture and Dionysius. The reflex arrives with the source text rather than with the man.
The Pessoan seat existed already and was too thin to answer anything. Four voices, 111,449 tokens, with Ricardo Reis at 5,548 — a quantity that cannot fill a six-part apparatus whatever Pessoa did, so any null would have measured the seat rather than the corpus. Arquivo Pessoa lists thirty-nine authorial positions and 4,529 texts. Twenty-two hundred were fetched and twenty-five positions seated at 1,254,436 tokens, an elevenfold expansion. Bernardo Soares, absent entirely, is now 139,455 tokens; Reis is 42,955; Alexander Search and António Mora are new; the orthonym stands at 2,810 of 2,814. Seven signature loci were verified before seating, each by the vocabulary that identifies its voice.
Two defects are declared rather than repaired silently. The Aquinas mislabelling is recorded in the seat's locus field, including which files were wrong and how the error surfaced. The Theophrastus seat was built earlier the same day with its verified-loci field empty, and was verified only retroactively; the note says so.