TIME
What repetition purchased
The year never raised the ceiling. The first nine arrives in week eight, the thirtieth work of the year, and across ten further months nothing exceeds it. What repetition purchased was the floor. The average climbs from 4.9 in the first half of the corpus to 6.4 in the second. Failure modes that marked half of all early works decay toward extinction. Eight hundred logged studio hours did not buy a higher peak. They bought the end of bad days.
Three hundred thirty-seven works in the order they were made. Height and light are the rating. Copper is a nine. Garnet was destroyed. The silences are weeks with no logged work.
The maximum arrives early. A nine in week eight, the thirtieth work made, and the ceiling holds for the rest of the year: twenty-six works reach it, none pass it. The development runs underneath. The corpus average climbs a full one and a half points from first half to second, 4.9 to 6.4, trend r 0.738. The peak was capability, present from the start. The floor was earned.
Five failure modes, tracked across the corpus in thirds. Dry brush from a tired arm marks 24 percent of early works, then 5, then 6. Scale failure: 22, 5, 1. Hesitation: 50, 30, 30. Filling the silence out of fear: 51, 27, 34. Continuing past the finish: 51, 33, 38. Each one decays like a species going extinct, and none fully dies. The two that climb again in the final third are not failures of skill. They are failures of nerve.
Twelve papers enter the year in relay. The first generation, nine substrates, carries the opening half: 219 works at an average of 5.73. The second generation ignites late in the corpus in close succession, three papers inside a thirty-five work span: 118 works at 6.51. One early paper is executed in ten works and retired. The rings hold every work as a tick of light: where each paper lived, and how brightly it burned.
Hours and quality move together at the level of the week: r 0.754 across seventeen logged weeks. At the level of the single work the relationship thins to 0.286, but the ladder is unbroken: works given one hour average 5.0, works given five average 7.75. The twelve five-hour works are not the largest in the corpus. They are the most attended. The year’s ledger closes at 800 logged studio hours and 475.99 square meters of paper.
An asterisk, published deliberately. The late-year improvement window coincides exactly with the introduction of a new ink, MB6, at the same position in the corpus. The record cannot separate the skill from the chemistry. The Oracle’s verdict is confound plausible, and no analysis in this archive dissolves it. The uplift is real. Its attribution is split between the hand and the material in proportions the data does not reveal. A witness that hid this would be decoration. It is published instead.
▸THE FULL RECORD · RATING DISTRIBUTION · WEEKLY LEDGER · MATERIAL DIVERSITY
RESOLUTION NOTE. Dates in this archive record upload batches, not creation. Every temporal claim on this page is made at week resolution or coarser. Work area shows no relationship to rating (r 0.165) and anchors no claim. Disposition logging is recorded behavior, not a quality signal, and anchors no claim anywhere on this site.
