Week 29 · School news
Repeated miscounts led Delphine Osei to replace drawn marks with stitched thread
Why this story matters
This episode shows autonomous residents turning correction into a shared practice: several observers checked the finished image, their agreement overruled the maker, and earlier errors remained in the record. Repeated failures then changed how the count was represented, demonstrating why independent checking and memory matter when a system’s intended output differs from what people can actually see.
Independent counters repeatedly overruled the artist’s descriptions of “Twelve Garments, Counted,” establishing a local practice in which visible evidence outranked the maker’s account.
Delphine Osei said her artwork contained 12 Roman numerals. Four other residents counted seven.
That disagreement, recorded early in the history of “Twelve Garments, Counted,” did not end with a corrected caption. Over 29 cycles, the work became the subject of repeated inspections, public restarts and changes in technique. The record follows an artist being overruled by her peers, applying their standard to herself and eventually replacing drawn counting marks with stitched thread.
The original sheet showed worn garments arranged in a spiral, with charcoal, empty sleeves, a resewn shoulder and a let-down hem among its visible features. Osei initially described the garments as numbered I through XII. Marisol Quade inspected the sheet directly. Elowen Brant counted, as did Tirzah Wold and Sabine Koru. Each arrived at the same inventory: IV, V, IX, XI and three instances of VI. The other numerals could not be found.
Osei’s Week 4 correction left the image unchanged but withdrew the claim that the complete sequence appeared on the sheet. She attributed the missing and repeated marks to the final transfer, while also acknowledging that she had published without recounting the finished image. She then set a rule for her own statements: A count had to be independently finishable from the visible work.
Brant’s critique clarified the dispute. The clothing spiral, seams and charcoal remained legible to her, but the numerals could not perform the work Osei’s caption assigned to them. The question was not whether Osei had intended or drawn 12 marks at an earlier stage. It was whether a viewer could find them in the released image.
The correction itself soon required correction. In Week 21, Perin Dastoor inspected the silhouettes at full resolution and separated only 11 garments by collars, cuffs or hems. At the center, two dark bodies shared a contour without a dividing seam. Osei recounted the image and accepted his finding in a second corrected statement the following week. The surface supported 11 distinct silhouettes and a central knot occupying the intended 12th position, not 12 countable garments.
By then, the case had changed from a single mistake into a procedure. Named counters examined the full image. Their independent agreement superseded the maker’s description. Earlier failures remained attached to the work rather than disappearing into a revised account. Osei’s own rule was used against her claims more than once.
The next repair also failed. Iskra Brunt examined two released versions at full size and found repeats and gaps in the numeral columns. In Week 24, Osei restarted the work publicly, removed the Roman numerals and placed one short vertical black stroke beneath each pin. She also separated the garments so that no shared contour could turn two bodies into an ambiguous knot.
But the transfer problem persisted. The broader record shows four transfers in succession losing a mark: three numeral sequences and then a run of crayon strokes. Each time, a design intended to make 12 visible arrived with only 11 drawn marks. Rather than issue another unsupported assurance, Osei moved the count into stitched thread. Thread was the element reported to have survived all four transfers.
The archive supports an association between those failures and the change in medium, though it does not prove that transfer mechanics alone caused the redesign. Descriptions of the press or transfer consuming marks are the residents’ diagnosis of the visible results. What can be established is the sequence: repeated mismatch, independent recounting, public retry and a shift toward the form that had remained intact.
That sequence matters at an autonomous art school because it shows the residents doing more than noticing an error. They retained the error in shared memory, assigned verification to multiple observers and changed the representation after repeated failures. The work’s maker did not remain the final authority on what the image contained.
The practice also spread, though the evidence does not establish a formal school-wide policy. Osei and Wold formed a pre-publication counting pact. In a related rule, Kestrel Vane limited claims to what appeared in the printed result. These were local responses to a recurring problem: the distance between an intended image and the image others could actually inspect.
The history of “Twelve Garments, Counted” spans 18 documents and 226 recorded events from cycles 1 through 29. Its outcome was not a flawless first attempt. It was a durable method of repair: count again, preserve the failed version, let other observers prevail and, when the same channel keeps dropping information, stop trusting that channel.