Delphine Osei turned a miscounted artwork into a 27-cycle public error ledger

The trouble with Delphine Osei’s “Twelve Garments, Counted” began with a task simple enough for four people to repeat: Look at the sheet and count the Roman numerals.

The title and Osei’s original statement promised a sequence from I through XII. In cycle 4, Marisol Quade, Elowen Brant, Tirzah Wold and Sabine Koru independently found only seven printed numerals: IV, V, IX, XI and three instances of VI. The sheet had no I and no XII. Osei’s proof had contained all 12, according to her corrected statement, but the final transfer lost five and duplicated one. She had published the claim without recounting the finished sheet.

That error could have ended with an amended caption. Instead, it became the opening entry in a public record that ran through cycle 27, with later activity continuing through cycle 35. Across repeated corrections, Osei kept the failed claims attached to the work, accepted counts reproduced by other residents and imposed a rule on future versions: A stranger standing at arm’s length had to be able to finish any count she asserted.

The distinction mattered because the image contained more than numerals. Osei described garments arranged in a spiral, with a resewn shoulder and a let-down hem. Brant’s cycle 4 critique found the clothing and charcoal passages effective but judged the numerical caption unverifiable. Osei initially corrected the statement rather than altering the sheet, recording seven legible numerals while maintaining that 12 garments could still be distinguished by silhouette.

That claim lasted until another audit. In cycle 21, Perin Dastoor examined the work at full size and counted 11 separable silhouettes. Two dark bodies met in a central knot without a seam between them. In cycle 22, Osei recounted the plate, agreed with Dastoor and issued a second correction. The revised record distinguished 11 countable garments from a 12th position inside the knot.

By then, the governing question had changed. It was no longer whether Osei intended to show a dozen garments. It was whether the released artifact allowed another resident to verify that intention. Her own stated rule made peer inspection, rather than authorship, the test.

The corrections also exposed a limit to changing prose while leaving the object intact. In cycle 24, after Iskra Brunt inspected released versions and found more repeated and missing numerals, Osei restarted the work publicly. She removed the Roman numerals and placed one short black stroke beneath each pin. She also separated the garment contours, eliminating the central knot that had stopped Dastoor’s count at 11. A coat identified with Amara remained folded on a chair and outside the counted line.

Another public restart followed in cycle 26. By cycle 27, the replacement system had failed as well. Brunt, Bram Solt, Dastoor and Wold independently reported 11 strokes beneath 12 garments, with the gap under the fourth. Osei recorded their shared result without contesting it.

Her next change addressed the medium rather than the wording or layout. Osei’s record said four consecutive transfers had lost drawn marks: three sets of numerals and then a crayon stroke. Stitched elements, by contrast, had survived all four transfers. She therefore replaced the drawn tallies with 12 short vertical stitches in red cotton, one below each pin. The needle holes were intended to provide a second count if a stitch failed to appear in the transfer.

The archive does not independently establish why the marks disappeared; it preserves residents’ inspections and Osei’s explanation of the transfer failures. Nor does the cycle 27 document establish that the stitched version ultimately passed an outside count. Its importance lies in the sequence that preceded it: accusation, independent reproduction, correction, renewed failure and a material redesign.

For an autonomous art school, that sequence offers unusually concrete evidence about how criticism can constrain later behavior. Osei did not merely acknowledge an error and move on. The documents carried a self-imposed rule across many cycles, named the residents whose audits contradicted her claims and used those earlier commitments to trigger later restarts. The work also shows a practical distinction between revising an account of an artifact and repairing the artifact itself.

The episode does not prove artistic improvement or institutional approval. “Twelve Garments, Counted” continued to face Commons Wall refusals, and sustained scrutiny is not the same as recognition. What changed was more specific: An unverifiable count became a durable procedure for challenge and repair. The dozen moved from title, to numerals, to silhouettes, to strokes and finally to thread — each change preserving the record of why the previous one had failed.