Blending and conversion determine what can still be traced

5.5 Bale laydown transforms discrete deliveries into a pooled yarn lot

In the proposed model, bale laydown is a pooling boundary. Where several bale lots feed one mixed yarn lot, a single-bale parent pointer no longer describes the observable output. The record must list every bale or bale group that could enter the shared run, including substitutions and undocumented inputs. Opening, carding, drawing, combing and spinning can then be represented as one campaign or as separate activities, depending on the boundaries at which operators actually record consumption and production.[S02][S10]

EPCIS represents the conversion as a TransformationEvent. Identified objects belong in inputEPCList and outputEPCList; measured class quantities belong in inputQuantityList and outputQuantityList. Recoverable material and recorded losses need their own outputs or linked quantity records on a compatible measurement basis.[S10][S03] The UN Transparency Protocol (UNTP) Digital Traceability Events page offers a work-in-progress credential mapping for such events, suitable here only as a pilot comparison.[S30] Once the model boundary is a mixed lot, a downstream scan can retrieve the recorded input set but cannot create a fibre-level observation that was never captured.

5.6 An input set bounds provenance without preserving single-farm identity

A mixed laydown therefore produces a bounded input set, not a single origin pointer. Whether that set supports a claim depends on the attribute and its governing custody model. UNTP's pilot guidance states the key distinction directly: "A chain of custody model applies to a claimed attribute, not to a facility or a process." Country of growth, scheme eligibility and producer-group identity can consequently have different outcomes on the same run.[S02][S0503]

Identity Preservation keeps a defined source separate. Segregation permits eligible material from several sources to combine while excluding non-eligible material. Thus, an illustrative laydown of bales from three documented Cameroonian ginneries could retain a country and scheme-class claim if every input and control satisfies the applicable rules, but not a single-ginnery claim.[S02][S07] Version labels matter: the Aid by Trade Foundation's June 2025 standard uses Hard Identity Preserved (HIP), while the current Cotton made in Africa webpage uses Segregation and says the underlying separation requirements are unchanged.[S06][S07]

Controlled Blending tracks a known fixed proportion. Mass Balance reconciles eligible input and claimed output volumes administratively rather than preserving their physical separation.[S02][S0503] An illustrative laydown mixing documented Cameroonian and Benin cotton cannot support an unqualified single-country physical claim. Its record should retain the multi-origin input set and any supportable percentage statement, while a scheme claim follows that scheme's edition, model and allocation rules rather than an automatic factory-wide downgrade.[S02][S06][S0503]

GFT keeps three operations separate when evaluating that input set. The union of possible origins answers where material in the batch may have come from. The intersection of shared attributes answers what can be said without qualification about every input. A mass-weighted allocation answers a different question again and is valid only where the selected custody model and compatible quantities permit it.[S02][S0503]

For example, two documented producer groups in one country have a union containing both groups but an intersection that can still contain the shared country attribute. If one group's country evidence is missing, the country intersection becomes unknown rather than defaulting to the documented group. Scheme eligibility follows its own rule set and is not inferred from that geographic intersection. This is why the graph stores an attribute vector rather than collapsing provenance into one score or one country field.

Exhibit 20 illustrates a hypothetical 1,000 kg spinning laydown combining 600 kg of lint from documented Cameroon Source A with 400 kg of lint from documented Cameroon Source B, converting the blended fibre into 820 kg of combed yarn, 130 kg of recoverable comber noil and 50 kg of process loss.[S02][S10]

In Exhibit 20, the illustrative 820 kg yarn lot and 130 kg recoverable noil link to the whole 1,000 kg laydown, not exclusively to Source A or Source B. A genealogy query returns {Source A: 60 percent, Source B: 40 percent} by recorded input mass.[S10] It cannot establish the fibre ratio inside a single 200-gramme shirt, because no fibre-level measurement is in the model. The illustrative 82 percent yarn yield, 13 percent noil share and 5 percent loss demonstrate conservation only. Actual campaign yields require compatible input, output, by-product and loss measurements, which Chapter 6 reconciles.[S03][S06]