"What" is an Epit: Mathematical Foundations of Epistemic Structure

The same parts can form different wholes: a message chain differs from a broadcast even

when the same people take part. This paper develops a small unit for describing such

arrangements and rules for combining its records into larger descriptions. An epistemic unit,

or epit, has three ordered positions (A,B,C) and three pair channels, AB,BC,AC. Each

channel can be forward, inactive, or reverse, represented by +1, 0,−1. Each channel is one

trit, a three-valued entry, so the three channels give 27 configurations. Larger records are

assembled by checking their shared relations and allowing a completed record to become

part of another. For applications to knowledge, the proposed OETMI scheme connects

the subject (Ontology), observation (Epistemology), intended use (Teleology), assessment

(Meta), and accepted result (Integration). The foundational result is that the epit is the

minimal complete unit of relational distinction: three ports and three channels are necessary

and sufficient for the seven local arrow patterns when self-connections and duplicate channels

are excluded. If each named channel may be changed independently and the resulting

arrow pattern must remain known, all 27 configurations must stay distinguishable. Four

compatible triangular face records determine a tetrahedron’s six edge values. When a

finite observation leaves several possible wholes, the largest group of possibilities fixes the

minimum number of extra labels for exact recovery. A selected vocabulary of twenty names

and four declared return rules gives 556 initial objects, including 536 new outputs. Every

expression has one simplified form, and repeated construction follows an exact counting rule.

Deciding the assertions such records encode requires a proof theory and an interpretation.

In a logical realization, an epit carries a theory, a proof, and a claim; we show that proof

status cannot be separated from its theory when records from consistent effective arithmetic

theories are compared, that consistency certification has no fixed point under the standard

second-incompleteness hypotheses, and that no internal arithmetic predicate supplies truth

for all encoded arithmetic sentences.