When two theoretical levels connect — statistical mechanics to thermodynamics, quantum chemistry to molecular structure, molecular genetics to organismal phenotype — the connection requires a construct that neither theory provides. Statistical mechanics needs the Stosszahlansatz (molecular chaos hypothesis) to derive thermodynamic entropy from particle dynamics. Quantum chemistry needs bonding analysis to connect wavefunctions to molecular architecture. Molecular genetics needs a gene concept to bridge DNA sequence to organism.
Sticker identifies a pattern: these bridge constructs occupy a third theoretical role, distinct from both the lower-level dynamical theory and the higher-level observational theory. The bridge theory maps between them through a many-to-one relation — many dynamical microstates correspond to each observational macrostate, creating a contingent space of dynamical configurations compatible with observations but not selected by them.
Three conditions complete the bridge in sequence. A Partition defines which dynamical states count as observationally equivalent. A Magnitude characterizes the geometry of the contingent space — how large it is, how its elements relate. A Closure selects or weights elements within that space. The Stosszahlansatz is a Closure condition — it selects the molecular-chaos distribution from the space of all distributions consistent with macroscopic variables.
The framework generates a Mirror Test: any proposed closure should be identifiable as either closing the contingent space (legitimate) or introducing new structure (overreach). This distinguishes productive bridge constructs from smuggled assumptions.
The structural point: inter-level scientific disputes persist not because one side is wrong but because the bridge construct connecting the levels is incomplete or contested. The bridge is not a derived consequence of either theory — it is an independent theoretical contribution, as load-bearing as the theories it connects.
Note (2026-11-11): checked against the full text of the paper (Sticker, arXiv 2603.09626, 'The Architecture of Inter-Level Representation'). The three conditions (Partition, Magnitude, Closure) in that order, the contingent space, the many-to-one inter-level map, and the Stosszahlansatz as a closure rule are as in the paper. One account is not. The essay describes the Mirror Test as sorting a proposed closure into 'closing the contingent space (legitimate)' or 'introducing new structure (overreach)', and says this separates productive bridge constructs from 'smuggled assumptions'. In the paper the test asks whether the selected subset or distribution is invariant under the dynamical theory's symmetry group G. A closing rule is G-invariant and the emergence it generates is 'provisional'; an introducing rule is not G-invariant and the emergence is 'permanent', because no development of the dynamical theory that preserves G could supply the selection. The paper's own paradigm of an introducing rule, in the temporal case, is the Stosszahlansatz, which the essay uses as its example of a legitimate closure. The test is thus a classification of kinds of emergence (the paper's three types: reductive, provisional, permanent), not a test of which constructs are legitimate; the words legitimate, overreach and smuggled do not appear in the paper in this sense. The paper also gives its quantum chemistry example as four incompatible analyses of bonding for one quantum state (molecular orbital, valence bond, natural bond orbital and atoms-in-molecules), and says the gene concept is a problem of Partition stabilization, points the essay's summary compresses.