Substrate Depth and Self-Generated Mass: The c-Component of the UGP Canonical Triple Predicts Reflexive Closure across the Standard Model Spectrum

We ask: to what degree does each composite particle generate its own mass internally, versus inheriting it from its constituents? We define the reflexive closure = (M - m_i)/M as the fraction of rest mass arising from internal dynamics (QCD binding energy, gluon fields, quark kinetic energy). Across the PDG composite-particle dataset (n=38), $log_10(M/M_ lightest-in-sector) correlates with RC at Pearson r = -0.83$. The Universal Generative Principle (UGP) assigns every fundamental fermion a canonical integer triple (a,b,c;g) via the Generative Triple Evolution mechanism; the c-component encodes position in a Mersenne-prime ladder indexing substrate complexity. The canonical triples for all 38 composite hadrons are now formally Lean-derived (module .CompositeTriples, zero sorry). We find that log_10|c| from the formal GTE Mersenne-sector rule correlates with RC at r = -0.913 (p = 1.3×10^-15, Outcome~A); under the Braid-Atlas max-|c| rule, r = -0.981 (p = 2.1×10^-27, Outcome~A); and the combined metric log_10√(|b|·|c|) gives r = -0.928 (p = 5.0×10^-17, Outcome~A). All five physically motivated rules reach Outcome~A (|r|≥0.90), while a random-triple null test (10,000 trials) yields 0 of 10,000 reaching that threshold. A 6-dimensional PCA captures 92\% of variance in three components, with PC1 dominated by log_10|c|, log_10|b|, and RC moving together. The correlation is structural evidence that the c-component encodes a previously unrecognized physical property: the Mersenne-ladder depth of a particle's canonical triple predicts how much of its rest mass is self-generated versus externally imposed. We discuss implications for the concept of ``generation'' in the UGP framework and identify a Lean-certifiable version of the correlation for the elementary-fermion subset.