# The metaplectic phase lives on a congruence extension **Date:** 25 October 2026 **Status:** direct finite lift obstructed; explicit relation rail constructed; genuine scalar \(C_4\) lift ruled out **Evidence:** [congruence-lift ledger](../../../evidence/data/moonshine-metaplectic-congruence-lift-gate-v1.json) **Certificate:** [finite-lift obstruction verifier](../../../verification/verify/verify_moonshine_metaplectic_congruence_lift_gate.mjs) ## Result The quarter-turn phase (S\mapsto i) is algebraically available on (\mathrm{Dic}_{11}^{ab}\cong C_4), and eighth-root theta multipliers exist on the genus-two metaplectic cover. Those two facts do not yet define a common finite subgroup upstairs. The level-23 instruction group sits in \[ \overline D\cong\mathrm{Dic}_{11}4). Thus the level-23 transition skeleton has no faithful finite-order integral lift. Its integral preimage is an infinite congruence extension. ## 1. What this changes about the (C_4) test Composing (\widehat D\to\mathrm{Dic}_{11}\to C_4) still gives an abstract character on the congruence preimage. The obstruction is subtler: the metaplectic cover pulls back to \[ 1\longrightarrow\mu_2\longrightarrow\widetilde{\widehat D} \longrightarrow\widehat D\longrightarrow1, \] The initially desired scalar assignment would send the metaplectic center to (-1) and the quarter-turn lift to (i). The subsequent explicit lift proves that these requirements are inconsistent: a quarter-turn lift has fourth power equal to the metaplectic center, whereas \(i^4=1\). The finite abelianization (C_4) fixes only the quotient values. It says nothing about the phase accumulated by relations that close only modulo the principal congruence subgroup. That kernel phase is exactly the missing datum. ## 2. The corrected next calculation Do not search for a finite embedding (\mathrm{Dic}_{11}