diff --git a/v2/crates/wifi-densepose-signal/tests/cir_synthetic.rs b/v2/crates/wifi-densepose-signal/tests/cir_synthetic.rs
index 697731e6..b72af04c 100644
--- a/v2/crates/wifi-densepose-signal/tests/cir_synthetic.rs
+++ b/v2/crates/wifi-densepose-signal/tests/cir_synthetic.rs
@@ -154,6 +154,7 @@ fn save_fixture(path: &str, k_active: usize, csi: &[Complex64], expected_dominan
 }
 
 // ---------------------------------------------------------------------------
+
 // Shared test logic: inject 3-tap channel, run estimator, assert
 // ---------------------------------------------------------------------------
 
@@ -340,7 +341,7 @@ fn should_return_tof_at_40mhz() {
 // ---------------------------------------------------------------------------
 
 #[test]
-#[ignore = "ADR-134 P2 (remaining): RMS delay spread inflated by far ISTA taps above the 1% cutoff; needs delay-window/robust-spread work (dominant-ratio tuning now landed)."]
+#[ignore = "ADR-134 P2 (remaining): ISTA emits a spurious ~15-20%-of-dominant tap at an implausible far delay (~bin 150, ~3us) that inflates rms_delay_spread to ~390ns. Too close to tap2 (~30%) for a safe magnitude cutoff; needs ISTA recovery-quality work (grid de-aliasing / stronger far-tap suppression), not a band-aid. Dominant-ratio tuning landed separately."]
 fn should_produce_positive_rms_delay_spread() {
     let cfg = CirConfig::for_bandwidth_mhz(20);
     let k_active = cfg.delay_bins / 3;