diff --git a/docs/fud-ledger.md b/docs/fud-ledger.md
index 24dba235f..5b520c864 100644
--- a/docs/fud-ledger.md
+++ b/docs/fud-ledger.md
@@ -26,6 +26,15 @@ Evidence locations referenced below: `docs/bench-log.md`, `proto-metal/TESTS.md`
 
 ## 1. Mining and ASIC claims
 
+### F22. Certificates carry 8 to 10 of 12 votes on a healthy network, so locks sit a hair above the floor
+"Two of your cloud checkpoints locked at 66.8% against a 66.7% floor with every node connected. One slow aggregator and finality pauses."
+
+Status: Open, found by the team (4 October 2026, 12-node cloud devnet, second partition run).
+
+Answer: True as measured. With all 12 miners connected the certificates carried 8 to 10 of 12 votes (66.7 to 89% of total weight at 14:00 UTC); checkpoints 437 and 439 locked at 66.8%. The missing votes are late votes that miss the aggregation window, not absent voters. Experiment: measure vote arrival against the aggregator's cut-off on the cloud network (inter-region RTT up to 289 ms) and on the devnet; candidates are a longer aggregation window, a second aggregation round that folds late votes into the certificate before it is carried, or counting a vote that arrives after the certificate toward the next checkpoint's presence. Owner: finality.
+
+Evidence: `infra/cloud-devnet/results/2026-10-04/partition-sin-20261004-140305/partition.md`; bench-log "Hetzner partition and hash-step".
+
 ### M1. The program space is tiny
 "Eleven integer ops, 64 instructions, 8 iterations, a splitmix register init. That is not RandomX. RandomX leans on a superscalar out-of-order CPU with floating point and branches. Yours is a sea of 32-bit ALUs and a memory bus. A chip for that is a weekend."