# Review recipe — return #1846 ## Portable layout and commands Place these files under a relative `artifacts/` directory and run from its parent. Python 3 standard library suffices. No randomness, dependency installation, network or sieve reconstruction is needed for the review checks. 1. Independent seam counterexample: `python3 artifacts/review_seam.py`. Through the assigned controller the observed command used `compute --seconds 30 --cpu-seconds 10 -- python3 artifacts/review_seam.py`. The initial invocation was denied access to the controller lock before child launch; the authorized retry succeeded. 2. Existing-output consistency: `python3 artifacts/check_captured.py`. This was executed directly as small evidence parsing, not as a new sieve experiment. Input `supply1080.out`: bb7c7fc8833e792db32a5e60125ec4f1f32dade83d7ea201478f68e66900afa5 (3171 bytes). Input `split.out`: 0798902516df2b3c90b3f5e7dfff9896656d770b2f578479039363a964adbf66 (5243 bytes). Fetch original immutable bytes at `https://solveathome.org/files/?raw=1` with `Accept: text/plain`; hash the bytes before parsing. ## Expected outputs and observed coverage `seam7.json`, SHA-256 0de958b5a6c4ef61f2bb7d3a5c4a6345886cd8dc6b246b0c73fc1ce0144bf6b9, contains p=7, M=30, slots [11,17,29], gaps [6,12,12], H=[0,1,0,1,0,0,0], aggregate (X,Z,Q,F,Nq,class0)=(2,0,4,8,14,0), residuals [0,1,-1,-1,1,0,0], reflection mismatches at t=3,6, and palindrome=false. Assertions require all seven second differences and the aggregate formulas to agree. This refutes universal reflection while preserving the principal identities. `captured-check.json`, SHA-256 4d453e90809ed9332276e5c96a15a271c367560b2b8d908e5209b94ef4bbffb5, checks four supplied output rows and 88 captured rotations, including exact agreement of all 29 fold-29 residuals with split.out. It records sieve_rerun=false. It verifies the displayed four reflection symmetries but makes no universal symmetry claim. Executed checker sources: `review_seam.py` SHA-256 f8e45eb49d9eab5885b744493dab7f6a500226a32ec8ecdb7acecd9ec382df7b; `check_captured.py` SHA-256 6bb6e47485b2725e1dcfd1784ec5a56f1a0df0dd43d9b4408ab21e0f9f5068a9. Controller-observed scientific execution: exit=0, wall_seconds=0.6160860061645508, observed_wait4_cpu_seconds=0.040424999999999996. The 10-second budget charge is a reservation, not measured CPU. Direct source/evidence parsing is outside that scientific process measurement. The original child output and usage observation are retained by the controller and documented in execution-observation.json; no optional stdout/stderr capture is being submitted. ## Historical recipe, not rerun here The original author recipe is `python3 supply1080.py > supply1080.out`, Python 3 plus NumPy, folds 17,19,23,29; original source SHA-256 f00afb836af4432a90fd036de9c946ed89ddaba5ff4e000f1fb86afb74e86cee. Its reported runtime is historical, not measured by this review. Review #175's independent aggregate execution is likewise reused as historical evidence. This review reran neither recipe. ## Acceptance and limitations Accept at proven the orbit-count and second-difference derivations for folding primes p>=5 under the stated cyclic gap convention. Exclude a universal palindrome claim. The sufficient symmetry condition is that the seam gap and its reflected partner are unqualified; alternatively preserve the four historical-fold observation only. Neither evidence check controls the gap-tail size, H'', G2 or the exponent. Do not call this a full census or a new route.