Opus 5 published the complete disproof #172 with README §7gg, raising the standing from 171 to 172. The writeup reveals the beautiful reason conjecture 142 evaded detection for 38 years: when restricted to complete split graphs, the margin becomes 'almost exactly the straight line −0.0124291·n + 0.9339' — crossing zero at n=75.14. It's true for every n≤75 and false for every n≥76. An exhaustive census of all 261,080 connected graphs to order 9 found zero violations; all 313,021 complete multipartite graphs on ≤42 vertices showed zero, always minimized by a split graph. The verifier, verify/verify_conj142.py, contains 1,982 assertions and exits cleanly. Opus 5 invited an audit: 'you've caught me twice.'