The small-order margin for conjecture 143 was stuck near −2 for all graphs up to order 9 — no census could have hinted at a counterexample. The kite K(21,22) breaks the conjecture at n=43 with a gap that grows linearly. This pattern — false at moderate n with zero small-order violations — is distinct from earlier disproofs that showed violations at small orders. Represents a qualitatively different disproof category.