Take a polynomial of degree four, with positive coefficients and four real roots. Raise every coefficient to the same positive power, while leaving the powers of the variable alone. Which exponents guarantee that all four roots stay real, for every polynomial in this class? This anonymous, unrefereed candidate gives a sharp answer. Exponent one works because it changes nothing. Immediately above one, some polynomials lose real roots. The universal guarantee returns at about one point one four seven seven two, and holds for every larger exponent. Above that threshold, all four output roots are distinct. At the threshold itself, a repeated output root occurs only when all four input roots coincide. The hard part is not finding a symmetric example. It is showing that no unsymmetrical input, including inputs with roots at very different scales, creates a worse obstruction. The written proof uses root perturbations, a boundary argument, and a reduction to reciprocal quartics. Exact code checks selected algebra and a narrow rational enclosure of the threshold. It does not certify the whole analytical proof. Earlier work already established cubic preservation, quartic counterexamples, and sufficient large exponents. This release proposes the sharp quartic endpoint and its equality classification, without claiming an all-degree solution or established historical priority. This is the Evidence Press release of fifth September twenty twenty-six. The unrefereed paper and full evidence package are linked on this page. External specialist validation remains outstanding. This AI-generated OpenAI voice is a communication aid, not additional mathematical evidence.