Corollary 3 as a word RAM program: the dual table of Section 3 for equal preprocessing times. Read the word, sort the jobs by start time, and fill the table over sets of thresholds written as base-$(n+1)$ numbers exactly as for Theorem 2, but with rows the $n + 1$ instants $u · p$ (the instant a partial solution has spent after selecting $u$ jobs) instead of the $W + 1$ weights: the cell for the instant $u · p$ is $max (T[X₁, u]) (min W (w_j + T[X₂, u + 1]))$, the second term present when $(u+1) p + q_j ≤ d_j$. The guard is tested as $u < lim_j$, where $lim_j$ is computed once per job by a division, so the machine never forms the product $u · p$ (only the entries of the word are bounded by the word length). The table has $(n+1)^m · (n+1)$ entries, filled from the largest number down, and the answer is one entry. The domain restricts to positive processing times, the paper's standing assumption for Lemma 1's recursion.