She nodded, not understanding, but grateful.
Punmia’s example 8.4 showed a classic case: a hidden leak in a secondary branch, impossible to find by listening, but mathematically obvious if you calculated the nodal residuals. The margin note— “leak suspect” —was from some long-dead student, but for Arjun, it was prophecy.
The first iteration failed. Residuals scattered like frightened birds. The second, worse. By the fourth, a pattern emerged. Node 12, a junction near the old Hanuman temple, showed a correction term of +0.32 m³/hr—small but persistent. According to Punmia’s logic, that meant water was leaving the system there, not reaching the end users.