>Jack Good, a veteran of Colossus practice at Bletchley Park, later claimed that, if appropriately configured, Colossus could almost have carried out a multiplication but that this would not have been possible in practice because of constraints on what could be accomplished in a processing cycle. We have no reason to doubt this, though it would presumably have required special settings of the code wheels and message tape and been, even if possible, a rather inefficient alternative to a desktop calculator. This fact has been offered as proof of the flexibility of Colossus, which in a sense it does attest to: a device designed without any attention to numerical computations could almost have multiplied thanks to the flexibility with which logical conditions could be combined. Yet it also proves the very real differences between Colossus and devices designed for scientific computation. Multiplications were vital to computations, and a device that could not multiply would not, by the standard of the 1940s, be termed a “computer “or “calculator.”

https://www.sigcis.org/files/Haigh%20-%20Colossus%20and%20th...

The limitation seems to have been physical rather than logical.