This is one of the things that reading D.K. Brown's pentology on design of RN ships (_Before the Ironclad_, _Warrior to Dreadnought_, _Grand Fleet_, _Nelson to Vanguard_, and _Rebuilding the Royal Navy_) drives home: in the pre-WWI era almost everything is done with fudge factors and building off of what was done before, but a little different, and even into the 1950's there are lots of room for error.
One story that has stuck with me is of a destroyer class in WW2 that performed particularly horribly because of an error on calculating the metacentric height. The proper procedure for calculating it was to have two different dudes each spend a week calculating it independently and hoping that their numbers matched. Apparently, in the rush of what had to be done, the RCNC cut some corners and only had one draftsman do it in this case, and he made an error and now the ships rolled horrifically. But even Bouguer and Euler- the people inventing metacentric height calculations- are working a century after the Vasa, so a master shipbuilder in that era just has his working experience and no real math to help him.
If you follow the excellent YouTube channel Drachinifel you will learn that many ships of that era (including WW2) had design flaws, like heavy rolling, flooding into compartments, breaking of the hull etc. This was the time before computer models and computer simulation.
One story that has stuck with me is of a destroyer class in WW2 that performed particularly horribly because of an error on calculating the metacentric height. The proper procedure for calculating it was to have two different dudes each spend a week calculating it independently and hoping that their numbers matched. Apparently, in the rush of what had to be done, the RCNC cut some corners and only had one draftsman do it in this case, and he made an error and now the ships rolled horrifically. But even Bouguer and Euler- the people inventing metacentric height calculations- are working a century after the Vasa, so a master shipbuilder in that era just has his working experience and no real math to help him.