Let me focus on the case that introduced the thread. First, I want to say that I do not know about the particulars of the case. However, there is a point that everyone seems to have missed, and that is whether the accused was following the standard operating procedure (SOP) for preparation of the dose. I worked for about 16 years in the medical field, specifically clinical laboratory science. Clinical laboratory science is not necessarily a perfect analogy to preparing a dose of medication, but for purposes of discussion I am going to assume that the analogy is quite close.
In my profession every laboratory method was strictly specified by an SOP. Let us assume that the preparation of a therapeutic dose is similarly specified. (Someone correct me if you know this to be wrong.) In this case, if the type of scales used to prepare the dose are specified in the SOP, and if the amount of the dose was within the SOP specification, and if the person preparing the dose followed the SOP (i.e. used the scales specified at the amount specified in the SOP), then the person who prepared the dose was not at fault. The fault would be with the person who wrote the SOP as well as everyone who signed off on the SOP.
There is also another subtlety. In my profession the calibration of all scales had to be checked on a specified schedule, such as yearly. Also, any instrument that had been exposed to some kind of change would also need to undergo some kind of re-validation. For example, if an instrument were moved from one location to another a re-validation would be required. I don't recall if scales fell under this regulation, but I would be surprised if they did not.
An additional subtlety relates to training. I think this might have been mentioned in the thread already, but if so I am mentioning it again. If the SOP specified the use of a certain scale and that the scale be operating within specific parameters, then the organization is responsible to train the person preparing the dose on these requirements. If proper training was not carried out, including a proficiency test, then logically the fault lies with the organization, not necessarily with the person preparing the dose.
Finally, I would like to briefly mention error budgets. I won't go into detail, but a proper analysis of the error budget for a measurement folds both accuracy and precision together into an overall error budget, and typically the error budget is given by a graph that encompasses the allowable measurement range.