Where should the type information be 335

I'm familiar with that kind of problem. Usually it is not a case where a particular answer is right, but a case where you learn to live with the best data and best data reduction or analysis that is avaialable even though you could conceive of better data or a better reduction-analysis.

By that definition there are no wrong answers. That's probably not a useful criteria for evaluating computer arithmetic.

That's not a valid example because neither of the techniques is particularly superior to the other unless you are using extreme container sizes. For large diameter containers the bottom of the meniscus (actually the tangent rather than the chord) is demonstrably superior to the top and the error grows as the 1-3rds power of the container size, whereas the chord error grows as the 2-3rds power of the container size.

For small diameter containers of considerable length (think fine glbutt tubes) the meniscus tends to be irrelevant. For short narrow containers the only acceptable answer is the average meniscus point because the curve of the meniscus is an appreciable fraction of the volume to be measured.

In all cases there is a stratospheric view in which minor details, such as rounding, sampling bias, or measurement conventions-limitations are not important or even relevant. But in most cases the transition of those details to relevance or even importance is silent. It is that opporunity for quiet corruption that needs to be eliminated in order to declare a method, calculation, approximation, or answer "best" and justifies however long it takes to defend against silent, hidden errors.

tj3