[Test-Equipment] Standard cell averages

[Alan Melia]

I think Bob is taking a rather narrow view of the problem. What one is
trying to do, if one is not a standards lab, it to determine the most likely
value of a volt. Like the man with two watches the most likely time is the
average of the two....it doesnt mean its the RIGHT time!!
If you have no way of telling which cell is "right" you can get a "best
estimate" by taking an average. You would also be justified in "censoring"
the cells that did not lie withing a certain number of standard
deviations....even if the distribution was not known to be gaussian. You
have a rule and you use it to define the "your" volt as best you can. As to
the chemical effects not necessarily generating a gaussian error
distibution, why not? What mechanism causes the differences? Might it not be
minute levels of contamination, which could well be randomly distributed. If
you had enough cells you could test whether it was close to gaussian,but I
guess not many amateurs have more than one or two cells. The important thing
when using "statistics" on such a small sample is to remember that the
result may not be particularly significant. The more likely cause of
differences for older cells might be misuse (drawing excess current), but I
have no idea what that might do or how permanent the effect. My cells are
the unsaturated variety which are more rugged but less trustworthy.

Interesting topic ...standards and how best to do it.

Cheers de Alan G3NYK