a +/- 0.1% meter, reading "100.00"

[WhyDoYouAsk]

The likelihood of the true value of what you're measuring actually being 100.00 is virtually zero.
There's a 95% chance the true value is between 100.10 and 99.90.
There's a 68% chance the true value is between 100.05 and 99.95.

Agreed?

 

[VEBill]

Your estimate of "virtually zero" would actually be about 1-in-21 (about 5%), given an even distribution of the possible counts within that stated error range.

And, lacking any evidence to the contrary, you'd have to assume an even distribution. The error might be quite systematic and primarily composed of an unknown fixed offset guaranteed to be within the specified limits. Of the total error budget, it's likely that only a small part is made up of natural variations where bell curve math would apply.

 

[WhyDoYouAsk]

Thanks.
This may come up a lot for me, so aside from the uniform distribution assumption do you have a link for this kind of not-very-well-known info (interpreting meter tolerance specs)?

 

[waross]

If this is going to come up a lot, consider selecting test instrumentation with greater precision.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[VEBill]

You could start here:

http://en.wikipedia.org/wiki/Accuracy

But keep in mind that some people literally spend a lifetime immersed in this subject.

 

[WhyDoYouAsk]

Thanks for the link.
Re: lifetime, I might have once applied to the NIST; I can't recall. It's only up the road from me.

 

[IRstuff]

NIST does not use the term "accuracy," they use "uncertainty."

http://physics.nist.gov/cuu/Uncertainty/index.html

TTFN

FAQ731-376

 

[WhyDoYouAsk]

Ask, and I shall receive! Thanks again.

 

[Comcokid]

There are other reasons for that last decimal point other than for makng exact measurements. Afterall, the world is dynamic, not static. That extra decimal place serves to show slow drift or variations in the readings.