Calculate Measurement Error

[ludzinc]

This is a question that I got wrong 20 years ago in my Uni Physics days, and I still don't know how to get it right....

Assume I'm measuring the results of an experiment and I need to take three different measures (assume length, weight and temperature).

If my length measurement has an error (accuracy) of 1%, my weight measurement has an error of 2% and my temperature measurement has an error of 5%, what is the overall error of my system?

Do I just sum the percentages? (HINT:  No, I did that and failed, and I never learnt why....)

Thanks all.

[zlymex]

Normally in a RSS(root-sum-square) way if these three variables are all directly related to the final result.
overall error  = sqrt(1%^2+2%^2+5%^2)=5.5%

[Krampmeier]

@zlymex: Correct, that is how measurement uncertainties are usually added. That is, if all the errors are uncorrelated. It also depends on how the system processes the individual values (add/substract or multiply/divide/power/...).

@ludzinc: You may want to start reading https://www.wmo.int/pages/prog/gcos/documents/gruanmanuals/UK_NPL/mgpg11.pdf at chapter 7.2 or, if you really do like formulas, check chapter 5 of http://www.bipm.org/utils/common/documents/jcgm/JCGM_100_2008_E.pdf.

[ludzinc]

Thanks!

I can see I have some light reading at bed time tonight