## ABSTRACT

Every observation contains error and this chapter considers the error's effect on the results computed from the observation. For effective analysis, systematic, random, calculation, sampling, and other errors must be properly estimated and propagated. The observation's accuracy indicates its closeness to the true value either known or presumed, whereas its precision refers to its repeatability, and its bias is the difference between its accuracy and precision. The differences between absolute and relative accuracy as well as precision are reviewed. Also considered is the number of significant figures required to report an observation and its use in addition, subtraction, multiplication, division, and round-off operations. Taylor's theorem gives the propagated error for a function involving m-parameters with related uncertainties. The function's first order error, accordingly, is the sum of m-terms where each is the product of the parameter's uncertainty times the function's change due to the parameter's change. Dividing the function into the propagated error yields the propagated relative error. Error analysis complements the graphical and numerical methods as a third fundamental approach for quantifying data variations in terms of environmental parameters.