

DOESTD112198
2.5 CONVENTIONS FOR ROUNDING AND SIGNIFICANT FIGURES
The need to distinguish between exact numbers, such as dose limits, and imprecise numbers, such as
the results of measurements, leads to the use of some conventions in this DOE Standard.
Any legislated number, as well as any integer or ratio of two integers, is an exact number with no
uncertainty1. Examples of exact numbers are the 5rem annual TEDE limit, the DAC for radon progeny
have tolerances, but when tolerances are not specified, the exact numbers must be treated as arbitrarily
precise: a 5rem limit is 5.000 000 000 rems.
Measurements are often uncertain and imprecise, and inferences of dose from measurements using
calculational models with uncertain parameters are also uncertain and imprecise. Confusion sometimes
results when comparing uncertain or imprecise numbers with exact standards. Furthermore, difficulty
arises when exact numbers are converted from one set of units to another and the result is rounded. This
difficulty becomes particularly acute for quantities and units associated with radon and thoron. Thus, the
DOE has decided to derive all radon and thoron concentration values from 10 CFR 835, Appendix A,
PAEC limits (or ICRP/IAEA PAEE limits for newer recommendations), rather than from 10 CFR 835,
Appendix A, equilibrium equivalent DACs, which give slightly different answers and lead to confusion
(Strom et al. 1996).
Excellent, detailed guidance on significant figures and rounding for measurements is given by the
ASTM (ASTM 1993). Unfortunately, ASTM E38093 does not recognize the exact nature of regulatory
limits nor address the problems of significant figures when converting exact numbers between unit
systems. Also, it does not address radon and thoron quantities and units.
To minimize roundoff errors, it is recommended that all calculations be performed using numbers
specified to at least "single precision" (six to seven significant figures) or as rational numbers if
appropriate (ratios of integers, e.g., 1/3, 5/12, etc.) For reporting purposes, it should be acceptable to
round to three significant figures or to the precision of the reporting field, whichever is less. For example,
3.84 mrem may be rounded to 4 mrem if only integral numbers of mrem can be reported in a given field.
More detail on recording and reporting is given in Section 9.
Similarly, in this DOE Standard, all numbers that are simply unit conversions are expressed to 5
significant figures to prevent contradictions or inconsistencies.
Some irrational numbers are exact, such as B, o2, and e, the base of the natural logarithms.
1
Sometimes, unit conversions are exact, such as 37,000,000,000 Bq/Ci.
19

Privacy Statement  Press Release  Copyright Information.  Contact Us 
Integrated Publishing, Inc. 