Uncertainty and significant figures

Measurements always have uncertainties. If you measure the thickness of the cover of a hardbound version of this book using an ordinary ruler, your measurement is reliable to only the nearest millimeter, and your result will be 3 mm. It would be wrong to state this result as 3.00 mm; given the limitations of the measuring device, you can’t tell whether the actual thickness is 3.00 mm, 2.85 mm, or 3.11 mm. But if you use a micrometer caliper, a device that measures distances reliably to the nearest 0.01 mm, the result will be 2.91 mm. The distinction between the measurements with a ruler and with a caliper is in their uncertainty; the measurement with a caliper has a smaller uncertainty. The uncertainty is also called the error because it indicates the maximum difference there is likely to be between the measured value and the true value. The uncertainty or error of a measured value depends on the measurement technique used.

   We often indicate the accuracy of a measured value—that is, how close it is likely to be to the true value—by writing the number, the symbol {, and a second number indicating the uncertainty of the measurement. If the diameter of a steel rod is given as 56.47 { 0.02 mm, this means that the true value is likely to be within the range from 56.45 mm to 56.49 mm. In a commonly used shorthand notation, the number 1.64541212 means 1.6454 { 0.0021. The numbers in parentheses show the uncertainty in the final digits of the main number.

   We can also express accuracy in terms of the maximum likely fractional error or percent error (also called fractional uncertainty and percent uncertainty). A resistor labeled ;47 ohms { 10%< probably has a true resistance that differs from 47 ohms by no more than 10% of 47 ohms—that is, by about 5 ohms. The resistance is probably between 42 and 52 ohms. For the diameter of the steel rod given above, the fractional error is 10.02 mm2>156.47 mm2, or about 0.0004; the percent error is 10.000421100%2, or about 0.04%. Even small percent errors can be very significant (Fig. 1.7).

   In many cases the uncertainty of a number is not stated explicitly. Instead, the uncertainty is indicated by the number of meaningful digits, or significant figures, in the measured value. We gave the thickness of the cover of the book as 2.91 mm, which has three significant figures. By this we mean that the first two digits are known to be correct, while the third digit is uncertain. The last digit is in the hundredths place, so the uncertainty is about 0.01 mm. Two values with the same number of significant figures may have different uncertainties; a distance given as 137 km also has three significant figures, but the uncertainty is about 1 km. A distance given as 0.25 km has two significant figures (the zero to the left of the decimal point doesn’t count); if given as 0.250 km, it has three significant figures.

   When you use numbers that have uncertainties to compute other numbers, the computed numbers are also uncertain. When numbers are multiplied or divided, the result can have no more significant figures than the factor with the fewest significant
figures has. For example, 3.1416 * 2.34 * 0.58 = 4.3. When we add and subtract numbers, it’s the location of the decimal point that matters, not the number of significant figures. For example, 123.62 + 8.9 = 132.5. Although 123.62 has an uncertainty of about 0.01, 8.9 has an uncertainty of about 0.1. So their sum has an uncertainty of about 0.1 and should be written as 132.5, not 132.52. Table 1.2 summarizes these rules for significant figures.

   To apply these ideas, suppose you want to verify the value of p, the ratio of the circumference of a circle to its diameter. The true value of this ratio to ten digits is 3.141592654. To test this, you draw a large circle and measure its circumference and diameter to the nearest millimeter, obtaining the values 424 mm and 135 mm (Fig. 1.8). You punch these into your calculator and obtain the quotient 1424 mm2>1135 mm2 = 3.140740741. This may seem to disagree with the true value of p, but keep in mind that each of your measurements has three significant figures, so your measured value of p can have only three significant figures. It should be  stated simply as 3.14. Within the limit of three significant figures, your value does agree with the true value.

   In the examples and problems in this book we usually give numerical values with three significant figures, so your answers should usually have no more than three significant figures. (Many numbers in the real world have even less accuracy. An automobile  speedometer, for example, usually gives only two significant figures.) Even if you do the arithmetic with a calculator that displays ten digits, a ten-digit answer would misrepresent the accuracy of the results. Always round your final answer to keep only the correct number of significant figures or, in doubtful cases, one more at most. In Example 1.1 it would have been wrong to state the answeras 341.01861 m>s. Note that when you reduce such an answer to the appropriate number of significant figures, you must round, not truncate. Your calculator will tell you that the ratio of 525 m to 311 m is 1.688102894; to three significant figures, this is 1.69, not 1.68.
   When we work with very large or very small numbers, we can show significant figures much more easily by using scientific notation, sometimes called powers-of-10 notation. The distance from the earth to the moon is about 384,000,000 m, but writing the number in this form doesn’t indicate the number of significant figures. Instead, we move the decimal point eight places to the left
(corresponding to dividing by 108) and multiply by 108; that is,

384,000,000 m = 3.84 x10⁸ m

In this form, it is clear that we have three significant figures. The number 4.00x10⁷also has three significant figures, even though two of them are zeros. Note that in scientific notation the usual practice is to express the quantity as a number between 1 and 10 multiplied by the appropriate power of 10.

   When an integer or a fraction occurs in an algebraic equation, we treat that number as having no uncertainty at all. For example, in the equation v_x ² = v_0x ² + 2a_x (x - x₀)², which is Eq. (2.13) in Chapter 2, the coefficient 2 is exactly 2. We can consider this coefficient as having an infinite number of significant figures (2.000000…….). The same is true of the exponent 2 in v_x² and v_0x².

   Finally, let’s note that precision is not the same as accuracy. A cheap digital watch that gives the time as 10:35:17 a.m. is very precise (the time is given to the second), but if the watch runs several minutes slow, then this value isn’t very accurate. On the other hand, a grandfather clock might be very accurate (that is, display the correct time), but if the clock has no second hand, it isn’t very precise. A high-quality measurement is both precise and accurate.

Frequently Asked Questions