When you report, mention or show estimated numbers with THREE SIGNIFICANT FIGURES (say: 101 or 0.0 526 or 968 000 ) you in fact suggest that your number has an accuracy of better than + or - 0.5%, because the number 101 is either exactly 101 or a rounded number of a value between 100.5 and 101.4 An engineer with helicopter view will never be interested in such a high accuracy when doing calculations.

Only ACTUALLY SPENT MONEY figures deserve this high accuracy and then only for financial and/or government reports. Also in actual design drawings three or more significant figures should be used.

Reporting numbers with TWO SIGNIFICANT FIGURES (say: 0.11 or 49 or 97 000 000) suggests an accuracy of better than 5% , 1% and 0.5% respectively. Even here the 49 and 97 000 000 may be too accurate in many cases and should be rounded to 50 and 100 000 000 respectively.

I still have to meet the first engineer that can estimate, whatever matter, with higher than + or - 5% accuracy!!! And of course a round number such as $ 140 million is better remembered than $ 142.5 million.

WHY BOTHER TO REPORT ESTIMATED OR ORDER OF MAGNITUDE NUMBERS WITH MORE THAN TWO SIGNIFICANT FIGURES!!!!! IN ESTIMATES DO NOT REPORT THE SMALL ITEMS BUT PUT THEM TOGETHER.

Examples in oil and gas engineering are:

Field gas volume or mass measurement is at best say 5% plus or minus accurate, so what was said above for estimating is also true for gas quantity reporting. Only government reports require more "significant" figures.

Do not get fooled by implied laboratory accuracies, for instance in gas compositions. A methane content, reported as 86.55% mol has two meaningless numbers behind the decimal point, even if high accuracy of the gas chromatograph is claimed. The real problem is: How to take a good representative sample!! Only by taking many samples one gets a feel for the range of possible variation. The above number in isolation probably means that the methane content is between 85 and 88% mol. It is very important when writing terms of reference for a study to state a range or the assumed accuracy of the numbers; namely the design contractor may claim that, when the plant does not perform according to design during the test run , the methane content was not 86.55 but different and give this as a reason why the plant does not perform.

Do not report rounded numbers expressed in Imperial units in numbers with many significant figures when expressed in SI units but round the SI number also.