One Reading Is Not a Measurement: Repeat It, and Look at the Spread

Reading the same dial twice tells you nothing. Repeat from the START: set the work down and pick it up again, re-close the caliper, restart the timer, re-tare the scale. Most of the variation in a hand measurement comes from how the instrument was PLACED, not from the instrument. A caliper read three times without moving it gives three identical numbers and a false sense of precision; re-closed three times it gives the real spread. Five or six repeats is the practical minimum. Three tells you only whether something has gone badly wrong; six starts to give a spread you can quote.

Record all of them, in order, as you take them. The reading that looks wrong is the most informative one in the set, and it is the first to vanish if you wait and write down what you remember. A reading may be DISCARDED, but only for a reason you can state — the piece slipped, the timer was started late, the balance had not settled. Write the reason next to it and keep the number. Deleting an outlier because it is inconvenient is how a measurement quietly becomes a wish. In order matters too: a set that drifts steadily upward is telling you something a shuffled set cannot. Drift usually means something is warming up, settling, drying, or being worn away by the measuring.

Write the result as the mean plus or minus the scatter of the mean — for six readings, the sample standard deviation divided by the square root of six. That one line says what you found and how hard you looked. Say how many repeats went into it. "2.01 s" and "2.01 plus or minus 0.05 s, n=6" are different claims, and only the second can be checked or repeated by anyone else. And round the uncertainty first, to one or two digits, then round the value to match it. A result quoted as 2.0134 plus or minus 0.05 is announcing four digits it cannot support. The measurement-uncertainty rung takes that arithmetic further; this rung is only the part you can do with six numbers and a calculator.