Sections 1.9 and 1.10 in brief
No measurement is perfect. Section 1.9 sorts errors into three kinds, each reduced in its own way, and section 1.10 states how uncertain a reading is. Precision, accuracy and rounding off (1.12 and 1.13) have their own Precision and accuracy page.
Three kinds of error (1.9)
| Error | Where it comes from | How to reduce it |
|---|---|---|
| Human (personal) | The observer: misjudging a pointer between two marks, an eye not straight in front of the mark, a wrong reading method, or slow reactions with a stopwatch | Training, careful technique and no distractions; best of all, automatic or digital instruments |
| Systematic | A fault that changes every reading by the same amount: a zero error, poor calibration or wrong marking of the scale | Compare the instrument with a more accurate one, then apply a correction to each reading |
| Random | Unpredictable changes the observer cannot control, such as a draught, or small changes in humidity, room temperature or the mains voltage, so repeated readings differ | Take several readings and use their mean |
A mean of many readings cannot cancel a systematic error, because every reading carries the same shift.
Worked numerical 1: a worn metre rule
The first 3 mm of a metre rule have worn away, so its end now sits at the 0.3 cm position. A pencil laid against that end reaches the 15.8 cm mark.
Safer still, avoid the worn end: start the pencil at a clear mark such as 1.0 cm. With the tip at 16.5 cm, the length is again cm.
Uncertainty in one reading (1.10)
Every measured value is uncertain: a scale shows nothing finer than its least count, its smallest division. Only counting, such as the bricks in a pile, is exact.
Your book reads a scale using the midpoint of a division. If the end of the object stops short of halfway between two marks, record the lower mark; if it reaches or passes halfway, record the higher one. Either way, the true end lies within half a division of the recorded mark:
Worked numerical 2: the halfway rule
A wire's end lies between the 12.6 cm and 12.7 cm marks of a metre rule, whose least count is 0.1 cm.
- If it stops before 12.65 cm, record 12.6 cm.
- If it reaches or passes 12.65 cm, record 12.7 cm.
So the first reading is written cm. The same rule gives ±0.005 cm on Vernier Callipers and ±0.005 mm on a screw gauge. On a digital display the last digit may flicker; Table 1.7 rates a digital stopwatch at ±0.01 s and an analogue one at ±0.1 s.
Worked numerical 3: thirty swings
Section 1.10 adds that repeating and averaging shrinks the uncertainty of short times and small lengths. A student times 30 oscillations of a pendulum: 48.6 s, uncertain by about ±0.2 s because of reaction time at the start and the stop.
Timing a single swing would leave the whole ±0.2 s on a reading of about 1.6 s.
Using the simulation
In the simulation above, switch to Readings. The Ruler is set up, and no readings have been taken yet.
- Press Read 5. Its worn end gives this ruler a zero error of +2 mm: every reading of the 86.6 mm pencil comes out at 88 or 89 mm, and Mean − true value shows +2.0 mm. Press Read 5 again: the gap stays. Drag Zero error (worn end) to 0 to remove it.
- Under Random error, press Large. The readings spread out and the Range grows, but the mean hardly moves.
- The One reading card shows ±0.5 mm, half the ruler's 1 mm least count. The Half the range card is the estimate in the box above.
- Choose Stopwatch from the instrument menu. Every time is stopped late: the note under the slider explains that your book calls this a human error, though it acts like a systematic one. One reading now shows ±0.01 s.
- Turn on Include an odd reading (number 4). The lab note says what went wrong: a human error. Choose Leave out or Repeat it.
On a phone, the Controls tab under the stage holds the sliders and switches, and the Readings tab holds the cards.