The Stream After Three Days of Rain
You are taking 34 Form 3 learners to a perennial stream 3 km from school for a two-hour fieldwork exercise on discharge. Permission has been granted. After three days of rain the stream is high and brown, and the bank at your intended first cross-section is undercut. You have one 30 m tape, two metre rules, two floats, a stopwatch and three 1 m ranging poles. Two learners cannot swim. The bus leaves at 09:00 and must return by 11:30. A parent has emailed to ask whether the trip is safe. Last year, one group recorded a cross-sectional area of 4.8 m2 at a site where the tape had been held diagonally across the channel. Their discharge figure was nearly double the other two groups' figures, and the class concluded that the stream 'speeded up in the middle'.
1. Decide whether to proceed, relocate, or cancel. Give your reasoning with reference to both safety and data validity.
Read the worked answer
Safety first. The undercut bank is the deciding factor: it can fail without warning, and a learner standing on it would enter fast, turbid water. That site must be abandoned regardless of its teaching value. The high, brown water also makes the original plan to wade mid-channel inappropriate.
Decision. Proceed, but relocate to a straight, shallow riffle with a stable gravel bank. Restrict in-channel work to two adults plus three learners who are confident swimmers, each supported by a ranging pole and never beyond 0.4 m depth. The two non-swimmers take bank roles, holding the tape, timing and recording, so they are fully included without entering the water. Cancel only if no safe riffle is found within 15 minutes of arrival; a lesson can be recovered in class, an incident cannot.
Data validity. The diagonal tape inflated width, and therefore area; the conclusion that the stream speeds up in the middle was an artefact of measurement error, not a finding. Fix the protocol before departure: tape perpendicular to flow, checked against a pole on each bank, width measured three times, depth read at five equidistant points. State both decisions to the parent in those terms: what changed and why.
2. How will you organise 34 learners so that the data from each group can be compared?
Read the worked answer
Structure. Divide into three groups of 11 or 12, each with a named recorder and a named safety spotter. Groups rotate every 20 minutes through three stations: channel geometry, velocity, and bank sketch with site description. Rotation means every learner performs every measurement, so the data set is comparable and nobody is idle near the water.
Standardising the measurement. Mark the cross-section permanently with the three ranging poles: one on each bank and one at the midpoint. Every group measures the same section, so differences between groups reflect measurement error rather than different sites. Width is measured three times and the median recorded. Depth is read at five equidistant points on the tape, not at chosen 'interesting' spots. Velocity is the mean of three float runs, with the float released upstream of the 10 m mark.
Recording. Use a single pre-printed sheet with units in the column headings: m, s, m/s, m2, m3/s. This prevents the unit confusion that produced last year's impossible values. Collect sheets at the bus and photograph them as a backup.
Why rotation rather than division of labour. If each group performs only one task, no learner can explain the whole calculation, and the examination asks them to.
3. How will you handle last year's anomalous 4.8 m2 value when the class analyses the data?
Read the worked answer
Do not delete the outlier. It is the most useful number in the data set because its cause is traceable. Use it in the follow-up lesson as a worked error analysis.
Sequence. Show the three area figures alongside the measured widths and ask which is inconsistent. Then reveal the field note that the tape was diagonal. Learners calculate the effect: a tape held at about 45 degrees across a 6 m channel measures roughly 8.5 m, inflating area by around 40% and discharge by the same factor. That explains the near-doubling without any change in stream behaviour.
Name the concept. This is systematic error: a biased method applied consistently. It does not average out with more readings; only changing the method removes it. Random error, from timing or depth reading, does shrink with repeats. Distinguishing the two is the learning objective.
Follow-through. Ask learners to state one protocol change that would have prevented the error, then to identify which of this year's measurements are most exposed to random error. Recompute the class mean with and without the outlier and discuss why a median is more robust in a small data set. The outlier becomes evidence for method evaluation, a higher-order skill than the calculation itself.