The Duckweed Fertiliser Trial
Priya has heard that liquid fertiliser makes pond plants "explode with growth", and she wants to test this for her science fair. She fills five identical 250 mL glass jars with 200 mL of pond water and adds ten duckweed plants to each. Her plan is simple: jar A gets three drops of fertiliser and sits on the sunny windowsill; jars B, C, D and E get no fertiliser and sit inside a dark cupboard, because she wants a "control group". After seven days she will count the duckweed fronds in each jar. On day 2 a classmate knocks over jar C and about half the water spills out, so Priya tops it back up from the tap. On day 5 she notices a film of green slime on the inside of jar A. On day 7 she counts 46 fronds in jar A, 12 in B, 14 in D and 11 in E. She concludes that the fertiliser quadrupled the growth.
1. Identify the flaws in Priya's experimental design and explain how each flaw weakens her conclusion.
Read the worked answer
Priya changed two things at once: fertiliser and light. Jar A is the only fertilised jar and the only jar in sunlight, so a rise in growth cannot be credited to the fertiliser - duckweed photosynthesises, and more light alone would increase growth. Second, she has only one fertilised jar (no replication), so she has no way to know whether 46 fronds is typical or a fluke. Third, she never recorded the starting number of fronds - she added ten plants, but each plant may already carry several fronds, so the baseline is unknown. Fourth, topping up jar C with tap water changed both the volume and the water chemistry (chlorine, minerals), so jar C is no longer comparable. Fifth, the green slime is almost certainly algae, which competes with duckweed for light and nutrients and is itself stimulated by fertiliser - another uncontrolled influence. Any one of these could produce the result; together they make "quadrupled" an overclaim.
2. Rewrite Priya's investigation as a fair test and describe one improvement to how she measures growth.
Read the worked answer
Fair-test design: use ten jars, five randomly assigned to fertiliser and five to no fertiliser; all ten on the same windowsill, with positions rotated every two days so no jar gets a permanently shaded or sunnier spot; identical pond water from the same container, identical volume, no tap-water top-ups (mark the water level and top up with stored pond water only if needed). Record the initial frond count for every jar so growth is a change, not a final total.
Measurement improvement: counting fronds alone ignores frond size. She could also photograph each jar from directly above against a grid and estimate percentage surface coverage, or at the end blot the duckweed dry and measure total mass on a balance. Report the mean and the range for each group rather than single numbers; a range that does not overlap between groups is far stronger evidence than "46 versus 12".
3. What should Priya do with the data from jar C, and how should she report it?
Read the worked answer
Jar C should not be pooled with B, D and E. Half the water was lost and replaced with tap water, so both the volume and the dissolved chemistry changed after the experiment began. That makes it a different treatment, not a replicate of the control.
The honest approach is to report it transparently: include a note that jar C was disturbed on day 2 and was refilled with tap water, state its count (14 fronds) separately, and exclude it from the control mean. If she has enough time she should re-run the jar so the control group has five undisturbed replicates again. Silently dropping the data would be worse than keeping it - a reader needs to know the accident happened to judge whether the exclusion was reasonable. Excluding it is defensible because the reason is known in advance and is unrelated to the result; excluding a jar because its number looked inconvenient would not be.