Have each group's prototype ready on their table before the lesson starts, together with their Design Brief from earlier in this project. Do not hand out test kit yet.

Ask two or three groups to read one success criterion aloud. Keep the hook light: the point is to reconnect to the brief, not to start testing.

Key question: If we only say "it looks good", have we really tested it?

Nature of STEM angle to plant now and return to later: real engineers judge a solution with evidence against the brief, then improve and explain the process.

Name the three ideas in under a minute from the short board text, then go straight into the live worked model so pupils see a real object tested within the first one to two minutes of this step. Use the table below as your own reference while you talk. Classroom shorthand for the one-change rule can stay as "improve once" if useful, but keep pupil language as "one improvement" or "change one thing".

Worked model (do this live, start within the first 1–2 minutes)

Hold up one spare or teacher-made mini example (a simple card stand or barrier strip is enough). Think aloud:

• I wonder whether this meets the criterion "stands without help for 10 seconds".
• I predict it will stand, because the base is wider than the top.
• I test by placing it on the desk and timing 10 seconds with no hand near it.
• I observed that it stood for 10 seconds, but rocked when I tapped the desk gently.
• I think it meets the stand-alone criterion, but a wider base would make it more stable next time.

Misconception to head off: pupils treat a pretty build as a pass. Push them back to the written criteria. Also stop groups changing three things at once; one improvement only.

Differentiation: less confident groups may tick each criterion met / not met with one short note; confident groups add a measured result (time, count, distance) where the criterion allows it.

Before this step

Set out shared test kit in a central place: small weights, two water stations (jug, tray and teaspoon each), two soft balls for barrier tests, two stopwatches or class timers, metre sticks or 30 cm rulers, a pool of classroom books for stand tests, and scrap card, tape and string for later improvements only (keep improve materials aside until step 5).

Classroom logistics (scarce kit)

Groups will not all need the same kit. Sort groups loosely by problem type on the board (shelter / barrier / stand / draught) and run short turns:

• Weights and books: stay at tables; most stand/holder tests need no queue.
• Water stations (2): two shelter groups test at once; next pair queues; waiting groups watch and check fairness (same pour count, same height).
• Soft balls (2) and marked roll lines: two barrier groups at a time.
• Timers (2): book out for timed stand tests; return when done. Teacher phone timer is a spare if needed.
• Draught tests: use door/board gaps with paper fans; little shared kit needed.

Call the next problem-type turn every few minutes so nobody sits idle for the full 15 minutes. Fold waiters in with questions: Was that pour the same height? What would count as a pass?

If many groups share one problem type

Two stations can still bottleneck (for example mostly shelters). Add a third water tray from classroom substitutes (roasting tin or lunchbox lid plus bottle), or stagger starts so only half the cluster tests in the first five minutes while the rest run a named fairness-watch: one observer times or counts, one checks same height or load, and both report one fair or unfair detail before their own turn. Keep waiters judging the live test, not off-task.

How groups should test (match the problem type on the Design Brief)

• Shelter or cover: place over a dry tissue or paper towel; pour a counted small amount of water (e.g. 3 teaspoons) from the same height; check whether the tissue stays dry.
• Ball barrier: roll the same soft ball three times from a marked start line; count how many times the barrier stops it entering the "flowerbed" zone.
• Book stand or holder: load the same number of books each time; check if it stays upright for a counted time.
• Draught-stopper: place at the base of a closed door or against a board gap; use a folded paper fan to make a gentle draught and feel or watch a tissue strip for movement.

Keep tests fair within each group: same load, same pour, same roll distance, same timing method.

Recording during the test: pupils jot on the Investigation Journal testing page (Prototype Evaluation sheet) as each criterion is tested. Step 4 is only a short gap-fill and star-the-weakest checkpoint, not a second full write-up. If a group has only opinions, send them back for one concrete observation.

Circulate and ask: Which criterion are you testing now? What will count as a pass? What are you keeping the same between tries?

Safety: stand back if a structure collapses; no throwing weights; wipe water at once; adult helps with any tricky cut if a repair is needed later.

What good looks like: every criterion is attempted; results are honest, including fails; groups do not rebuild mid-test; waiters are watching and judging fairness, not chatting off-task.

This is a short checkpoint only for the Prototype Evaluation testing page, not a second full write-up. Most groups should already have jotted results during step 3. Use these five minutes to finish gaps, separate observation from opinion, and star the weakest criterion, then exit crisply into the improve step.

Prompt them only if something is missing:

• each success criterion
• how they tested it
• the result (met / not met, plus a measure or clear observation)
• which one part of the design they will change next, and why the evidence points there

Do not invent extra columns or named boxes beyond what the journal page already provides. Groups that recorded well during the test should star and move on. If a group has only opinions written, send them back to add one concrete observation.

Common pitfall: writing "it was good" with no link to a criterion. Ask: Which criterion? What did you see or count?

Rules for this beat

One improvement only. If a group wants to rebuild everything, stop them and ask which single change the evidence supports.

Useful one-change examples: widen the base; add one brace or triangle; tape a gap in a shelter roof; raise a barrier lip; add weight low down on a stand; lengthen a draught-stopper.

After the change, groups must re-test with the same method as the first test so the comparison is fair.

Pacing shared kit and adult help (inside 18 minutes)

Rough guide: about 9 minutes to improve, then 9 minutes to re-test and note the result on the results page.

• Groups that only need tape, card or string improve at tables first; they do not hold the water or ball stations while building.
• Re-open the same problem-type turns as step 3 for re-tests: two water stations, two soft balls, two timers. Call shelter re-tests, then barrier, then timed stands.
• If several groups need adult cuts, queue them at your table during the improve window so re-test time is not lost later. Simple tape fixes need no queue.
• Groups waiting for a station should prepare their re-test set-up (mark the roll line, count teaspoons, stack the same books) so their turn is quick.

On the Investigation Journal results page, pupils add the re-test result and a short conclusion: what changed, what stayed the same in the test method, and whether the design is closer to meeting the brief. First-test evidence stays on the testing page; do not send groups hunting for a page labelled with an internal code.

Call a two-minute warning before presentation.

If a re-test still fails: that is useful engineering evidence, not a disaster. They should say so honestly in the presentation.

Safety: adult makes any difficult cuts; mind fingers under loads; wipe water; keep walking routes clear of prototypes on the floor.