G2 Science K223, K224 and K225 systems reveal useful information when they are disturbed. A small perturbation can test whether a system returns toward its earlier state, settles at a new level, overshoots, oscillates, or fails to recover. The recovery pattern can distinguish competing mechanisms.
This Learner’s Guide is designed to upgrade capability rather than merely add facts. It makes the mechanism explicit, shows how to diagnose failure, defines the smallest useful repair, keeps competing explanations visible, states model limits, and ends in transfer to a changed problem surface.
Mechanism
Identify the relation that should remain true and the condition that controls how the answer changes. Write the causal, logical or quantitative handoff explicitly enough that another learner could predict the next state before seeing the result.
Diagnosis and smallest repair
When the answer fails, locate the first broken handoff. Repair that node only: source, role, denominator, sign, condition, mechanism, boundary or measurement. Recheck one downstream consequence before changing anything else.
1. small temperature perturbation
For small temperature perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
2. small concentration perturbation
For small concentration perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
3. light perturbation
For light perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
4. water perturbation
For water perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
5. nutrient perturbation
For nutrient perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
6. humidity perturbation
For humidity perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
7. airflow perturbation
For airflow perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
8. load perturbation
For load perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
9. force perturbation
For force perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
10. voltage perturbation
For voltage perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
11. heat-input perturbation
For heat-input perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
12. population perturbation
For population perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
13. resource perturbation
For resource perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
14. predator perturbation
For predator perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
15. prey perturbation
For prey perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
16. pH perturbation
For pH perturbation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
17. baseline disturbance
For baseline disturbance, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
18. control disturbance
For control disturbance, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
19. short pulse
For short pulse, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
20. step change
For step change, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
21. temporary disturbance
For temporary disturbance, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
22. sustained disturbance
For sustained disturbance, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
23. remove disturbance
For remove disturbance, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
24. return toward baseline
For return toward baseline, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
25. new steady state
For new steady state, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
26. overshoot
For overshoot, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
27. undershoot
For undershoot, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
28. oscillation
For oscillation, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
29. slow recovery
For slow recovery, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
30. fast recovery
For fast recovery, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
31. incomplete recovery
For incomplete recovery, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
32. no recovery
For no recovery, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
33. delayed recovery
For delayed recovery, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
34. feedback
For feedback, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
35. negative feedback
For negative feedback, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
36. positive feedback context
For positive feedback context, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
37. buffering
For buffering, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
38. capacity
For capacity, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
39. threshold crossing
For threshold crossing, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
40. path dependence
For path dependence, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
41. hysteresis-like history effect without formalism
For hysteresis-like history effect without formalism, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
42. baseline drift
For baseline drift, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
43. response lag
For response lag, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
44. bottleneck
For bottleneck, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
45. competing timescales
For competing timescales, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
46. dynamic balance
For dynamic balance, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
47. spatial gradient recovery
For spatial gradient recovery, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
48. compartment transfer
For compartment transfer, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
49. dose-response recovery
For dose-response recovery, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
50. measurement interval
For measurement interval, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
51. recovery metric
For recovery metric, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
52. time to recovery
For time to recovery, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
53. recovery final rule
For recovery final rule, state the observation or given information first. Then state the inference, calculation or mechanism separately. This keeps what is seen apart from what is concluded.
Construct a competing explanation or near-miss that fits part of the evidence. Identify one discriminating test, counterexample, changed condition or independent check that the two alternatives predict differently. Support raises confidence; it does not turn a model into unlimited proof.
For transfer, change the surface while preserving the deep relation. Alter names, numbers, source, graph, location, time, boundary or apparatus. Rebuild the reasoning rather than recalling the example, and state where the method would stop being reliable.
Delayed transfer and measurement
Return later to a mixed problem where the framework is not named. Measure success by independent reconstruction: can the learner identify the mechanism, diagnose a near-miss, choose the smallest repair, distinguish observation from inference, test a competing explanation and state the boundary without prompts?
Internal learning links
Use the Science Hub, Vol 0127 Dynamic Balance, Vol 0139 Competing Timescales, the Examination Craft hub and the PSLE Learner’s Guide.
Official-source discipline
For the current 2027 Singapore-Cambridge Secondary Education Certificate G2 school-candidate framework, use the official SEAB G2 syllabus directory and its linked subject syllabus. The current directory lists K200 English Language, K210 Mathematics and K223–K225 Science combinations. This article’s framework is an eduKateSengkang learning method, not an additional SEAB syllabus topic or marking rule.
Final rule
A strong learner can reconstruct the relationship, locate failure, repair minimally, test an alternative and transfer the method to an unseen surface. That capability—not the length of the answer—is the standard.
