How Mathematics Examination Works | How to Check Maths Answers and Find Careless Mistakes

Checking answers in a mathematics examination works best when the check is different enough from the original solution to expose a different failure. Repeating the same calculation can catch a copied digit or arithmetic slip, but it can also reproduce the same wrong model, sign assumption or calculator entry. Strong maths checking uses substitution, inverse operations, estimation, units, bounds, alternative representations and contextual constraints to challenge the answer from another direction.

Understanding how to check maths answers, find careless mistakes and verify working in exams turns checking from a vague final ritual into a mathematical skill. The aim is not to distrust every line or solve the whole paper twice. It is to identify the transitions most likely to fail and use a check capable of detecting that particular failure.

This world-facing guide extends How Mathematics Examination Works, word problems and multi-step questions, reasoning and proof, and method selection in mixed questions.

The 50-second answer

IDENTIFY RISK → CHOOSE AN INDEPENDENT CHECK → TEST THE ORIGINAL CONDITIONS → LOCATE DISAGREEMENT → REPAIR ONLY WHAT FAILED → RECHECK THE DEPENDENT RESULT.

1. Checking is not the same as rereading

Ryan rereads 17×6=112 three times and sees nothing wrong because the written line has become familiar. Jo asks him to estimate: 20×6 is 120, so 17×6 should be a little below 120. That does not identify the exact answer, but it makes 112 plausible enough to inspect. Direct multiplication then gives 102. The estimate created a different route into the error.

2. A check has a detection target

Substitution detects many equation-solving errors. Units can detect incompatible quantities. Bounds can reject impossible magnitudes. Testing the next integer can verify a capacity maximum. A probability range check can reject values below zero or above one. No single check detects everything, so “check your work” becomes useful only when the student knows what the check is capable of finding.

3. Check the model before polishing the arithmetic

A perfectly recalculated wrong equation remains wrong. Before spending time redoing arithmetic, ask whether the equation, diagram, ratio or formula actually represents the question. If a sale price is 80 after a twenty-percent reduction, checking 80×1.2 twice cannot repair the wrong reverse-percentage model.

4. Substitute into the original, not only the simplified line

If x=5 is obtained from 3(x−1)=12, substitute into 3(x−1), giving 12. A transformed equation can hide an earlier algebra error or an extraneous root. The original question is the strongest reference point because it contains the conditions the solution was meant to satisfy.

5. Estimate before trusting precision

A calculator may display many digits with complete confidence. Estimation supplies an independent scale. If 49.8×20.1 is entered, expect roughly 50×20=1000. An output near 100 or 10,000 suggests a place-value, bracket or entry problem even before exact recalculation.

6. Units can check the kind of answer

Area should have square units, volume cubic units and speed distance per time. If a calculation intended to produce time ends in kilometres, something structural is wrong. Dimensional reasoning cannot prove every numerical detail, but it can reject answers of the wrong quantity type.

7. Boundaries deserve explicit checking

When the answer is the maximum number affordable, check that number and the next larger integer. When the answer is the minimum number of containers, check the chosen count and one fewer. This verifies both feasibility and extremality. A single substitution cannot establish both.

8. Check scope as well as value

A trigonometric equation may have another solution in the requested interval. A proof may cover positive integers but claim all integers. A statistics answer may describe a sample and then overgeneralise to a population. Checking asks not only “Is this value right?” but “Is this the complete and correctly scoped answer?”

9. Error repair should begin at the first failing transition

If the model is correct and only 7×8 was miscalculated, preserve the model and repair the arithmetic. If the first equation was wrong, later algebra may be internally flawless but dependent on the wrong start. Rebuilding only from the first invalid line saves time and keeps successful work visible.

10. A final scan should be personalised by error history

Mira often loses units; Ethan hides domain restrictions; Ben misses changed reference wholes. Their final scans should differ. Practice records can reveal which transitions deserve attention under time pressure. A generic reread of every line treats low-risk and high-risk work as though they were equally likely to fail.

Part II. Eighty ways to check a mathematics answer

11. Substitution

Best used for: equations and candidate roots. The checking move is to put the proposed value into the original relation. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid copying the same transformed equation. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

12. Inverse operation

Best used for: arithmetic and algebra. The checking move is to undo the operation independently. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid using the same mistaken forward operation. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

13. Estimation

Best used for: calculator and arithmetic magnitude. The checking move is to compare with a rounded benchmark. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid treating plausibility as exact proof. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

14. Unit analysis

Best used for: rates, geometry and applied problems. The checking move is to track quantity types through operations. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid adding units only at the final line. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

15. Sign check

Best used for: directed quantities and algebra. The checking move is to predict positive, negative or zero where possible. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid assuming a familiar answer must be positive. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

16. Range check

Best used for: probability, angles and bounded quantities. The checking move is to compare with mathematical limits. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid accepting a value outside the permitted range. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

17. Integer feasibility

Best used for: containers, people and discrete counts. The checking move is to test the chosen integer against the constraint. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid ordinary rounding without checking feasibility. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

18. Next-integer check

Best used for: maxima/minima in discrete contexts. The checking move is to test one neighbouring integer. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid showing feasibility but not optimality. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

19. Original-condition check

Best used for: multi-step problems. The checking move is to reconstruct every stated condition from the answer. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid checking only the final arithmetic. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

20. Reference-whole check

Best used for: fractions and percentages. The checking move is to ask ‘of what?’ for each fraction or percentage. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid reusing the original whole after it changed. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

21. Ratio reconstruction

Best used for: ratio problems. The checking move is to rebuild actual quantities and simplify their ratio. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid checking only total. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

22. Total reconstruction

Best used for: part-whole problems. The checking move is to add the solved parts back to the stated total. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid repeating the division. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

23. Alternative representation

Best used for: word problems and algebra. The checking move is to compare equation, diagram, table or graph. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid redrawing the same wrong representation. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

24. Graphical check

Best used for: equations and functions. The checking move is to inspect intersection, sign or shape. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid trusting a rough sketch as exact evidence. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

25. Coordinate substitution

Best used for: line equations. The checking move is to test supplied points in the final equation. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid checking gradient but not intercept. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

26. Gradient direction

Best used for: coordinate geometry. The checking move is to compare sign with visual/coordinate movement. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid using inconsistent difference order. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

27. Pythagorean check

Best used for: right-triangle lengths. The checking move is to square sides and test relation. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid using Pythagoras without a right angle. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

28. Angle-sum check

Best used for: geometry. The checking move is to sum relevant angles to known totals. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid assuming diagram scale. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

29. Similarity ratio check

Best used for: similar figures. The checking move is to compare corresponding side ratios. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid mixing correspondence. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

30. Scale-dimension check

Best used for: similarity. The checking move is to confirm linear/area/volume power. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid using same scale factor for all dimensions. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

31. Probability total check

Best used for: complete event sets. The checking move is to confirm exhaustive disjoint probabilities sum to one. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid adding overlapping events. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

32. Complement check

Best used for: probability. The checking move is to compare event with its complement. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid using complement when events are not exhaustive. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

33. Sample-space check

Best used for: probability. The checking move is to enumerate or structure outcomes. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid assuming named outcomes equally likely. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

34. Mean-total check

Best used for: statistics. The checking move is to multiply mean by count to recover total. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid averaging means without weights. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

35. Median-order check

Best used for: statistics. The checking move is to order data before locating middle. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid taking middle of unsorted list. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

36. Spread check

Best used for: data comparison. The checking move is to compare the requested spread measure. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid using mean to answer consistency. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

37. Bound interval check

Best used for: rounded measurements. The checking move is to test values near both endpoints. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid treating recorded value as exact. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

38. Rounding check

Best used for: numerical answers. The checking move is to look one digit beyond required precision. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid rounding every intermediate value. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

39. Significant-figure check

Best used for: scientific quantities. The checking move is to start count at first non-zero digit. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid counting leading zeros. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

40. Decimal-place check

Best used for: fixed decimal precision. The checking move is to count positions after decimal point. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid confusing with significant figures. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

41. Exactness check

Best used for: fractions, surds and π. The checking move is to verify no unnecessary approximation was introduced. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid giving a decimal when exact form requested. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

42. Domain check

Best used for: functions and algebra. The checking move is to test denominator, root and log restrictions. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid forgetting restrictions after simplification. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

43. Interval check

Best used for: trigonometric and inequality solutions. The checking move is to compare every candidate with requested interval. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid reporting principal value only. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

44. Extraneous-root check

Best used for: squared or transformed equations. The checking move is to substitute candidates into original. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid accepting all transformed roots. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

45. Missing-root check

Best used for: factorised or periodic equations. The checking move is to enumerate every permitted case. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid stopping after first solution. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

46. Inequality sign check

Best used for: inequality manipulation. The checking move is to inspect any multiplication/division by negative quantity. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid copying equality habits. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

47. Endpoint check

Best used for: inequalities and bounds. The checking move is to decide whether equality is permitted. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid using open/closed endpoints by habit. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

48. Function-output check

Best used for: functions. The checking move is to evaluate final input through original function. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid confusing input with output. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

49. Inverse-composition check

Best used for: inverse functions. The checking move is to test f(g(x)) and g(f(x)) where appropriate. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid reversing operations without domain check. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

50. Sequence transition check

Best used for: sequences. The checking move is to count changes rather than labels. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid using year/stage number directly as exponent. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

51. Recurrence check

Best used for: recursive sequences. The checking move is to verify initial term and transition rule. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid checking only later numerical pattern. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

52. Compound-growth check

Best used for: percent change. The checking move is to rebuild first one or two periods. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid using additive change accidentally. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

53. Financial-timeline check

Best used for: interest and payments. The checking move is to place each cash flow at correct time. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid combining values from different times. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

54. Calculator-mode check

Best used for: trigonometry and statistics. The checking move is to verify device mode before entry. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid blaming mathematics for mode mismatch. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

55. Bracket-entry check

Best used for: calculator expressions. The checking move is to compare typed grouping with written expression. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid trusting display without expression. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

56. Memory-register check

Best used for: calculator use. The checking move is to clear or verify stored values. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid carrying an old value into new work. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

57. Order-of-operations check

Best used for: arithmetic. The checking move is to rewrite grouping explicitly. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid reading left-to-right mechanically. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

58. Place-value check

Best used for: decimals and standard form. The checking move is to estimate exponent and magnitude. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid moving decimal without scale check. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

59. Fraction-decimal check

Best used for: exact rational values. The checking move is to convert one form to the other independently. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid rounding too early. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

60. Reverse-percentage check

Best used for: percentages. The checking move is to apply the stated percentage change forward. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid adding percentage of final value. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

61. Rate-definition check

Best used for: speed/density/unit rate. The checking move is to rebuild numerator and denominator from units. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid inverting rate. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

62. Average-speed check

Best used for: journeys. The checking move is to reconstruct total distance and total time. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid averaging listed speeds. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

63. Area-perimeter check

Best used for: geometry. The checking move is to state quantity type and unit. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid using same dimensions with wrong formula. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

64. Surface-volume check

Best used for: solids. The checking move is to track square versus cubic units. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid confusing covering with filling. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

65. Derivative check

Best used for: calculus. The checking move is to differentiate reconstructed antiderivative or inspect numerical slope. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid assuming symbolic manipulation correct. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

66. Integral check

Best used for: calculus. The checking move is to differentiate antiderivative and verify limits. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid forgetting constant or sign. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

67. Stationary-point check

Best used for: optimisation. The checking move is to classify and compare candidates/endpoints. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid calling f’=0 a maximum. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

68. Proof-scope check

Best used for: proof. The checking move is to match domain and quantifier of conclusion. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid proving examples only. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

69. Theorem-condition check

Best used for: proof/geometry. The checking move is to list hypotheses before applying theorem. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid using theorem by name alone. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

70. Counterexample check

Best used for: logic. The checking move is to verify example meets hypothesis and violates conclusion. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid choosing an irrelevant exception. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

71. Converse check

Best used for: logic. The checking move is to write implication arrows explicitly. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid reversing a theorem. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

72. Dimension check

Best used for: modelling. The checking move is to verify physical dimensions/units. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid combining unlike quantities. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

73. Assumption check

Best used for: modelling. The checking move is to state what must remain constant or negligible. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid treating model as reality without conditions. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

74. Reasonableness check

Best used for: applied answers. The checking move is to compare with context and order of magnitude. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid rejecting unfamiliar but valid answers. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

75. Answer-target check

Best used for: all questions. The checking move is to reread exactly what was requested. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid submitting an intermediate quantity. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

76. Command-word check

Best used for: explain/prove/calculate. The checking move is to verify response form matches command. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid giving calculation when explanation requested. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

77. Working-completeness check

Best used for: method-mark questions. The checking move is to ensure decisive method is visible. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid adding irrelevant lines instead. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

78. Dependency check

Best used for: long solutions. The checking move is to identify which later answers depend on corrected line. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid repairing one value but not downstream work. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

79. Copying check

Best used for: multi-line algebra. The checking move is to compare transferred numbers/signs with previous line. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid recalculating correct arithmetic around a copied error. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

80. Transcription check

Best used for: tables/graphs. The checking move is to verify values copied from source. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid assuming a graphing error is algebra. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

81. Final-unit check

Best used for: applied questions. The checking move is to attach correct unit and power. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid using a unit from an intermediate quantity. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

82. Final-form check

Best used for: exam instructions. The checking move is to match exact/decimal/fraction/vector/interval form. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid stopping at equivalent but disallowed presentation. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

83. Completeness check

Best used for: multi-part and multi-solution tasks. The checking move is to count required outputs against prompt. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid answering only first subtask. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

84. Independent-route check

Best used for: high-risk answers. The checking move is to use a genuinely different method where efficient. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid performing cosmetic variation of same route. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

85. Boundary-case check

Best used for: general formulas. The checking move is to test zero, one, endpoints or sign changes. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid testing only convenient middle values. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

86. Special-case check

Best used for: general reasoning. The checking move is to use a known simple case to challenge formula. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid assuming one passing case proves formula. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

87. Reverse-story check

Best used for: word problems. The checking move is to start from answer and reconstruct story stages. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid redoing forward arithmetic. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

88. Constraint-led check

Best used for: optimisation/capacity. The checking move is to test all active constraints. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid checking objective but not feasibility. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

89. Precision-policy check

Best used for: calculator work. The checking move is to preserve sufficient intermediate digits. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid rounding repeatedly. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

90. Error-location check

Best used for: all work. The checking move is to find first invalid transition. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid rewriting entire solution unnecessarily. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

91. Repair-regression check

Best used for: corrected work. The checking move is to rerun a fresh check after correction. This creates a second source of evidence rather than merely asking the original working to certify itself.

What it can catch. This check is especially useful when the original route could hide a modelling, arithmetic, sign, range, condition or interpretation failure. It is not universal. A passing check establishes only what that check actually tests.

Failure mode. Avoid assuming correction cannot create a new error. A checking ritual becomes weak when it reproduces the same assumption that generated the answer. The strongest check is often short precisely because it attacks one vulnerable transition directly.

Practice drill. Create one correct answer and one plausible wrong answer that this check can distinguish. Apply the check without looking at the original solution. Then explain what disagreement would send you back to the working and where you would restart the repair.

Part III. Thirty original verification laboratories

Laboratory 1. linear equation

Task. Solve 5x−7=28. Proposed answer: x=7.

Independent check. Substitute:35−7=28.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 2. quadratic

Task. Solve x²−7x+12=0. Proposed answer: x=3 or4.

Independent check. Substitute both into original; both give zero.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 3. square-root equation

Task. Solve √(x+6)=x. Proposed answer: candidate x=3 after rejecting −2.

Independent check. Original substitution rejects −2.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 4. reverse percentage

Task. After 20% discount price is72. Proposed answer: original90.

Independent check. 20% of90=18;90−18=72.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 5. ratio total

Task. Ratio3:5 total64. Proposed answer: 24 and40.

Independent check. 24+40=64 and24:40=3:5.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 6. ratio change

Task. 24 red,40 blue; add blue to ratio1:2. Proposed answer: add8 blue.

Independent check. Final 24:48=1:2; red unchanged.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 7. average speed

Task. 60 km at30 then60 at60. Proposed answer: 40 km/h.

Independent check. 40×3 total hours=120 km.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 8. capacity

Task. 55 people,12 per vehicle. Proposed answer: 5 vehicles.

Independent check. 4 hold48 insufficient;5 hold60 sufficient.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 9. budget

Task. 35+8n≤150. Proposed answer: max14.

Independent check. 14 costs147;15 costs155.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 10. Pythagoras

Task. legs9,12. Proposed answer: hypotenuse15.

Independent check. 9²+12²=225=15².

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 11. triangle angles

Task. angles44,68,68. Proposed answer: valid triangle.

Independent check. Sum=180 and equal base angles match condition.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 12. line equation

Task. through(2,5),(6,13). Proposed answer: y=2x+1.

Independent check. Both points satisfy equation.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 13. probability

Task. 4 blue3 red, two without replacement, same colour. Proposed answer: 3/7.

Independent check. Complement different colours=4/7; sum1.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 14. combined mean

Task. 8 mean6,12 mean11. Proposed answer: mean9.

Independent check. 20×9=180 equals48+132.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 15. bounds

Task. 12.4 nearest0.1. Proposed answer: 12.35≤x<12.45.

Independent check. Test12.35 rounds12.4;12.45 rounds12.5 under standard rule.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 16. inequality

Task. (x−1)(x−4)≤0. Proposed answer: 1≤x≤4.

Independent check. Test x=2 works; x=0 and5 fail; roots satisfy equality.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 17. exact form

Task. circle radius5 exact circumference. Proposed answer: 10π.

Independent check. C=2πr gives10π; decimal not needed.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 18. trig angle

Task. sinθ=.5,0≤θ<360. Proposed answer: 30°,150°.

Independent check. Both have sine .5; interval contains no further cycle.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 19. compound growth

Task. 200 grows10% 3 stages. Proposed answer: 266.2.

Independent check. Reverse one stage successively by ÷1.1 returns200.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 20. sequence

Task. arithmetic first5 difference3, n=10. Proposed answer: 32.

Independent check. a_n=5+9×3=32; step backward nine differences returns5.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 21. function inverse

Task. f(x)=3x+7, output31. Proposed answer: input8.

Independent check. f(8)=31.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 22. log equation

Task. 2^x=10. Proposed answer: x≈3.3219.

Independent check. 2^x numerically returns10.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 23. area

Task. rectangle12×7. Proposed answer: 84 square units.

Independent check. Alternative decomposition 12×(5+2)=60+24=84.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 24. volume

Task. cuboid3×4×5. Proposed answer: 60 cubic units.

Independent check. Layer view: area12 repeated5 times=60.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 25. similar volume

Task. length ratio2:3, small volume80. Proposed answer: large270.

Independent check. 270/80=27/8=(3/2)^3.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 26. derivative

Task. f=x³−4x. Proposed answer: f’=3x²−4.

Independent check. Difference quotient/numerical slopes at test points can challenge sign/scale.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 27. integral

Task. ∫0^2(3x²+2)dx. Proposed answer: 12.

Independent check. Differentiate x³+2x to recover integrand; evaluate limits again.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 28. stationary max

Task. A=100−(x−10)². Proposed answer: max100 at10.

Independent check. Square is nonnegative, so expression cannot exceed100; equality at10.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 29. proof parity

Task. odd+odd even. Proposed answer: 2(a+b+1).

Independent check. Final form is twice an integer.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Laboratory 30. word problem

Task. 360 collected,135 spent, remainder/9. Proposed answer: 25 each.

Independent check. 9×25+135=360 reconstructs story.

Why this check is useful. It approaches the answer through a relationship that is at least partly independent of the original execution. If the two routes disagree, do not average them or choose the nicer-looking result. Return to the first point where their assumptions or calculations diverge.

Fault injection. Deliberately alter one digit, sign, condition or operation in the proposed solution and see whether the check detects it. Then invent an error the check would not detect. This teaches the limits of verification as well as its power.

Part IV. A 20-day checking programme

Day 1. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 2. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 3. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 4. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 5. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 6. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 7. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 8. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 9. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 10. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 11. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 12. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 13. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 14. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 15. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 16. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 17. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 18. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 19. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Day 20. Train one check until it becomes available under pressure

Select five solved questions from material already learned. Before looking at the solutions, name the most likely failure for each: model, sign, arithmetic, unit, domain, interval, rounding, integer interpretation or completeness. Choose a check designed for that failure rather than applying the same ritual to every question.

Inject one deliberate error into two solutions and test whether the chosen check catches it. If not, record the check’s blind spot and choose a second check. This exercise builds a realistic understanding of verification: checks are detectors with coverage limits, not magical certificates of correctness.

Finish with a fresh mixed question under modest time pressure. Solve it, choose one targeted check and record whether the check changed anything. Over time, build a personal list of high-yield checks tied to recurring error patterns rather than a long universal checklist.

Part V. Frequently asked questions

Should I check every question?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

What is the fastest useful check?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

Why does rereading miss mistakes?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

Should I redo the whole calculation?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I check a calculator answer?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I check algebra?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I check fractions?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I check percentages?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I check geometry?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I check probability?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I check statistics?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I check a proof?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I check an inequality?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I check trigonometry?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I check exact answers?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I check rounding?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

What if two methods disagree?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

When should I change an answer?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I stop changing correct answers from doubt?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

How do I build a personal checking routine?

Choose the check according to the risk. A short substitution may be enough for an equation; a capacity problem needs a neighbouring-integer test; a calculator result benefits from estimation; a proof needs a scope and theorem-condition audit. Checking becomes efficient when it is selective and diagnostic.

During practice, record what the check can and cannot detect. Then create one plausible wrong answer and see whether your chosen check rejects it. This converts “be careful” into a concrete skill that can be retrieved during an examination.

A creative-writing lens: continuity checking

Creative writers perform a related but different form of verification when they check continuity. If a character leaves a key on a desk, the key cannot open a later door unless the story explains how it returned. Mathematics is stricter because relationships are formal, but the editing habit is similar: trace consequences back to their source and challenge contradictions rather than merely rereading familiar sentences.

Use the eduKate ecosystem as a route

For prerequisite concepts, use the Mathematics Learning Hub and Additional Mathematics Hub. For detailed post-correction verification in advanced mathematics, use Additional Mathematics Post-Repair Regression Testing. For broader learning from errors, use How Mistakes Improve Learning.

Scope note and final answer

The appropriate amount of checking depends on the paper, available time and examination rules. This article’s tasks are original teaching material, not official marking instructions. Adrian, Jo, Ben, Aisha, Ryan, Mira, Clara and Ethan are fictional teaching characters.

The central habit is: do not ask your solution to prove itself. Identify the likely failure and attack it from another direction. A good check is small, independent and specific enough to disagree when something important has gone wrong.