Singapore School Mathematics Operating Manual · Chapter 17
Before a calculation can be correct, the objects inside it must be compatible.
You can add 3 metres to 5 metres. You cannot meaningfully add 3 metres to 5 square metres as though they were the same kind of quantity. You can divide distance by time to obtain speed. You cannot add a probability to a temperature. You can substitute a number into a function when it lies in the function’s domain. You cannot treat a point, a vector and a scalar as interchangeable simply because they are all written using numbers.
This chapter uses the phrase mathematical type checking as BTT teaching language. The idea is simple: before accepting a line of working, ask whether every operation is legal for the kind of object being used.
Computer programs use type systems to prevent incompatible operations. School Mathematics has an analogous discipline even when it is not named explicitly. Units, dimensions, domains, scalar-versus-vector distinctions, count-versus-measure distinctions, probability bounds, angle units and function input types all act like mathematical type information.
Quantity types · Units · Dimensions · Object types · Vectors and scalars · Probability and data · Practice · Worked answers
1. Numbers do not carry enough meaning by themselves
The numeral 5 can represent five pupils, 5 cm, 5 cm², 5 seconds, $5, probability 0.5 after scaling, a coordinate, an index or a pure number.
The numeral alone does not tell us what operations are appropriate.
If a rectangle is 5 cm long, “5” is a length magnitude. If its area is 5 cm², “5” belongs to a different physical dimension. If five pupils are present, the quantity is discrete and whole-number valued.
A strong solver therefore reads every number with its type attached.
Instead of seeing only 12, see “12 metres”, “12 pupils”, “12%”, “12 degrees”, “12 dollars” or “term 12”.
2. Addition requires compatible quantity types
3 m + 5 m = 8 m is meaningful.
3 m + 500 cm is also meaningful after unit conversion because both represent length.
3 m + 5 m² is not an ordinary physical addition because length and area are different quantity types.
Likewise, 20 seconds + 3 metres is not a sensible direct sum.
This is one of the cheapest error checks in applied Mathematics: before adding, ask whether the terms represent the same kind of quantity.
3. Subtraction has the same compatibility requirement
If a tank contains 50 L and 12 L are removed, 38 L remain.
If a journey lasts 50 minutes and one stage lasts 12 km, the subtraction 50 − 12 has no direct quantity meaning unless the problem has first converted the objects into a common type.
Subtraction compares like with like.
A numerical answer can look plausible while the operation that created it is type-invalid.
4. Unit conversion changes representation, not quantity type
2.4 m = 240 cm.
The numeral changes because the unit changes, but both values represent the same physical length.
A valid conversion preserves the underlying quantity.
This is why conversion factors such as 100 cm / 1 m can be treated as equal to one. Multiplying by the factor changes the unit representation while preserving the length.
The existing Units, Scale and Measurement guide owns detailed school-level unit work. This chapter focuses on the operating rule underneath it: unit compatibility acts as a type check.
5. Area units square the length conversion
1 m = 100 cm, but 1 m² is not 100 cm².
Because area has two length dimensions, 1 m² = (100 cm)² = 10,000 cm².
Similarly, 1 cm² = 100 mm².
A learner who applies a linear conversion factor to an area has ignored the dimension of the object.
The same principle extends to volume: 1 m³ = 1,000,000 cm³.
6. Dimensional analysis checks whether a formula can even be plausible
Suppose a proposed speed formula is speed = distance × time.
The right-hand side has units such as km·h, not km/h.
The dimensions do not match speed.
By contrast, distance ÷ time produces km/h, which matches the target type.
Dimensional consistency cannot prove a formula is correct, but inconsistency can prove it is wrong.
7. Both sides of a physical equation should have compatible dimensions
Consider distance = speed × time.
If speed is measured in m/s and time in s, the right side has units m, matching distance.
If a student writes distance = speed + time, the units become m/s + s, an invalid addition.
The equals sign claims the two sides represent the same type and same quantity.
Dimension checking therefore supports both equation construction and verification.
8. Pure numbers can appear inside dimensional formulas
The factor 1/2 in triangle area A = 1/2 bh is dimensionless.
b and h are both lengths, so bh has area dimension.
Multiplying by 1/2 changes magnitude but not dimension.
Constants such as 2, π and percentage multipliers may be dimensionless, though other constants in physics or applied models can carry units.
The learner should distinguish a dimensionless multiplier from a quantity with its own physical type.
9. Scalars, points, vectors, equations and sets are different mathematical objects
A scalar is a single magnitude such as 5 or −2.
A coordinate point such as (3, 4) represents location.
A vector such as (3, 4) can represent displacement or direction-and-magnitude information.
The same ordered pair notation can therefore encode different object types depending on context.
Operations should follow the object, not just the visual form.
10. A point is not automatically a vector
The point A(2, 5) locates A relative to the chosen coordinate frame.
The vector OA may have components (2, 5) if O is the origin.
The vector AB requires subtraction of coordinates: B − A.
Treating every coordinate pair as the same type can produce incorrect geometry.
The notation may look similar, but the roles differ.
11. Vector addition is legal when the vector spaces match
If a = (2, 3) and b = (4, −1), then a + b = (6, 2).
Both are two-dimensional vectors expressed in the same coordinate basis.
Adding a two-dimensional vector to a three-dimensional vector is not defined in the ordinary way because their component structures differ.
Likewise, adding a vector directly to a scalar is generally type-invalid unless a special definition has been introduced.
12. Scalar multiplication changes magnitude, not object class
3(2, −1) = (6, −3).
The scalar 3 multiplies every vector component.
The result remains a vector.
By contrast, the dot product of two vectors produces a scalar. The cross product in three-dimensional vector geometry produces a vector.
Different operations can deliberately change type.
13. A legal operation can have a different output type from its inputs
Length × length gives area.
Area × length gives volume.
Distance ÷ time gives speed.
Change in velocity ÷ time gives acceleration.
Two vectors under a dot product give a scalar.
A probability distribution can produce an expected value.
Type checking does not mean every operation preserves type. It means the input-output type relationship must be legitimate.
14. Function notation includes input-type restrictions
If f(x) = √(x − 2) over the reals, then the input must satisfy x ≥ 2.
The expression f(1) is not a real output under this definition.
If g(n) is defined only for positive integers n, then substituting n = 2.5 may be outside its intended domain even if the algebraic formula can be evaluated numerically.
The input domain is part of the function’s type contract.
15. Inverse functions require an appropriate domain and codomain relationship
For f(x) = x² over all real numbers, f is not one-to-one because f(2) = f(−2).
Therefore a single-valued inverse function on all real inputs is not available without restricting the domain.
Restricting to x ≥ 0 gives an inverse f⁻¹(y) = √y for y ≥ 0.
The inverse question is partly a type question: does each allowed output correspond to exactly one allowed input?
16. Counts and measurements have different admissible output types
95 ÷ 4 = 23.75.
If 95 metres are divided into four equal lengths, 23.75 m is meaningful.
If 95 pupils are divided into four groups, 23.75 pupils is not an admissible group size.
The arithmetic is identical; the output type is not.
This links directly to Discrete and Continuous Models.
17. A rate is not the same type as a total
60 km/h is a rate.
120 km is a distance.
2 h is a time.
The equation 60 km/h × 2 h = 120 km is dimensionally coherent because the hour units cancel.
Adding 60 km/h + 2 h would mix different types.
Many word-problem errors disappear once rate, total and duration are kept distinct.
18. Percentages are dimensionless ratios—but their reference matters
25% can be represented as 0.25.
It is dimensionless in the sense that it represents a ratio.
But “25% of what?” remains essential.
25% of $80 is $20, while 25% of 80 kg is 20 kg.
The percentage multiplier is dimensionless, but the result inherits the type of the base quantity.
19. Probability has its own admissible range
A probability must lie between 0 and 1 inclusive.
If a calculation produces 1.3 or −0.2, the output fails the probability type contract.
This does not identify the exact earlier error, but it proves the final value is inadmissible.
The same idea applies to percentages where context may restrict values, lengths that must be positive, and counts that must be whole numbers.
20. Expected value is not the same type as an individual outcome
A fair die has outcomes 1, 2, 3, 4, 5 and 6.
Its expected value is 3.5.
No individual roll can equal 3.5.
The expected value is a summary statistic, not an admissible sample outcome.
This is a type distinction between distribution-level quantities and event-level outcomes.
21. A mean can have a type different from a raw observation
If family sizes are whole-number counts, an average family size can be 2.4.
The mean is a real-valued summary even though each observed family size is an integer.
Rejecting a non-integer mean because “you cannot have 0.4 of a child” confuses statistic type with observation type.
The statistic describes the data set; it is not claiming to be one of the observations.
22. Angles need a unit convention
sin 30° = 1/2.
sin 30 when 30 is interpreted in radians is a completely different value.
Calculator degree/radian mode is therefore part of the input type.
An angle value without a clear unit can be ambiguous.
This is why a calculator-state error can produce a numerically precise but mathematically mistyped result.
23. Logarithms also depend on base
log₂8 = 3, while log₁₀8 and ln 8 have different values.
The base is part of the logarithm’s definition.
A calculator key labelled log often means base 10, while ln means base e.
Using the wrong base is not merely an arithmetic mistake. It changes the function being evaluated.
24. Coordinates depend on a reference frame
The point (3, 4) has meaning relative to chosen axes and origin.
Changing the origin can change the coordinate numbers without changing the physical point.
This is another example of separating representation from object.
Coordinate values are typed by their frame.
25. A matrix has shape information
A 2 × 3 matrix cannot be added to a 3 × 2 matrix because addition requires matching dimensions.
Matrix multiplication has a different rule: an m × n matrix can multiply an n × p matrix, producing an m × p matrix.
The inner dimensions must match.
Even where matrix work lies beyond a learner’s current syllabus, the principle is useful: legal operations depend on object structure.
26. Algebraic expressions also have domains as types
(x² − 4)/(x − 2) simplifies to x + 2 for x ≠ 2.
The simplified expression looks like an ordinary polynomial, but the original rational expression has a hole at x = 2.
Removing the visible denominator does not erase the original domain information.
The type contract must be carried through the transformation.
27. Equations should not equate incompatible object types
Writing “area = 12 cm” is type-inconsistent because area requires square-length units.
Writing “speed = 50 km” confuses a rate with a distance.
Writing “probability = 70% kg” combines incompatible notation.
The equals sign is a strong claim: both sides should represent the same kind of mathematical object.
28. Type checking can locate errors before numerical work
Suppose a proposed formula for the area of a circle is A = 2πr.
The right side has length dimension, not area dimension.
The formula is therefore dimensionally inconsistent before any value of r is substituted.
By contrast A = πr² has area dimension.
This kind of pre-calculation check is highly efficient.
29. Type checking can compare two candidate formulas
Suppose a student is unsure whether cylinder volume is πr²h or πrh.
The first has dimensions length² × length = length³, appropriate for volume.
The second has dimensions length², appropriate for area rather than volume.
Dimensional analysis identifies which candidate could possibly be correct.
It does not derive the formula from first principles, but it eliminates the mistyped one.
30. Mathematical object types matter in proofs too
A statement about integers cannot automatically be extended to all real numbers.
A proof using parity applies to integers because even and odd classification is integer structure.
Trying to call 2.5 “odd” is a type error: parity is not defined in the ordinary way for non-integers.
The domain of discourse determines which predicates are legal.
31. Set membership and equality are different relationships
3 ∈ {1, 2, 3} is a membership statement.
{3} ⊆ {1, 2, 3} is a subset statement.
3 = {3} is false because a number and a one-element set are different object types.
Notational discipline protects these distinctions.
32. A graph point and a function value are related but not identical
If f(2) = 5, then (2, 5) is a point on the graph y = f(x).
The scalar output 5 is not itself the point (2, 5).
The equation links an input-output relation; the ordered pair packages both coordinates.
Keeping these object types distinct helps in coordinate and function questions.
33. A type error can survive correct arithmetic
Suppose a learner computes 3 m × 4 m = 12 m.
The multiplication 3 × 4 is correct, but the unit type is wrong.
The correct result is 12 m².
Arithmetic accuracy is therefore not sufficient for mathematical correctness.
34. A practical type-checking routine
Before calculation, label the key quantities: count, length, area, volume, time, rate, probability, angle, coordinate, vector, scalar, set or function input.
During calculation, ask whether each operation accepts those types.
After calculation, ask whether the output type matches the question.
Finally, check whether units, domain restrictions and discrete/continuous conditions agree with the interpretation.
This routine can be done mentally once the habit becomes fluent.
35. Independent practice
1. Which of these sums are type-valid without further interpretation: 3 m + 4 m; 3 m + 4 m²; 3 min + 4 min?
2. Convert 2.5 m² to cm².
3. A proposed speed formula is v = dt. Use dimensions to test it.
4. A rectangle is 8 cm by 5 cm. State the type and unit of its area.
5. Why is 23.75 a valid result for 95 m ÷ 4 but not for an equal division of 95 pupils into four whole groups?
6. If f(x) = √(x − 3), is f(2) a real-valued function output?
7. A fair die has expected value 3.5. Why is 3.5 not a possible individual roll?
8. A calculator returns sin(30) ≈ −0.988. What type/state issue should be checked first if the intended angle is 30°?
9. Explain why area = 2πr is dimensionally suspicious.
10. Add vectors (2, 3) and (−1, 4).
11. Explain why a number and a set containing that number are not the same object type.
12. A student writes probability = 1.4. What immediate type check fails?
36. Worked answers
1. 3 m + 4 m and 3 min + 4 min are valid additions. 3 m + 4 m² mixes length and area and is not an ordinary compatible sum.
2. 1 m² = 10,000 cm², so 2.5 m² = 25,000 cm².
3. If d has length dimension and t has time dimension, dt has length×time, not length/time. The formula cannot represent speed.
4. Area is a two-dimensional length quantity. A = 8 × 5 = 40 cm².
5. Length is continuous in the idealised model, so fractional metres are allowed. A pupil count is discrete and whole-number valued.
6. No. x − 3 must be non-negative for a real square-root output. At x = 2 the radicand is −1.
7. Expected value is a distribution summary. The die’s individual outcome type is one of the integers 1 to 6.
8. Check whether the calculator is in radian mode instead of degree mode.
9. 2πr has length dimension, while area requires length squared. The type mismatch proves the formula cannot be an area formula.
10. (2, 3) + (−1, 4) = (1, 7).
11. 3 is a scalar number; {3} is a set whose only element is 3. Membership and equality are different relationships.
12. Probabilities must lie in [0, 1]. The output is inadmissible before any deeper interpretation.
37. Continue through the School Mathematics manual
Use Exhaustive Solutions, Completeness and Coverage when the main risk is missing a legal case, Iteration, Convergence and Stopping when an approximate numerical process repeatedly changes a value, and Construction, Verification and Impossibility when the task requires a witness or refutation.
Return to the BTT Mathematics Hub for Batch 05.