Singapore School Mathematics Operating Manual · Chapter 3
A structured mathematics question is not necessarily a row of separate exercises. One part may establish a quantity, another may change its representation, and a later part may ask for a conclusion that depends on both. A student can understand each technique and still lose the question at the point where one result becomes the next input.
This chapter concerns that handoff. What may be carried forward? What does “hence” ask a learner to do? When can a later part be attempted without completing an earlier one? Is a printed result a fact to use, a statement still to be proved, or both in different parts of the task? How much precision should survive the move from one answer space to the next?
The worked questions are original teaching sets. “Result handoff” and the status labels used below are BTT teaching language, not official assessment categories. The examples range from whole-number problems to coordinate geometry and calculus. Use the later material only after its mathematical prerequisites have been taught.
Read the dependencies · Understand hence · Study worked sets · Try linked practice · Check complete answers
1. Read a question as a set of dependencies
Suppose part (a) asks for a circle’s radius, part (b) asks for its area, and part (c) asks for the cost of covering that area at a stated price per square metre. The dependency is radius to area to cost. An error in the radius can travel into both later calculations unless the student notices it.
Another question may ask for the gradient and midpoint of a line segment in separate parts, then ask for its perpendicular bisector. The gradient and midpoint are sibling results: neither necessarily depends on the other, but the later line equation needs both.
The printed order does not prove that every part depends on the immediately preceding one. A final part may return to the original data and ask a different question. Reading every subpart as a single unbroken chain can trap a student unnecessarily.
Before calculation, identify the target of each part and the information it needs. For a substantial question, a short note such as “(c) needs midpoint from (a) and gradient from (b)” can make the structure visible. This is a planning aid, not a requirement to draw an elaborate flowchart in an examination.
The existing BTT guide to accuracy in long multi-step questions addresses memory, copying and execution across long solutions. This chapter has a narrower job: deciding which results and instructions legitimately connect one question part to another.
2. A result needs a name, a meaning and a status
The number 12 written beside part (a) is not enough information for a safe handoff. Is it a radius, a diameter, a total, a percentage, a coordinate or a probability denominator? A useful result carries its identity: “radius = 12 cm” or “number of children = 12”.
It also needs a status. A given result is supplied by the question. An established result has been justified in the working. A target is a statement the learner has been asked to show. A provisional result is an unverified answer being explored. A reported result is the version rounded or formatted for a particular answer space.
These labels need not all be written. Their purpose is to prevent a subtle mistake: treating everything visible on the page as equally trustworthy. A printed “show that” expression and a guessed numerical answer may look equally available, but their roles in the argument are different.
For example, if the task asks to show that a width is 7 cm, using 7 cm at the start of that same proof would assume the conclusion. In a later part, using the stated width to explore an area may be a reasonable conditional continuation, depending on the directions. It still does not retrospectively complete the missing proof.
Good handoffs preserve meaning and distinguish what is known from what is being assumed. This is mathematical bookkeeping, not decorative presentation.
3. “Hence” asks for a connection, not a restart
In the original practice questions in this chapter, “hence” means that the next result should follow using what has just been established. The task is not only to obtain the next answer but to recognise the useful relationship between the parts.
Suppose part (a) gives x² − 6x + 5 = (x − 3)² − 4. Part (b) says, “Hence state the minimum value of x² − 6x + 5 for real x.” Since a square is non-negative, the minimum is −4, attained when x = 3. The completed-square form is doing the work.
Differentiation could also find the minimum, but it would bypass the connection deliberately established in part (a). In a practice task designed to use completing the square, that bypass misses the instructional purpose. In an actual assessment, the exact wording and applicable marking arrangements govern method requirements.
“Hence” does not mean that any previous number should be inserted into any available formula. The earlier result must supply a logically relevant input. If a preceding part found a gradient, a later area calculation does not automatically need that gradient merely because it was found most recently.
Before moving on, finish a short sentence: “This result now lets me…” The answer might be “identify the minimum”, “replace this expression”, “find the missing count”, or “write a line through the midpoint”. A meaningful completion reveals the handoff.
4. “Hence or otherwise” leaves room for another valid route
When an instruction says “hence or otherwise”, it explicitly allows an alternative to the route suggested by the earlier result, subject to the rest of the question’s conditions. The alternative still has to establish the requested conclusion.
For example, after factorising x² − 7x + 12, a question might ask, “Hence or otherwise solve x² − 7x + 12 = 0.” Using the factorisation (x − 3)(x − 4) gives roots 3 and 4. Applying the quadratic formula also gives those roots and is an alternative valid derivation.
The wording does not mean that justification is unnecessary. It does not authorise use of a prohibited calculator method or permission to ignore a positive-value restriction. Freedom of method sits inside the task’s other boundaries.
A student who already has a reliable earlier result will usually benefit from using it. Restarting can add work and create a new opportunity for error. But when an earlier method has become tangled, recognising an expressly allowed alternative can preserve progress.
During learning, compare the routes after solving. Which information did each method use? Which steps were shared? Did either route require assumptions that the other did not? This comparison teaches method choice without turning “hence” into a word to memorise mechanically.
5. Primary worked set: a count becomes a revenue calculation
At an invented school event, there are 84 visitors. Every visitor is either an adult or a child, and there are twice as many adults as children. A child ticket costs $9 and an adult ticket costs $15. Assume all visitors buy exactly one ticket at the stated price.
Part (a): Find the number of children. One equal group represents the children and two equal groups represent the adults. Three equal groups contain 84 visitors, so one group contains 28. There are 28 children.
Part (b): Hence find the number of adults. Using the result from (a), twice 28 is 56. Check the original total: 28 + 56 = 84. The answer is 56 adults, not 56 tickets of unspecified type.
Part (c): Find the total ticket revenue. Child revenue is 28 × 9 = $252. Adult revenue is 56 × 15 = $840. The combined revenue is $1,092. This part needs both counts and both prices, not simply the latest number written down.
Part (d): The organiser sets aside one quarter of the revenue for materials. Find the remaining amount. Three quarters remains, so the result is (3/4) × 1,092 = $819. The input here is revenue, not the visitor count or the adult revenue alone.
Now inspect the handoffs. Part (a) establishes a count. Part (b) produces a second count. Part (c) changes counts into money using two different prices. Part (d) changes the money allocation. Each transition needs a named quantity, not a bare result.
A helpful repair is to ask the pupil to label only the intermediate answers. “Children 28; adults 56; total revenue $1,092.” These three labels can prevent more confusion than a page of unexplained arithmetic. They also make it clear where to return if the final allocation appears implausible.
6. A later part can branch away from the main chain
Continue the event example, but add another question: “The hall has 100 seats. How many seats are unoccupied?” This part uses the original total of 84 visitors, so the answer is 16 seats. It does not require the ticket prices, the adult count or the revenue.
A student who failed to calculate the revenue can still answer the seat question. An earlier unresolved part has not removed access to the original facts. The dependency structure, not the emotional feeling of being stuck, determines whether progress remains possible.
Now change the seat question to “How many additional children could attend if the existing adults all remain and the total attendance must not exceed 100?” The available places are still 16, but the conclusion now counts additional children under a specific condition. If the problem also required the adult-to-child ratio to remain unchanged, that would introduce another constraint and change the task.
Read every added condition. A later part may retain an earlier model, modify it, or introduce a new scenario. Words such as “instead”, “now”, “a different group”, and “under the same conditions” tell the reader which assumptions travel forward.
This is why the best recovery question is not “Can I finish the whole problem?” It is “Which information does this particular part require, and do I have it?” That question is smaller and mathematically answerable.
7. Coordinate geometry: two sibling results feed one line
Let A be (1, 2) and B be (7, 10). The task asks for the midpoint of AB, its gradient, the perpendicular bisector, and the point where that bisector meets the y-axis. These are original coordinate choices, not an extracted examination question.
Part (a): Midpoint. Average the corresponding coordinates: ((1 + 7)/2, (2 + 10)/2) = (4, 6). Label this point M. The answer is an ordered pair, not the two separate differences between coordinates.
Part (b): Gradient of AB. The change in y is 8 and the change in x is 6, so the gradient is 8/6 = 4/3. Part (b) can be completed directly from A and B even if the midpoint calculation is unfinished.
Part (c): Perpendicular bisector. A perpendicular line has gradient −3/4 here, and the bisector must pass through M(4, 6). Therefore y − 6 = (−3/4)(x − 4). Rearrangement gives 3x + 4y = 36.
Part (d): Intersection with the y-axis. Set x = 0 in the line equation. Then 4y = 36, so the point is (0, 9). The final answer is a coordinate point. Writing only 9 supplies the intercept value but not the requested ordered pair.
Part (c) needs both earlier results. Using only the midpoint gives infinitely many possible lines through M. Using only the perpendicular gradient gives infinitely many parallel lines. The pair of conditions is what determines the bisector.
The output of (c) can also be checked against the inputs. Substituting M gives 3(4) + 4(6) = 36. Multiplying the two gradients gives (4/3)(−3/4) = −1. These checks test the two different properties required of the line rather than merely repeating the rearrangement.
8. A printed result can support continuation without supplying its proof
Suppose a question says, “Show that the perpendicular bisector has equation 3x + 4y = 36,” then asks where this line meets the y-axis. A student who cannot complete the derivation may still understand the later substitution x = 0 and obtain (0, 9).
There are two separate pieces of mathematical work. One establishes the equation from the geometric data. The other derives an intercept from the equation. Completing the second does not complete the first, but leaving the first incomplete does not make the second calculation meaningless.
During practice, a learner can mark the continuation honestly: “Using the stated equation…” This avoids pretending that an unproved step has been proved. Whether and how any examination credit is available depends on the actual paper and marking scheme; this guide makes no follow-through or partial-credit guarantee.
Do not generalise this into permission to trust any self-generated answer. A printed target and an unsupported guess are not interchangeable. If a student guessed a line equation, the later intercept is only conditional on that guess. It should not be presented as independently established from the original geometry.
The useful principle is limited and practical: distinguish a missing derivation from a missing input, and distinguish mathematical continuation from a claim about marks.
9. Completed-square results can feed several different conclusions
Consider f(x) = x² − 6x + 5 for real x. First show that f(x) = (x − 3)² − 4. Expanding the right-hand expression gives x² − 6x + 9 − 4, which matches the original function. The equivalence is now established for every real x.
Hence find the minimum value. Since (x − 3)² is at least zero, f(x) is at least −4. Equality occurs at x = 3. The minimum value is −4; the minimum point is (3, −4). Read which of those objects the part requests.
Hence solve f(x) = 0. The completed-square form gives (x − 3)² = 4, so x − 3 = 2 or −2. The roots are 5 and 1. The minimum result itself is not needed; both the minimum and the roots branch from the completed-square identity.
Hence solve f(x) ≤ 5. The same identity gives (x − 3)² ≤ 9. Thus −3 ≤ x − 3 ≤ 3, and 0 ≤ x ≤ 6. This is an interval answer, not just the two boundary values.
These parts demonstrate a reusable representation. Once the square form is available, different features become visible. The earlier expression is more useful than a single rounded output because it preserves a relationship for all allowed inputs.
A result handoff is therefore not always numerical. It may transfer an identity, an inequality, a diagram property or a functional relationship. Students who carry forward only boxed numbers can miss the most valuable result in a structured question.
10. Definitions have a scope: the same letter may change jobs
Within one connected question, a variable normally retains the meaning assigned to it unless the wording changes that meaning. If x is the width of a rectangle in centimetres, a later expression involving x should preserve that definition. Substituting a value for the rectangle’s length because it is the most recent result is a meaning error.
Across separate questions, the same letter can be reused for a completely different object. The x in a coordinate problem need not equal the x found in a previous algebra problem. Carrying a numerical value across that boundary silently imports an assumption.
A later part can deliberately introduce a new model: “For a different rectangle, let x be…” In that case, the definition has changed. Read the new conditions before reusing an earlier formula. Similar lettering does not prove that the old physical situation remains in force.
Units are part of the definition. If t originally measures minutes and a later variable s measures seconds, a formula written for t cannot receive the raw value of s without conversion. The numbers may look compatible while their meanings are not.
A compact safeguard is to retain a short definition beside an important intermediate variable. This is especially useful when several quantities have similar names, such as distance and displacement, total cost and unit cost, or frequency and cumulative frequency.
11. Do not let a later result prove the earlier statement it depends on
Suppose part (a) asks a student to derive an area expression from a rectangle’s dimensions. Part (b) then uses that expression to find a maximum. Using the maximum found in part (b) as the sole reason why the area expression in part (a) is correct reverses the dependency and can create circular reasoning.
Working backwards while planning is not automatically circular. A student may inspect the desired expression, recognise a factorisation and then build a forward derivation from the given dimensions. What matters is the logical support in the finished argument.
For a simple algebraic illustration, consider the statement x² − 4 = (x − 2)(x + 2). Expanding the right side independently verifies the identity. But saying “the factorisation is correct because its roots are 2 and −2, and those roots are correct because of the factorisation” supplies no independent justification for the relationship as written.
Even roots alone do not determine a polynomial’s leading coefficient. Both x² − 4 and 5x² − 20 have those roots. A valid reconstruction needs the remaining information as well. This is where result handoffs meet data sufficiency.
When checking a proof, ask what each line depends on. If a line relies on a later conclusion that itself relies on that line, find a genuinely independent starting fact or a valid reversible equivalence argument.
12. Calculus: a derivative is not a point on the original curve
Let f(x) = x³ − 3x² + 4. Differentiate to obtain f′(x) = 3x² − 6x = 3x(x − 2). A later part asks for the stationary points. The derivative supplies their possible x-coordinates through f′(x) = 0, giving x = 0 and x = 2.
The y-coordinates must come from the original function, not the derivative. We obtain f(0) = 4 and f(2) = 8 − 12 + 4 = 0. The stationary points are (0, 4) and (2, 0). Substituting both x-values back into f′ would merely return zero gradients.
To classify them, inspect the sign of 3x(x − 2). It is positive for x < 0, negative for 0 < x < 2, and positive for x > 2. Therefore the curve changes from increasing to decreasing at x = 0, producing a local maximum, and from decreasing to increasing at x = 2, producing a local minimum.
Now a separate part asks for the tangent at x = 1. This branch needs f(1) = 2 and f′(1) = −3. The tangent is y − 2 = −3(x − 1), or y = −3x + 5. The stationary-point coordinates are not the inputs for this tangent.
The two functions have different jobs throughout the question. The original function supplies position on the curve; its derivative supplies gradient. The notation should preserve that distinction at each handoff rather than treating both formulas as interchangeable expressions in x.
13. Motion: one turning time separates distance from displacement
For an original one-dimensional motion model, let a particle have velocity v(t) = 6 − 2t metres per second for 0 ≤ t ≤ 5 seconds, with position s(0) = 0. Take the positive direction as fixed throughout. This is a mathematical teaching model, not a description of a measured journey.
Part (a): Find when the particle is momentarily at rest. Set v = 0, giving 6 − 2t = 0 and t = 3. Before this time the velocity is positive; after it the velocity is negative. The result therefore identifies a change of direction within the stated interval.
Part (b): Find the position function. Integrating velocity gives s(t) = 6t − t² + C. Since s(0) = 0, C = 0. The position is s(t) = 6t − t² metres. The initial condition fixes the integration constant; it cannot be discarded.
Part (c): Find displacement over the five seconds. Calculate s(5) − s(0) = 30 − 25 = 5 m. This is a signed change of position, not the total distance travelled.
Part (d): Hence find total distance travelled. Use the turning time from (a) and the position function from (b). At t = 3, s = 9 m. The particle travels 9 m forward, then 4 m back to position 5 m. Total distance is 9 + 4 = 13 m.
Part (e): Compare average velocity and average speed. Average velocity is displacement divided by elapsed time, 5/5 = 1 m/s. Average speed is distance divided by elapsed time, 13/5 = 2.6 m/s. The denominator is the same, but the numerators represent different quantities.
A learner can integrate correctly and still answer the distance part incorrectly by handing forward displacement without changing its meaning. The turning time is not an extra answer to forget; it tells the learner where the distance calculation must be split.
14. Preserve the working value when the displayed answer is rounded
Suppose part (a) establishes a positive length r = √65 cm and asks for its value correct to two decimal places. The reported answer is 8.06 cm. Part (b) asks for the exact area of a circle of radius r. The appropriate input is still √65, giving area 65π cm².
Squaring the displayed 8.06 gives 64.9636, not 65. This is not a new geometric fact. It is an artefact of replacing the exact result with a rounded report before the value had finished doing its work.
Keep an exact expression or the unrounded calculator value for continuation when the instructions allow it. The answer space may require a short decimal, while the internal working value remains more precise. These are two representations of related information with different jobs.
There is an important exception: if a later task explicitly instructs the learner to use a stated approximation, follow that instruction. Likewise, if the original data are already rounded measurements, symbolic manipulation cannot recover the missing input precision.
The existing BTT exact-form and decimal guide develops that representation choice. The companion chapter on rounded data and guaranteed conclusions handles uncertainty in the inputs themselves.
15. Recover from an uncertain earlier answer without disguising it
When an earlier calculation is doubtful, first separate the part’s inputs from its result. Are the original quantities still readable and usable? Is an equivalent result supplied in a later statement? Does another branch avoid the uncertain value? These questions can reveal valid ways forward.
If a later calculation is performed using a provisional value, keep that dependency visible. For learning purposes, “Using my radius from (a)…” allows a teacher to see whether the next method is understood even if the radius needs repair. It does not make the final answer independently correct.
Do not silently change an earlier answer to fit a later result. If a printed check reveals a disagreement, identify where the original derivation failed and repair that step. A solution assembled backwards to resemble the expected answers can conceal the very misunderstanding the practice was meant to expose.
During an assessment, question directions and permitted methods remain controlling. This chapter offers mathematical organisation, not a universal strategy for earning method marks. No claim is made that an incorrect carried-forward value will receive credit.
After practice, revisit the first unresolved dependency. A learner who can finish parts (c) and (d) once supplied with the result from (b) may need a targeted repair in (b), not a complete reteaching of every later technique.
16. Independent practice: three complete linked sets
Try each set before reading its solution. Beside each subpart, identify the earlier result or original fact it needs. The objective is to preserve the dependencies, not to write the greatest number of lines.
Set A: boxes and delivery cost
An invented school order contains 150 workbooks. Each box holds exactly 12 workbooks when full. Every workbook must be packed, and partly filled boxes are allowed. Delivery costs $4 per box plus a fixed charge of $7.
(a) Find the minimum number of boxes. (b) Hence find the number of empty workbook spaces across those boxes. (c) Find the delivery cost. (d) A budget of $63 is available for delivery. Find the money remaining. (e) A separate question asks how many full groups of 12 workbooks can be formed. Explain why its answer is not the same as part (a).
Set B: a quadratic and its representations
Let g(x) = x² − 8x + 12 for real x. (a) Show that g(x) = (x − 4)² − 4. (b) Hence state the minimum value and the input where it occurs. (c) Hence solve g(x) = 5. (d) Solve g(x) ≤ 5. (e) Find g(0), and explain whether this part needs the answers to (b), (c) or (d).
Set C: motion with a reversal
A particle has velocity v(t) = 8 − 2t m/s for 0 ≤ t ≤ 6 seconds and initial position s(0) = 3 m. (a) Find its rest time. (b) Find its position function. (c) Find displacement from t = 0 to t = 6. (d) Find total distance travelled. (e) Find average velocity and average speed over the six seconds.
17. Worked answers: preserve each handoff
Set A
(a) Thirteen boxes. Twelve boxes contain only 12 × 12 = 144 workbooks, leaving six unpacked. Thirteen boxes provide capacity for 156. The whole-number capacity decision determines the minimum. The quotient 12.5 is useful working, not an available number of boxes.
(b) Six empty spaces. Use the number of boxes from (a): 13 × 12 − 150 = 6. The capacity must be calculated for the selected thirteen-box arrangement. Subtracting 150 from the capacity of twelve boxes would describe a shortage instead.
(c) Fifty-nine dollars. The delivery charge is 4(13) + 7 = 59. This part depends on the box count, not on the empty-space count. It can be solved even if part (b) is unfinished. The fixed charge is added once, not once per box.
(d) Four dollars. Subtract the delivery cost from the budget: 63 − 59 = 4. The budget and cost are comparable money quantities. Subtracting the box count from the budget would mix incompatible meanings.
(e) Twelve full groups, with six workbooks left over. This target asks how many complete groups can be formed, not how many containers are needed to hold everything. The numerical division is shared, but the requested objects and whole-number decisions differ.
Set B
(a) Expand (x − 4)² − 4 to obtain x² − 8x + 16 − 4 = x² − 8x + 12. This establishes the identity. Alternatively, complete the square directly from the original expression. Simply copying the requested form would not establish its equivalence.
(b) Minimum −4 at x = 4. The squared term is non-negative and becomes zero at x = 4. A coordinate answer would be (4, −4), but the question asks separately for the minimum value and its input. Name both accurately.
(c) x = 1 or x = 7. Using the square form, g(x) = 5 becomes (x − 4)² = 9. Thus x − 4 = −3 or 3. The two roots are a complete real solution set. Part (b)’s minimum value alone would not supply the roots without the function relationship.
(d) 1 ≤ x ≤ 7. The inequality becomes (x − 4)² ≤ 9, so −3 ≤ x − 4 ≤ 3. The boundary values from (c) help identify the interval, but a sign or square argument is needed to decide which region satisfies the inequality. Listing only 1 and 7 would omit all the interior solutions.
(e) g(0) = 12. Direct substitution in the original function gives 12. The completed-square form also gives 16 − 4 = 12. This part does not require the minimum, the roots or the inequality solution. It is a separate branch from the function definition.
Set C
(a) t = 4 seconds. Solve 8 − 2t = 0. The velocity is positive before 4 and negative afterwards in the stated interval, so the particle reverses direction there. The rest time will be needed for distance, not merely listed and forgotten.
(b) s(t) = 8t − t² + 3 metres. Integration gives 8t − t² + C. The initial position s(0) = 3 supplies C = 3. Omitting that constant would produce the wrong position function, even though some displacement differences would accidentally remain unchanged.
(c) Displacement 12 metres. At t = 6, s = 48 − 36 + 3 = 15 m. Subtract the initial position 3 m to obtain 12 m. The final position 15 m is not the displacement because the starting position is not zero.
(d) Total distance 20 metres. At t = 4, s = 32 − 16 + 3 = 19 m. The forward distance is 19 − 3 = 16 m, followed by a return distance of 19 − 15 = 4 m. Add magnitudes: 16 + 4 = 20 m. This uses both the reversal time and the position function.
(e) Average velocity 2 m/s; average speed 10/3 m/s. Divide displacement 12 m and distance 20 m separately by the elapsed 6 seconds. The exact average speed is 10/3 m/s, approximately 3.33 m/s to three significant figures. A reporting instruction would determine the required final form.
18. A short handoff review after practice
Review the transitions, not only the final answers. For each dependent part, ask what arrived from earlier work, what it represented, whether it was established, and whether its precision was preserved. Then ask what new operation or condition the current part introduced.
For an incorrect motion distance, the crucial question may be whether the learner carried forward the rest time. For an incorrect line equation, it may be whether both midpoint and perpendicular gradient were used. For a revenue error, it may be whether each count remained attached to the correct ticket price.
Use a new linked set with the same dependency pattern but different numbers. Then change the pattern itself: add an independent branch or ask for a different target. This reveals whether the learner understands the structure or has memorised the order of a single example.
A well-managed solution does not have to be long. It has to make the important transfers trustworthy. The student should know not only what was found, but what that result now permits.
19. Continue through the BTT School Mathematics manual
Return to the School Mathematics Operating Manual at the BTT Mathematics Hub. Use Command Words, Answer Forms and Task Contracts for the requested output, Data Sufficiency, Unique Answers and Counterexamples for whether the available inputs determine it, and Rounded Data, Thresholds and Guaranteed Conclusions for the strength of a numerical conclusion.
For the wider examination-preparation route, use Mathematics Examination Craft. For unresolved topic prerequisites, enter the Mathematics Knowledge Warehouse rather than treating every handoff problem as a need for another complete paper.
20. Sources and examination boundaries
The linked sets and their solutions are original teaching material. The interpretations of instructions are explained for these tasks; an actual paper’s wording and rules take priority. There is no universal promise about partial credit, error carried forward, or the acceptance of a particular alternative method.
For cohort-specific reference, use the 2026 PSLE formats, 2027 SEC G3 Mathematics and Additional Mathematics syllabuses, and 2026 A-Level syllabuses. These are different cohort references, checked on 6 September 2026, not interchangeable specifications.
An earlier answer becomes useful when the learner knows exactly what it means, how it was established, and which later conclusion it can legitimately support.