PRIMARY 4 MATHEMATICS LEARNING GUIDE · BATCH 9 · GUIDE 35
Seeing a pattern is the beginning of mathematical reasoning, not the end. A learner notices that every multiple of four in a short list is even. That observation is useful. The next questions are more powerful: What exactly is repeating? Can we predict a new case? Does the pattern still hold when the numbers become larger? Can we explain why it happens? Can we find a counterexample to a rule that only looked true?
This guide develops a Primary 4 route from observation to generalisation. The examples use number sequences, factors and multiples, parity, fractions, decimals, geometry, area and data. The aim is not formal algebraic proof. It is to build the habits that later proof and algebra will depend on: identify what changes, identify what remains invariant, test a conjecture, search for edge cases and explain the structure beneath the examples.
The official curriculum reference is the MOE Primary Mathematics Syllabus, updated October 2025, whose mathematical-process framework includes reasoning, communicating, representing and generalising. This article’s examples and progression are independently written.
Series route: return to the Primary 4 Mathematics Learning Hub. For testing a rule at its edges, use Special Cases, Boundaries and Counterexamples.
Navigate: observe · predict · test · explain · different structures · practice · answers.
1. Describe what changes before guessing the next term
Sequence: 4,250; 4,350; 4,450; 4,550.
The useful observation is not merely “the next number is 4,650”. It is “each term increases by 100”. The rule explains the next term and every later term generated by the same process.
State both direction and amount of change. “Gets bigger” is incomplete because many rules make numbers bigger.
Now inspect a decreasing sequence: 7,800; 7,500; 7,200; 6,900. Each term decreases by300.
A precise observation gives the learner something that can be tested and transferred.
2. Separate visual appearance from numerical structure
Sequence: 9,850; 9,950; 10,050; 10,150.
The digits change dramatically when the sequence crosses 10,000, but the rule remains “add100”.
A learner who tracks only visual similarity may think the pattern has changed at the place-value boundary.
Ask what numerical difference remains constant.
Generalisation follows value, not superficial digit shape.
3. A rule supports predictions beyond the visible examples
If a sequence begins 12, 18, 24, 30 and increases by6, the next term is36.
The tenth term can be found by extending or reasoning about repeated additions.
Starting at12, nine steps of6 lead to 12 + 54 = 66.
At Primary 4, a learner does not need formal nth-term notation to reason systematically about a distant term.
The key is that the prediction is tied to a stated rule rather than an unsupported guess.
4. Multiplicative patterns behave differently from additive ones
Sequence A: 3, 6, 9, 12, 15 adds3.
Sequence B: 3, 6, 12, 24, 48 multiplies by2.
The first two terms are identical, but the later structure diverges.
Do not decide a rule from a single gap. Use several consecutive transitions.
When two different rules can fit the visible information, more terms or an explicit condition may be needed.
5. Test a conjecture with cases you did not use to create it
Conjecture: “Every multiple of4 is even.”
The observed cases4,8,12,16 support the idea, but we should test new cases such as28, 44 and100.
Each remains even.
This evidence increases confidence, but the strongest Primary 4 explanation comes from structure: a multiple of4 can be written as groups of4, and every group of4 contains two pairs. Therefore the total can be arranged in pairs.
The explanation tells us why the pattern should continue beyond the tested examples.
6. Look for a counterexample before calling a rule universal
Conjecture: “Every number ending in2 is a multiple of4.”
Test 12: yes. Test 32: yes. Test 42: no, because42÷4 leaves a remainder.
One counterexample is enough to reject the universal statement.
The better generalisation is that every number ending in2 is even, not necessarily a multiple of4.
Counterexamples refine patterns instead of merely destroying them.
7. Explain why even plus even remains even
Take two even numbers, such as8 and14. They can be arranged as four pairs and seven pairs.
Combining the collections gives eleven pairs, so the total22 is even.
Try larger cases:36+48=84; 102+200=302.
The examples support the rule. The pair-grouping explanation shows the structure that survives across all positive whole-number cases and also works for zero.
This is an early form of general reasoning without formal symbolic proof.
8. Odd plus odd produces even
An odd number can be seen as some complete pairs plus one extra.
Combine two odd numbers. Their two extras make another pair.
Example:7=3 pairs+1; 11=5 pairs+1. Together18=9 pairs.
The visual structure explains why the result is even.
Now compare odd+even. The even number contributes only complete pairs, while the odd number contributes one unpaired item, so the result remains odd.
9. Factors appear in pairs because multiplication can be reversed
For36, factor pairs are1×36,2×18,3×12,4×9 and6×6.
After6×6, reversing the factors would repeat earlier pairs.
This explains why a systematic factor search can stop once the pair order begins to reverse.
A square number such as36 has one repeated middle pair,6×6. A non-square number such as24 has no equal middle pair.
The pattern connects factor listing with geometric arrays and multiplication symmetry.
10. Common multiples create repeating meetings
One event repeats every4 seconds and another every6 seconds.
They meet at12,24,36,48…
The gap between successive meetings is12 because12 is the first positive common multiple of4 and6.
After every meeting, the two cycles restart from a shared point, so another12-second block reproduces the same relative timing.
This gives a structural explanation for the repeating pattern of simultaneous events.
11. Equivalent fractions show an invariant under repartitioning
1/2=2/4=3/6=4/8.
The number and size of the parts change, but the represented amount remains invariant.
Each time numerator and denominator are multiplied by the same non-zero whole number, every original part is subdivided into equal smaller parts and the selected region is subdivided in the same way.
The fraction’s appearance changes; its value does not.
This pattern becomes useful for common denominators and fraction comparison.
12. Adding the same number to numerator and denominator is a different pattern
Starting from1/2, adding1 to both gives2/3.
The value changes.
This contrast is valuable: “same operation on top and bottom” is not enough to preserve a fraction. Multiplying numerator and denominator by the same non-zero number preserves the ratio of selected parts to total parts; adding does not.
A generalisation should be based on the relationship, not only the surface similarity of two operations.
Compare several cases to see the difference.
13. Decimal trailing zeros reveal a place-value pattern
0.6=0.60=0.600.
Each added trailing zero creates an extra place containing zero smaller parts.
The value remains six tenths.
But moving the six changes value:0.06 is six hundredths.
Generalisation depends on where the zero is placed, not simply on the presence of a zero.
14. Rounding patterns are controlled by a midpoint
Consider numbers around3,450 when rounding to the nearest hundred.
3,449 rounds down to3,400;3,450 rounds up to3,500 under the usual positive halfway-up school convention;3,451 also rounds to3,500.
The midpoint separates the two regions.
The pattern is not “five always makes a number bigger” in every context. It is a rule for choosing between two neighbouring rounded values under a stated convention.
Understanding the boundary makes the generalisation more precise.
15. Area patterns depend on which dimension changes
Take rectangles of width4 cm and lengths3,4,5,6 cm.
The areas are12,16,20,24 cm². Each extra centimetre of length adds4 cm² because a new strip measuring1 by4 is added.
If both length and width grow, the area pattern changes. A3×3 square has area9;4×4 has16;5×5 has25. The increases are7 then9, not constant.
The surface “bigger rectangle” is not enough to predict one pattern. Which dimensions change matters.
Generalisation requires identifying the controlled variable.
16. Perimeter patterns can remain linear when area does not
For squares with side lengths3,4,5,6, the perimeters are12,16,20,24.
Each one-unit side increase adds4 units to perimeter because all four sides lengthen by1.
The areas are9,16,25,36, whose increases are not constant.
One family of shapes can therefore produce different numerical patterns depending on the measured attribute.
This is another reason to name whether we are studying boundary length or covered region.
17. Graphs can display patterns but do not explain their causes automatically
A line graph may rise by5 units each day across four days.
That repeated difference supports the description of a constant recorded increase across those intervals.
It does not prove that the same increase will continue forever.
It also does not explain why the increase occurred.
Prediction from a data pattern should distinguish observed regularity from guaranteed future behaviour.
18. Input-output tables can reveal a rule
| Input | Output |
|---|---|
| 1 | 5 |
| 2 | 8 |
| 3 | 11 |
| 4 | 14 |
The output increases by3 whenever the input increases by1.
One description is: start at5 and add3 for each next input.
A more compact later description might connect output to “three times the input, plus two”. Check: for input4,3×4+2=14.
At Primary 4, ordinary language and tables are sufficient to explore the relationship without requiring formal algebraic notation.
The key is to test the proposed rule against every supplied row and new inputs.
19. More than one rule can fit a very short pattern
Suppose only the numbers2,4 are shown. Many rules can produce them: add2; multiply by2 once; begin an alternating pattern; or follow a longer hidden rule.
With more terms2,4,6,8, an add2 rule becomes more plausible, but even then the continuation depends on the assumption that the displayed rule persists.
In school pattern questions, the intended rule is usually chosen to be simple and consistent with the supplied terms.
The learner should still avoid pretending that two terms logically force one unique infinite sequence without any assumption.
This is a useful boundary between pattern recognition and mathematical certainty.
20. Generalisation can compress repeated calculation
Imagine a row of connected squares made from matchsticks. One square uses4 sticks. Adding a second square sharing one side needs only3 more, giving7. A third needs3 more, giving10.
The pattern is4,7,10,13…
Instead of redrawing every case, we can reason that each extra square adds3 sticks because one side is shared.
This explanation predicts the next case and identifies why the difference stays constant.
Generalisation saves work by capturing the repeated structure.
21. A conjecture should survive rotation of the surface
If a learner believes adding two even numbers gives even, test with small numbers, large numbers, zero and different representations such as pairs of counters.
If the rule concerns factors, test it as multiplication, exact division and arrays.
If it concerns equivalent fractions, test area models, number lines and symbolic forms.
Transfer across representation is stronger evidence that the learner understands the invariant.
A rule remembered only in one visual form may still be fragile.
22. Explain what changes and what does not
Many generalisations become clearer through two columns:
| Changes | Stays invariant |
|---|---|
| Numerator and denominator in equivalent fractions | Fraction value |
| Both numbers increased by same amount | Their difference |
| Rectangle length grows by1 with fixed width4 | Width; area increases by4 |
| Decimal trailing zero added | Decimal value |
| Cube is rotated | Face adjacency and opposite relationships |
Generalisation is often the study of an invariant under a controlled change.
23. Do not generalise beyond the tested domain
A learner notices that dividing positive whole numbers by whole numbers greater than1 often produces a smaller quotient.
That observation is useful inside the current domain.
Later mathematics introduces fractions and decimals as divisors, where new behaviour appears. A rule should state the domain in which it has been established.
At Primary 4, learners do not need to master every later exception. They can learn the language “for the positive whole-number cases we are using here”.
This protects useful patterns from becoming overconfident universal rules.
24. Pattern mistakes are often assumption mistakes
| Weak move | Better question |
|---|---|
| Predicts next term from first two terms only | Does the rule fit every transition shown? |
| Calls a pattern universal after three examples | Can I explain why it must continue or find a counterexample? |
| Changes several variables at once | Which quantity is being controlled? |
| Confuses correlation in graph with cause | What does the data actually measure? |
| Ignores zero/equality case | What happens at the boundary? |
25. Practice laboratory: observe, predict, test, explain
- State the rule and next two terms: 4,250;4,350;4,450;4,550.
- State the rule: 3,6,12,24,48.
- Find the tenth term of12,18,24,30,…
- Test the conjecture “every multiple of4 is even” using three new values.
- Find a counterexample to “every number ending in2 is a multiple of4”.
- Explain why even+even is even using pairs.
- Explain why odd+odd is even.
- List factor pairs of36 and explain the stopping point.
- Two events repeat every4 and6 seconds. Explain why simultaneous meetings repeat every12 seconds.
- Explain why1/2=2/4 using equal partitioning.
- Test whether adding1 to numerator and denominator preserves1/2.
- Explain why0.6=0.60.
- Compare the rounding of3,449;3,450;3,451 to nearest hundred.
- For width4 and lengths3,4,5,6, list the rectangle areas and describe the pattern.
- For squares side3,4,5,6, list perimeters and areas. Which pattern has constant differences?
- Complete the table rule: inputs1,2,3,4 give outputs5,8,11,14. Predict output for10.
- A connected-square matchstick pattern is4,7,10,13. Predict the fifth value and explain the constant increase.
- Give one pattern where a graph shows repeated increase but cannot prove the future will continue identically.
- Create a conjecture from three examples and then search for a counterexample.
- Name one situation where a value changes but an important relationship stays invariant.
26. Explained answers
1. Add100 each time. Next: 4,650 and4,750.
2. Multiply by2 each time.
3. Nine increases of6 after the first term:12+54=66.
4. Examples28,44,100 are all divisible by4 and therefore even. The stronger explanation is that each group of4 contains two pairs.
5. 42 ends in2 but is not divisible by4.
6. Each even number is made of complete pairs. Combining two collections of complete pairs still gives complete pairs.
7. Each odd number is pairs plus one extra. The two extras combine into another pair.
8. 1×36,2×18,3×12,4×9,6×6. After6×6 the reversed pairs repeat earlier cases.
9. Twelve is a common multiple. After each simultaneous meeting the cycles restart together, so another12 seconds reproduces the alignment.
10. Splitting each half into two equal smaller pieces creates two quarters while preserving the selected region.
11. 1/2 becomes2/3, so the value changes. The rule is false.
12. 0.6 is six tenths;0.60 is sixty hundredths. Those are equal quantities.
13. Under the stated convention:3,449→3,400;3,450→3,500;3,451→3,500.
14. Areas12,16,20,24. Each extra centimetre of length adds a1×4 strip, increasing area by4.
15. Perimeters12,16,20,24 have constant increase4. Areas9,16,25,36 do not have constant first differences.
16. The rule is +3 in output for each +1 input; equivalently output=3×input+2. Input10 gives 32.
17. Fifth value=16. Each new square shares one side, so it adds3 sticks.
18. Any invented steadily rising data series can show the issue: the observed pattern describes the recorded interval but does not guarantee identical future behaviour without another basis.
19. Many answers are possible. A sound response must give examples, state the conjecture, and either find a valid counterexample or explain why the tested cases do not yet prove universality.
20. Example: add the same amount to two numbers; both values change while their difference stays invariant.
27. Teaching routine: notice, name, stretch, challenge
Begin with three or four visible cases. Ask the learner to describe exactly what changes.
Request a prediction, then a new test case not already shown.
Ask why the rule might continue. Encourage drawings, grouping, place-value language or known relationships rather than demanding formal proof.
Then challenge the wording: what happens at zero, one, equality or another boundary? Can a counterexample be found?
The goal is to turn “I see it” into “I can describe it, test it and explain what makes it work.”
28. Handover to translating representations
A generalisation becomes stronger when it survives a change of representation. The next guide moves the same mathematical relationship through words, tables, number lines, bar models and equations.
Continue to Translating Representations: Words, Tables, Number Lines, Bar Models and Equations.
Final checkpoint: can the learner state a pattern precisely, use it to predict, test new cases, search for counterexamples and explain the invariant or repeated structure beneath the examples?
Source and editorial note
The curriculum boundary is referenced to the MOE Primary Mathematics Syllabus, updated October 2025. The conjecture-testing sequence and examples are independent eduKate teaching material.
Editorial control: Wintour House V1.0 · CivDJ · eduKate Publishing.
