A school wants to teach visitors about air quality, chance and careful public communication. The plan sounds simple: put up a display, run a probability game, and print a notice. Yet each part has a different kind of truth to protect. Chemistry explains substances and processes. Mathematics measures chances within a defined sample space. English helps readers distinguish an observation, a claim, a condition and an action.
This is an original integrated English–Mathematics–Science (EMS) learner case for the 2027 Singapore–Cambridge Secondary Education Certificate at G3. EMS is a teaching approach, not an official standalone SEC examination paper. The subject codes are K300 English Language, K310 Mathematics, and, where the candidate takes Pure Chemistry, K324 Chemistry. G3 combined Science uses the separate codes K326, K327 or K328, and requirements must be checked against the relevant subject combination. The official SEAB G3 syllabus directory is the source of truth.
For particular examination requirements, use the K300 English syllabus, K310 Mathematics syllabus, K324 Pure Chemistry syllabus and K326/K327/K328 combined Science syllabus. Do not assume that a learning activity combining three subjects creates a single integrated assessment with shared marks. Each formal subject has its own approved assessment structure.
The lessons that precede this case are precise and separate. Vol 0366 teaches pollutant sources, reactions and evidence. Vol 0367 teaches the choice between exclusive, independent and changing events. Vol 0368 teaches language for effect and responsible wording in notices. Earlier, Vol 0365 offered an ester-exhibition audit with a different scientific setting. This case transfers the new skills into a new problem rather than retelling the earlier exhibition.
Everything about the school, instruments, readings, display cards, visitors and notices in this guide is fictional. The figures are invented for teaching arithmetic and reasoning, not measurements of Singapore air quality. No environmental safety threshold can be inferred from them. Students should not generate combustion products or attempt risky testing; all tasks can be solved by reading and calculating at a desk.
Imagine a modest school event called ‘Read the Air’. The exhibition has three desks. The Chemistry Desk explains how carbon monoxide, nitrogen oxides and sulfur dioxide may arise. The Numbers Desk runs a game with ten paper cards. The Visitor Desk checks the wording of a temporary corridor notice. A capable learner must keep the evidence from one desk from wandering into the claims of another.
The central skill is to carry the right amount of certainty across subjects. A pollutant source is a chemical explanation, not a measured concentration. The probability of drawing a coloured card is a property of a toy selection process, not the probability that a student will breathe polluted air. A persuasive notice can make a valid instruction clear, but cannot manufacture evidence for an unverified cause.
The fictional information desk: exact facts available to every learner
The Chemistry Desk presents three short information cards. Card C1 says carbon monoxide can form during incomplete combustion of carbon-containing fuel. Card C2 says internal combustion engines can be a source of nitrogen oxides because of reactions associated with high-temperature operation. Card C3 says sulfur-containing fuel can produce sulfur dioxide when the sulfur reacts with oxygen. These are teaching statements. They do not claim that any specific school reading identified one of the gases.
The Numbers Desk contains ten equally likely paper cards: six marked VERIFIED and four marked REVIEW. The labels refer only to the status of fictional draft display statements. An organiser may draw one card, replace it, and draw again, or draw two without replacement. The drawing rule is written separately for every question. A draw is a random classroom game, not a model of environmental exposure, health risk or an official school decision.
The same desk has a demonstration table with five invented index readings: Monday 12, Tuesday 14, Wednesday 11, Thursday 16 and Friday 17. No physical unit is supplied because the figures are not concentration readings. The mean is 14, the median is 14 and the range is 6. A visitor may calculate those summaries but must not interpret them as micrograms per cubic metre or compare them with regulatory limits.
The Visitor Desk has a confirmed operational instruction: Corridor A is unavailable for inspection from 1 p.m. until 4 p.m.; Corridor B is the signed alternative. The cause of an unrelated equipment signal has not been established. The desk must publish a concise accessible message for visitors without suggesting that the corridor closure was caused by a particular gas.
Why the three kinds of evidence are different
Chemistry asks what a substance is and how it can be formed. If an answer names CO as carbon dioxide, the reasoning fails at identity. If it calls nitrogen oxides the inevitable result of any sulfur impurity, the source mechanism is confused. Those are scientific errors even if every sentence sounds fluent.
Mathematics asks for a probability under a particular selection rule. The chance of two VERIFIED cards without replacement is (6/10)(5/9), which is 1/3. With replacement, the chance is (6/10)(6/10), which is 9/25. Neither answer represents a proportion of actual school claims in the world beyond the ten-card toy collection.
English asks readers to act on information with the intended degree of certainty. ‘Corridor A is closed for inspection until 4 p.m.; please use Corridor B’ expresses a supported operational instruction. ‘A toxic chemical leak has forced us to close Corridor A’ asserts a cause the dossier explicitly does not establish. A sentence may be forceful and still be wrong.
The integrated challenge is to decide which evidence is admissible for which question. It is tempting to convert an arithmetic result into a safety prediction or to convert a label into chemical proof. Treat this as a category error. First ask what the evidence measures. Then ask whether the conclusion refers to that same thing.
First solve the Numbers Desk completely
Among ten cards, six are VERIFIED and four are REVIEW. For one draw, P(VERIFIED)=6/10=3/5 and P(REVIEW)=4/10=2/5. These events are mutually exclusive for a single card: one card has one of the two printed statuses. They are also exhaustive, so their probabilities add to 1.
For two draws without replacement, the four ordered status paths are VV, VR, RV and RR. Their probabilities are 6/10 × 5/9 = 1/3; 6/10 × 4/9 = 4/15; 4/10 × 6/9 = 4/15; and 4/10 × 3/9 = 2/15. They sum to 1. The chance of one of each status, in either order, is 4/15+4/15=8/15.
At least one REVIEW card has probability 1−P(VV)=1−1/3=2/3. Notice that this includes VR, RV and RR. Two REVIEW cards have probability 2/15, not 2/5 × 2/5, because the second count changes after a first REVIEW card is removed.
If the cards are replaced between selections, both-stage probabilities use the original 10-card composition. Then P(VV)=36/100=9/25. The change from 1/3 to 9/25 is not a rounding issue; it is a different experiment. Your working should state the drawing rule before the fractions.
Next read the measurement table without inventing a context
The five fictional index readings 12, 14, 11, 16 and 17 total 70. Divide by five to get the mean 14. In ascending order they are 11, 12, 14, 16, 17, so the median is 14. The highest value is 17 and the lowest 11, so the range is 6.
These calculations are useful precisely because their meaning is limited. The table cannot establish that an air-quality threshold was crossed. It does not name a pollutant, instrument, unit, sampling interval, location or calibration method. A visually rising sequence after Wednesday is not proof that a particular road, school appliance or weather pattern caused it.
A good report can say: ‘In the fictional five-day index, the values ranged from 11 to 17 and had a mean of 14.’ It should not say: ‘The school’s air became dangerous on Friday.’ The latter adds an environmental conclusion not contained in the numbers.
An examiner may ask you to identify one improvement. A scientifically useful request is to specify which pollutant is being measured, the instrument’s units and calibration, the times and sites of measurement, and the purpose of the comparison. More decimal places would not repair those missing foundations.
Finally write a useful visitor message
The operational instruction should lead with confirmed information. A practical first sentence is: ‘Corridor A will be unavailable for inspection from 1 p.m. to 4 p.m. today.’ The next sentence should provide the alternative: ‘Please use the signed route through Corridor B.’ Readers should not need to search through a scientific explanation before discovering where to walk.
If an equipment signal is still being checked, treat it separately: ‘The cause of the separate equipment signal remains under review.’ Do not present that review as a reason for closing the corridor unless the dossier states a connection. Separate paragraphs can help separate related event facts from unrelated investigation facts.
Language for effect should not sacrifice accuracy. A short imperative can make the instruction easy to act on. A clear time limit can help a parent plan a route. A calm tone can avoid unnecessary alarm. None of those effects require an invented pollutant reading or a fictional health outcome.
If a previously published message made a false claim, an updated notice should state what changed and what remains reliable. But in this exercise the task is to prepare a first accurate notice. Do not insert an apology for an error that the dossier never says was published.
Thirty-one decisions in the school information dossier
Checkpoint 01: Name a gas correctly
Desk question. The Chemistry Desk caption says ‘CO is carbon dioxide’. Identify the error and correct it.
Worked answer. CO is carbon monoxide; carbon dioxide is CO2. A small change in chemical formula changes the substance.
Evidence boundary. This is an identity correction, not an argument about whether the measured air is safe.
Read this as a three-desk evidence problem, not as a race to produce an impressive sentence. Ask which desk owns the claim and which evidence would be needed to support it. If a probability came from cards, it should stay attached to cards. If a gas was named in a mechanism, it should not become a measured on-site concentration. If an instruction was confirmed, it can be written firmly.
Checkpoint 02: Explain a possible source
Desk question. A fictional question describes incomplete combustion of a carbon-containing fuel and asks for one possible gaseous product.
Worked answer. Carbon monoxide is a defensible possible product of incomplete combustion. A complete statement links reduced oxygen availability to incomplete combustion.
Evidence boundary. Do not claim that carbon monoxide was measured at the school, because the dossier supplies no chemical test.
Build the working from the data that were actually given. Write the chemical formula if the case concerns identity; write the changing sample size if the case concerns card draws; write the audience, period and required action if the case concerns a notice. A useful answer is traceable from evidence to conclusion.
Checkpoint 03: Identify nitrogen oxides
Desk question. A panel mentions high-temperature combustion engines and asks why nitrogen oxides can appear.
Worked answer. High temperatures can enable reactions of nitrogen and oxygen from the air, contributing to nitrogen oxides.
Evidence boundary. Do not treat sulfur dioxide as synonymous with nitrogen dioxide.
Test the answer by changing one condition. Returning a card changes the second branch. Adding a missing unit would change what might be inferred from a graph. Replacing ‘may’ with ‘did’ changes the certainty of a chemical or causal claim. This changed-detail test reveals whether you understand the governing structure.
Checkpoint 04: Follow sulfur impurities
Desk question. A mock fuel sample is described as containing sulfur. What possible oxide should the panel mention after combustion?
Worked answer. Sulfur dioxide can form when sulfur-containing material reacts with oxygen during burning.
Evidence boundary. The statement is conditional on the composition and process described; it says nothing about an actual sample collected on campus.
Evaluate the limit of the evidence before you expand the answer. Numbers without units support certain arithmetic, not regulatory conclusions. Mechanisms explain possibilities, not specific measured events. A notice can be clear without making a scientific claim. Precise limits protect both marks and public understanding.
Checkpoint 05: Avoid an acid-rain shortcut
Desk question. A draft says ‘acid rain is pumped directly from every exhaust pipe’. What needs revision?
Worked answer. Certain oxides can undergo atmospheric reactions that contribute to acidic precipitation. Acid rain is not simply unaltered exhaust gas.
Evidence boundary. Describe the intermediate process when a question asks how emissions lead to environmental effects.
Under examination pressure, keep three separate questions visible: What is being asked? What evidence was provided? What is the smallest complete answer? That habit prevents an attractive but irrelevant paragraph from hiding the missing calculation, reaction explanation or reader instruction.
Checkpoint 06: Separate ozone locations
Desk question. A visitor asserts that all ozone must be harmful because a poster calls it a pollutant.
Worked answer. Ozone near ground level can be harmful at elevated concentrations, while ozone higher in the atmosphere plays a protective role.
Evidence boundary. Use context and location before attaching a simple ‘good’ or ‘bad’ label to a substance.
Read this as a three-desk evidence problem, not as a race to produce an impressive sentence. Ask which desk owns the claim and which evidence would be needed to support it. If a probability came from cards, it should stay attached to cards. If a gas was named in a mechanism, it should not become a measured on-site concentration. If an instruction was confirmed, it can be written firmly.
Checkpoint 07: Treat methane as methane
Desk question. A statement identifies CH4 as carbon monoxide.
Worked answer. CH4 is methane, a greenhouse gas; carbon monoxide is CO.
Evidence boundary. Do not use an environmental effect to excuse an incorrect chemical identity.
Build the working from the data that were actually given. Write the chemical formula if the case concerns identity; write the changing sample size if the case concerns card draws; write the audience, period and required action if the case concerns a notice. A useful answer is traceable from evidence to conclusion.
Checkpoint 08: Classify one draw
Desk question. A visitor picks one card from six VERIFIED and four REVIEW cards. Find P(REVIEW).
Worked answer. Four of ten equally likely cards qualify, so P(REVIEW)=4/10=2/5.
Evidence boundary. This probability describes paper cards only. It cannot be transferred to chance of inhaling a pollutant.
Test the answer by changing one condition. Returning a card changes the second branch. Adding a missing unit would change what might be inferred from a graph. Replacing ‘may’ with ‘did’ changes the certainty of a chemical or causal claim. This changed-detail test reveals whether you understand the governing structure.
Checkpoint 09: A mutually exclusive pair
Desk question. Can one printed card be both VERIFIED and REVIEW under the stated one-label rule?
Worked answer. No. The two events are mutually exclusive for a single draw, and 3/5+2/5=1.
Evidence boundary. State the one-label assumption clearly; a different classification system could permit multiple labels.
Evaluate the limit of the evidence before you expand the answer. Numbers without units support certain arithmetic, not regulatory conclusions. Mechanisms explain possibilities, not specific measured events. A notice can be clear without making a scientific claim. Precise limits protect both marks and public understanding.
Checkpoint 10: Two VERIFIED without replacement
Desk question. The organiser draws two cards without replacement. Find the probability both are VERIFIED.
Worked answer. P(VV)=6/10×5/9=30/90=1/3.
Evidence boundary. The second numerator and denominator change after a verified card is taken.
Under examination pressure, keep three separate questions visible: What is being asked? What evidence was provided? What is the smallest complete answer? That habit prevents an attractive but irrelevant paragraph from hiding the missing calculation, reaction explanation or reader instruction.
Checkpoint 11: Two VERIFIED with replacement
Desk question. The first card is returned before the next draw. Find the probability both are VERIFIED.
Worked answer. P(VV)=6/10×6/10=36/100=9/25.
Evidence boundary. Returning a card preserves the original composition and supports an independent repeated-draw model.
Read this as a three-desk evidence problem, not as a race to produce an impressive sentence. Ask which desk owns the claim and which evidence would be needed to support it. If a probability came from cards, it should stay attached to cards. If a gas was named in a mechanism, it should not become a measured on-site concentration. If an instruction was confirmed, it can be written firmly.
Checkpoint 12: One of each status
Desk question. Two cards are drawn without replacement. Find one VERIFIED and one REVIEW in any order.
Worked answer. Add VR=6/10×4/9=4/15 and RV=4/10×6/9=4/15. Total=8/15.
Evidence boundary. Two ordered paths are mutually exclusive complete outcomes; counting just one loses half the target event.
Build the working from the data that were actually given. Write the chemical formula if the case concerns identity; write the changing sample size if the case concerns card draws; write the audience, period and required action if the case concerns a notice. A useful answer is traceable from evidence to conclusion.
Checkpoint 13: Two REVIEW cards
Desk question. Calculate the probability of two REVIEW cards without replacement.
Worked answer. P(RR)=4/10×3/9=12/90=2/15.
Evidence boundary. After one review card is selected, three remain among nine cards.
Test the answer by changing one condition. Returning a card changes the second branch. Adding a missing unit would change what might be inferred from a graph. Replacing ‘may’ with ‘did’ changes the certainty of a chemical or causal claim. This changed-detail test reveals whether you understand the governing structure.
Checkpoint 14: At least one REVIEW
Desk question. Find the probability that two draws without replacement include at least one REVIEW card.
Worked answer. Use the complement of VV: 1−1/3=2/3.
Evidence boundary. The event includes VR, RV and RR; no second category is silently excluded.
Evaluate the limit of the evidence before you expand the answer. Numbers without units support certain arithmetic, not regulatory conclusions. Mechanisms explain possibilities, not specific measured events. A notice can be clear without making a scientific claim. Precise limits protect both marks and public understanding.
Checkpoint 15: Sum the four outcomes
Desk question. Check whether VV, VR, RV and RR describe an exhaustive pair of draws.
Worked answer. Their probabilities are 1/3, 4/15, 4/15 and 2/15. In fifteenths the sum is 5+4+4+2=15, or 1.
Evidence boundary. A total of 1 checks the branch model but does not prove the random-draw assumption for real-world behaviour.
Under examination pressure, keep three separate questions visible: What is being asked? What evidence was provided? What is the smallest complete answer? That habit prevents an attractive but irrelevant paragraph from hiding the missing calculation, reaction explanation or reader instruction.
Checkpoint 16: A card game misused
Desk question. A display caption says ‘Two thirds of pupils are at environmental risk’ based on the card-game answer.
Worked answer. The 2/3 value is the chance of drawing at least one REVIEW card in a particular toy selection process. It has no information about pupil exposure.
Evidence boundary. Remove the real-world risk claim; retain only the correctly labelled card-game probability.
Read this as a three-desk evidence problem, not as a race to produce an impressive sentence. Ask which desk owns the claim and which evidence would be needed to support it. If a probability came from cards, it should stay attached to cards. If a gas was named in a mechanism, it should not become a measured on-site concentration. If an instruction was confirmed, it can be written firmly.
Checkpoint 17: An unexplained index
Desk question. A table shows 12,14,11,16,17 with no measuring unit. Can it prove a legal threshold was crossed?
Worked answer. No. The quantity and unit are unspecified, as is any applicable threshold.
Evidence boundary. The mean and range can still be calculated as mathematical summaries of the invented numbers.
Build the working from the data that were actually given. Write the chemical formula if the case concerns identity; write the changing sample size if the case concerns card draws; write the audience, period and required action if the case concerns a notice. A useful answer is traceable from evidence to conclusion.
Checkpoint 18: Mean of five readings
Desk question. Calculate the arithmetic mean of 12,14,11,16,17.
Worked answer. Sum=70; divide by 5 to obtain mean=14.
Evidence boundary. Show the count of observations, not merely a calculator result.
Test the answer by changing one condition. Returning a card changes the second branch. Adding a missing unit would change what might be inferred from a graph. Replacing ‘may’ with ‘did’ changes the certainty of a chemical or causal claim. This changed-detail test reveals whether you understand the governing structure.
Checkpoint 19: Median of five readings
Desk question. Determine the median of the same values.
Worked answer. Sorted order is 11,12,14,16,17; the middle is 14.
Evidence boundary. A median is an ordered-position measure, not necessarily the same as the mean for every data set.
Evaluate the limit of the evidence before you expand the answer. Numbers without units support certain arithmetic, not regulatory conclusions. Mechanisms explain possibilities, not specific measured events. A notice can be clear without making a scientific claim. Precise limits protect both marks and public understanding.
Checkpoint 20: Range of the log
Desk question. Find the range of the five demonstration values.
Worked answer. Largest=17, smallest=11; range=6.
Evidence boundary. The units, if any, would follow the measured quantity, but none were supplied in this fictional index.
Under examination pressure, keep three separate questions visible: What is being asked? What evidence was provided? What is the smallest complete answer? That habit prevents an attractive but irrelevant paragraph from hiding the missing calculation, reaction explanation or reader instruction.
Checkpoint 21: A Friday claim
Desk question. A pupil writes ‘Friday was unsafe because its index was 17’. What is wrong?
Worked answer. The invented index has no stated physical interpretation or safety threshold. The number 17 alone does not justify ‘unsafe’.
Evidence boundary. A correct sentence may state that Friday had the highest listed value.
Read this as a three-desk evidence problem, not as a race to produce an impressive sentence. Ask which desk owns the claim and which evidence would be needed to support it. If a probability came from cards, it should stay attached to cards. If a gas was named in a mechanism, it should not become a measured on-site concentration. If an instruction was confirmed, it can be written firmly.
Checkpoint 22: A false cause
Desk question. A caption says a bus passing on Friday caused the value 17.
Worked answer. One timing coincidence cannot isolate a cause from other influences; the dossier contains no controlled evidence of emission source.
Evidence boundary. Replace the causal assertion with the observed value and a suggestion for valid further investigation.
Build the working from the data that were actually given. Write the chemical formula if the case concerns identity; write the changing sample size if the case concerns card draws; write the audience, period and required action if the case concerns a notice. A useful answer is traceable from evidence to conclusion.
Checkpoint 23: An effective first sentence
Desk question. Write the opening sentence for a notice when Corridor A is closed for inspection from 1–4 p.m.
Worked answer. ‘Corridor A is closed for inspection today from 1 p.m. to 4 p.m.’ is direct and complete.
Evidence boundary. It states the known operational restriction without attaching it to an unverified air-quality cause.
Test the answer by changing one condition. Returning a card changes the second branch. Adding a missing unit would change what might be inferred from a graph. Replacing ‘may’ with ‘did’ changes the certainty of a chemical or causal claim. This changed-detail test reveals whether you understand the governing structure.
Checkpoint 24: Give the alternative
Desk question. Write a sentence telling visitors what to do when Corridor A is closed.
Worked answer. ‘Please follow the signs and use Corridor B during this period.’ supplies the practical next step.
Evidence boundary. An actionable direction helps the audience more than an unsupported dramatic headline.
Evaluate the limit of the evidence before you expand the answer. Numbers without units support certain arithmetic, not regulatory conclusions. Mechanisms explain possibilities, not specific measured events. A notice can be clear without making a scientific claim. Precise limits protect both marks and public understanding.
Checkpoint 25: Don’t claim a leak
Desk question. A pupil suggests the headline ‘Toxic leak forces school corridor closure’. Does it fit the dossier?
Worked answer. No. Neither a toxic leak nor a causal connection to the closure is established.
Evidence boundary. An urgent-sounding headline is not justified unless the underlying event is confirmed.
Under examination pressure, keep three separate questions visible: What is being asked? What evidence was provided? What is the smallest complete answer? That habit prevents an attractive but irrelevant paragraph from hiding the missing calculation, reaction explanation or reader instruction.
Checkpoint 26: Make a condition visible
Desk question. The draft says ‘Use Corridor B’ but fails to state when. Improve it.
Worked answer. Add the relevant period: ‘From 1 p.m. to 4 p.m. today, please use Corridor B while Corridor A is closed.’
Evidence boundary. The time limit controls the instruction. Without it the notice is broader than the operational fact.
Read this as a three-desk evidence problem, not as a race to produce an impressive sentence. Ask which desk owns the claim and which evidence would be needed to support it. If a probability came from cards, it should stay attached to cards. If a gas was named in a mechanism, it should not become a measured on-site concentration. If an instruction was confirmed, it can be written firmly.
Checkpoint 27: Report the unresolved signal
Desk question. The equipment signal’s cause remains under review. Write a fair sentence.
Worked answer. ‘The cause of the separate equipment signal is still being investigated.’ accurately reports the status.
Evidence boundary. Do not link the signal to a particular pollutant or the corridor closure without evidence.
Build the working from the data that were actually given. Write the chemical formula if the case concerns identity; write the changing sample size if the case concerns card draws; write the audience, period and required action if the case concerns a notice. A useful answer is traceable from evidence to conclusion.
Checkpoint 28: Evaluate language for effect
Desk question. A pupil uses ‘Check the route. Check the time. Check the signs.’ What does repetition accomplish?
Worked answer. The three imperatives organise verification steps and create a concise memorable rhythm.
Evidence boundary. Explain the behaviour encouraged by the wording, rather than stating only that repetition is used.
Test the answer by changing one condition. Returning a card changes the second branch. Adding a missing unit would change what might be inferred from a graph. Replacing ‘may’ with ‘did’ changes the certainty of a chemical or causal claim. This changed-detail test reveals whether you understand the governing structure.
Checkpoint 29: Compare a headline and caveat
Desk question. A poster says ‘FREE ENTRY’ in large letters but ‘registration required’ in smaller print. Explain the meaning.
Worked answer. Entry may be free for registered visitors, but free entry does not mean there is no condition to attend.
Evidence boundary. A careful reader integrates the prominent invitation with the limiting condition.
Evaluate the limit of the evidence before you expand the answer. Numbers without units support certain arithmetic, not regulatory conclusions. Mechanisms explain possibilities, not specific measured events. A notice can be clear without making a scientific claim. Precise limits protect both marks and public understanding.
Checkpoint 30: Keep the three desks apart
Desk question. A final write-up mixes gas formulas, card probabilities and corridor instructions into one sentence with no boundaries.
Worked answer. Separate paragraphs by evidential purpose: first the chemical explanation, then the toy probability result, then the operational notice.
Evidence boundary. The integrated skill is correct handoff across subjects; each claim still needs its own source and conditions.
Under examination pressure, keep three separate questions visible: What is being asked? What evidence was provided? What is the smallest complete answer? That habit prevents an attractive but irrelevant paragraph from hiding the missing calculation, reaction explanation or reader instruction.
Checkpoint 31: Compose a final evidence note
Desk question. Write a three-sentence conclusion suitable for the fictional display record.
Worked answer. ‘The Chemistry Desk describes possible pollutant sources without claiming on-site measurements. The Numbers Desk calculates chances from ten labelled cards under stated draw rules. Corridor A is closed for inspection from 1 p.m. to 4 p.m.; visitors should use Corridor B.’
Evidence boundary. The conclusion is useful because it says what each source establishes and refuses to smuggle one desk’s evidence into another desk’s claim.
Read this as a three-desk evidence problem, not as a race to produce an impressive sentence. Ask which desk owns the claim and which evidence would be needed to support it. If a probability came from cards, it should stay attached to cards. If a gas was named in a mechanism, it should not become a measured on-site concentration. If an instruction was confirmed, it can be written firmly.
A final transferable EMS rehearsal
Make three columns headed chemistry, mathematics and English. Choose one chemical claim, one probability claim and one visitor instruction from the dossier. In each column write the original evidence, a correct answer, an attractive but false answer and a single sentence explaining the point of failure. This is more demanding than copying the model because you must identify the boundary of each kind of knowledge.
Then change the fictional card game to five VERIFIED and five REVIEW cards. Without replacement, the probability of two VERIFIED cards becomes (5/10)(4/9)=2/9. The probability of one of each becomes twice (5/10)(5/9)=5/9. This changed-count exercise tests whether the mathematics transfers rather than merely being remembered as 1/3 and 8/15 from the earlier ten-card setup. The sums and answers should change; the method should remain stable.
Finally rewrite the visitor notice in a version suitable for a younger pupil and another suitable for a parent. Both must preserve the 1 p.m. to 4 p.m. restriction, the inspection reason as stated and the use of Corridor B. Neither should invent a gas, health effect or reopening guarantee beyond the stated period. You have succeeded when the language changes appropriately while the supported facts remain the same.
Use the G3 SEC learner guide to continue, the PSLE foundation route where earlier reading or mathematics needs repair, the English practice guide, the Mathematics combined-events guide and the Science hub for their subject owners. The SEAB syllabus directory remains the authority for 2027 subject codes and formats. All 31 cases here are original teaching material.
