Start with the mathematical question, not the biological story
A recovery table can look simple. A runner stops exercising; pulse is measured immediately, then after one, three and five minutes. The numbers decrease. But the mathematical questions may be harder than subtracting the first value from the last. Are the intervals equal? Which reading is the baseline? Is a percentage change calculated from the resting value or the peak? Are two pupils being compared at matching times? Does a graph show actual continuous behaviour or a small number of recorded points?
This chapter follows Volume 0667, the Biology guide to exercise respiration. Biology explains why breathing and circulation may respond during activity and recovery. Mathematics keeps the measurements, their denominators and their time structure honest. Pulse rate is an indirect variable. The computations below are practice with time series and rates, not a way to diagnose fitness, calculate oxygen debt or prescribe exercise. The figures are fictional.
Current 2027 G3 Mathematics examination requirements
Under the 2027 Singapore-Cambridge SEC, G3 Mathematics has subject code K310, replacing reference code 4052. The official scheme has Paper 1 and Paper 2, each two hours fifteen minutes, each ninety marks and each carrying 50% of the subject total. Paper 1 has approximately 26 short-answer questions; Paper 2 contains nine to ten questions of varying lengths and the final question focuses on applying Mathematics to a real-world scenario. All questions are compulsory. Essential working is expected.
Relevant content includes tables, line graphs, mean and median, quartiles and standard deviation, straight-line relationships and the interpretation of data in real contexts. The syllabus convention for non-exact numerical answers is generally three significant figures unless a question specifies otherwise. The full document, not a remembered older code, is authoritative: SEAB G3 Mathematics K310 syllabus for 2027. This chapter’s fitness-style context is invented to practise mathematical transfer; it is not a promised examination topic.
Read every column heading before doing anything
A table might have headings ‘time after exercise, minutes’, ‘pulse rate, beats per minute’, and ‘person A’. The first column is elapsed time since the end of exercise, not clock time. The second is a rate estimated at each sampling time, not a count of all beats since the previous row. The third identifies whose data were measured. If the time column begins with zero, it means immediately after exercise in this example; baseline pulse could have been measured separately before activity.
Students who scan the entries without examining the headings often combine quantities that have different meanings. Subtracting 120 from 144 gives a change of 24 beats per minute in recorded pulse rate. It does not give the number of heartbeats occurring during the interval. The distinction matters whenever a question mixes the words count, rate and total.
The central dataset: keep it visible
Suppose the following records were made. Person A had a resting baseline of 72 beats per minute. At 0, 1, 3 and 5 minutes after exercise, their measured pulse rates were 144, 120, 96 and 80 beats per minute. Person B had a resting baseline of 78, followed by 150, 130, 94 and 82 at the same four post-activity times. Assume all data are illustrative and the measuring method is consistent within this constructed example.
Before calculating, copy the times as 0, 1, 3, 5 rather than writing the positions of the rows as 1, 2, 3, 4. The intervals are one, two and two minutes. A student who treats each adjacent row as one minute will obtain misleading gradients. Keep the baseline outside the timed recovery series unless the question explicitly gives it an elapsed time coordinate.
A rate already has a unit within its unit
Pulse rate is expressed as beats per minute. It is already a ratio, calculated from counted beats over a defined observation period or estimated by a measuring instrument. The unit is not simply ‘beats’. If a timed plot tracks how pulse rate changes over elapsed minutes, the gradient has the compound unit ‘(beats per minute) per minute’, sometimes written beats per minute squared. Do not simplify the interpretation to ‘the person lost 24 beats’ when you mean the recorded rate fell by 24 beats per minute.
Whenever a problem has a rate on the vertical axis and time on the horizontal axis, the gradient represents change in the rate per unit time. The physical meaning may be unfamiliar, but the mathematical method remains vertical change divided by horizontal change. Writing the units below the calculation is often enough to reveal a wrong denominator.
First calculate ordinary differences
For A, from the immediate post-exercise reading to one minute the recorded pulse rate falls from 144 to 120. The change, taking final minus initial, is 120 − 144 = −24 beats per minute. From one to three minutes it changes by 96 − 120 = −24. From three to five minutes it changes by 80 − 96 = −16. The signed changes show direction. The magnitudes 24, 24 and 16 show how far the rate moved at the sampled points.
The first two changes have equal magnitudes but occur over different times. That does not mean the rate of change is equal. Before interpreting a table, decide whether the question asks for a difference in the measured quantity or a change per minute. Mixing these operations is one of the fastest ways to lose otherwise accessible marks.
Use unequal-interval gradients correctly
Average gradient over a time interval is (pulse rate at final time − pulse rate at initial time) divided by (final time − initial time). For A, during 0 to 1 minute: (120 − 144)/(1 − 0) = −24 (beats/min)/min. During 1 to 3 minutes: (96 − 120)/(3 − 1) = −12. During 3 to 5 minutes: (80 − 96)/(5 − 3) = −8. The measurements suggest the average decline is steepest in the first interval and less steep in later intervals.
These are average gradients between observations, not instantaneous derivatives and not a proof of a particular physiological law. A curved real recovery pattern could change inside the interval. If the data are sparse, drawing smooth shapes between points adds assumptions. On paper, show the ordered pair, the subtraction and the interval length before interpreting the result.
The whole-period average rate of change
For A, the recorded rate changes from 144 at time zero to 80 at five minutes. The whole-period average gradient is (80 − 144)/(5 − 0) = −64/5 = −12.8 (beats/min)/min. It does not equal the simple average of the three interval gradients: (−24 − 12 − 8)/3 = −14.7 approximately. The intervals are not of equal duration, so equal weighting of the three gradients changes the meaning of the average.
An equivalent interval-weighted calculation is [−24 × 1 + (−12) × 2 + (−8) × 2]/(1+2+2) = −64/5 = −12.8. The weights here are elapsed minutes, not pulse readings. Students should understand why weighting works: the total change over each interval equals gradient times duration, and the overall gradient is total change divided by total duration.
Do not average four spot readings and call it a time average
For A, the arithmetic mean of the four sampled pulse rates is (144 + 120 + 96 + 80)/4 = 110 beats per minute. This is the unweighted mean of four observations. The sampling times, however, are not evenly spaced. It does not necessarily represent the average pulse rate across the full five minutes. The table does not reveal the values between observations. A defensible statement is ‘the mean of the four recorded spot readings was 110 beats per minute’.
One may use interpolation or a chosen model to estimate a time-weighted average, but doing so requires clearly stated assumptions. Without such instructions, avoid presenting a model-based value as an exact measured fact. This example differs from the whole-period gradient above, which depends only on the two endpoint measurements. A common examination trap is that both calculations use the word average but measure different things.
Compare baselines before interpreting differences
Person A started with a resting baseline of 72; Person B with 78. At the five-minute sample A was at 80, eight above baseline. B was at 82, four above baseline. The smaller absolute gap belongs to B, but that observation alone cannot establish who is fitter. We only know the measured differences under this constructed protocol. If the question asks for the gap, subtract each person’s own baseline rather than comparing both against 72.
A value of 80 beats per minute does not have the same relation to a baseline of 72 as it does to a baseline of 78. This is why ‘who has the lowest five-minute rate?’ and ‘who is closest to their own baseline?’ are separate questions. Explain the comparison standard before choosing a winner.
Find relative change from a named base
Immediately after exercise, A is at 144 against baseline 72. The elevation is 72 beats per minute, equivalent to (144 − 72)/72 × 100% = 100% above the baseline. B is at 150 against baseline 78; the elevation is 72 beats per minute but the percentage is 72/78 × 100% ≈ 92.3%. Identical absolute changes can represent different relative changes because the denominators differ.
A percentage change from peak to five-minute reading answers another question: for A, (80 − 144)/144 × 100% ≈ −44.4%. It would be wrong to use 72 as the denominator in that peak-to-five-minute percentage unless the problem specifically asked for a fraction of baseline. Always label the original reference quantity in words, then compute.
The difference between percentage and percentage points
If a chart reports the proportion of students whose pulse returns within a given band, the values may be percentages. A change from 40% to 55% is 15 percentage points, but a relative increase of (55 − 40)/40 × 100% = 37.5%. These are not interchangeable. The phrase ‘rose by 15%’ would be ambiguous or wrong if the writer intended a 15 percentage-point change.
Real-world context questions often put several kinds of percentage near one another. Ask whether the data are rates, proportions of a group, or changes in an individual’s measured variable. Write ‘percentage points’ for differences between percentages and ‘percent’ for a relative ratio using a clearly stated denominator.
Graph it without making a staircase by accident
Put elapsed recovery time on the horizontal axis, marked at 0, 1, 3 and 5 minutes using proportional spacing. Plot pulse rate on the vertical axis with a labelled unit. For Person A place the points (0,144), (1,120), (3,96) and (5,80). If the problem calls for a line graph, connect successive points appropriately. If it calls for a scatter plot, do not assume it needs connecting lines.
Incorrect graphs often space times 0,1,3,5 equally along the horizontal axis, as though they were four categories. That visually changes the gradients. A second error is making the vertical scale too compressed or hiding zero without explaining the choice. A truncated axis may be acceptable when clearly labelled, but it can exaggerate visual differences. Read the actual coordinates before making claims about steepness.
Which slope is steeper? Check signs and magnitudes
Suppose Person A has a gradient of −12 between one and three minutes and B has (94 − 130)/(3 − 1) = −18 over the same period. B’s rate of decrease has a greater magnitude over that interval. Saying ‘−12 is greater than −18’ is numerically true but does not answer a question about steepness or speed of decline. A line with gradient −18 falls more quickly with increasing time than a line of gradient −12.
Use three separate descriptions: numerical order, direction of the line, and magnitude of the gradient. For a negative gradient, increasing magnitude produces a more steeply downward line. Do not claim one recovery profile is medically superior on that basis; the mathematics describes the curve under the supplied conditions.
Comparing two pupils at different time points
A school report says that A’s pulse was 96 while B’s was 130 and concludes that A recovered faster. These measurements might be A at three minutes and B at one minute. Without matching the elapsed times, the comparison is not fair. Compare A and B at zero, one, three or five minutes when possible. If readings come from different times, a justified interpolation or other model would be required to make a same-time comparison.
Even with matching times, the initial values and baselines may differ. The correct statement depends on the question: an absolute rate, a fall from peak, a change relative to baseline or the recorded value at a fixed time. This discipline is the mathematical counterpart to experimental controls in Biology.
A real-world last-question strategy
The 2027 K310 Paper 2 includes a last question focused on applying Mathematics to a real-world scenario. In such tasks, start by extracting what each quantity means, then identify which formula or representation fits. Draw a small information map: baseline, starting observation, later observations, elapsed intervals, requested comparisons and restrictions. Do not immediately apply every formula you know.
If a question asks whether an exhibition’s chart supports a statement, calculate the necessary values and finish with a sentence tied to the numerical finding. A perfect calculation without interpreting the result may be incomplete. Conversely, an elegant recommendation unsupported by calculation may not earn the mathematical marks. The strongest answer shows its route: data, method, result and contextual conclusion.
When a straight-line model is reasonable and when it fails
A linear model y = mx + c can describe a constant-rate change over a chosen interval. Using A’s readings at one and three minutes gives m = −12 and c = 132, so a line through those points is y = −12x + 132, with x in minutes and y as pulse rate in beats per minute. At x = 1 it predicts 120; at x = 3 it predicts 96. That makes it an exact interpolation through the two selected observations.
At x = 5 it predicts 72, whereas the supplied reading is 80. The model does not fit that later point. Extrapolating indefinitely would eventually yield absurd negative pulse rates. The algebra may be correct while the model becomes unsuitable. State the domain for which a model is used and check it against available measurements.
An inverse problem from a line segment
For the same model y = −12x + 132 between one and three minutes, find the modelled time at which the pulse rate reaches 108. Solve 108 = −12x + 132, so −24 = −12x and x = 2 minutes. Because x = 2 lies within the one-to-three-minute interval, the interpolation is within the chosen model’s domain. It is still a modelled time, not a measurement taken precisely at two minutes.
If the answer had been x = 9, the result would lie far beyond the interval and should not be offered as a reliable predicted recovery time. A mathematics examination may accept an algebraic answer to a stated model; strong real-world reasoning adds a domain check.
How missing readings affect comparisons
Imagine Person B has no three-minute reading. Do not replace the missing value with zero in a mean or graph. Zero is a number with meaning, not a symbol for data that were not collected. Either leave the observation missing, omit it from the specific calculation while stating the denominator, or use an explicitly requested estimation method. Compare groups cautiously when one has fewer observations.
Another trap is to connect directly from one minute to five minutes and call the resulting slope the ‘one-to-three-minute rate’. It is actually the rate over one to five minutes. A missing midpoint cannot be conjured by renaming a longer interval. Label the interval covered by every calculation.
Outliers, transcription errors and honest data repair
If a row contains 800 beats per minute after the other entries are near 80–150, check the original record. The value might be a misplaced zero, a device fault or a genuine but unusual entry; numerical oddness alone does not authorise silently changing it. Explain any verified correction. If the raw source is unavailable, identify it as questionable and consider its effect on the summary.
An outlier can greatly affect a mean. Reporting the median or showing results with and without a flagged point can illuminate the effect, provided the choices are explained. Do not selectively discard only the observations that weaken a preferred conclusion. That is not mathematical analysis; it is altering the evidence.
Median and mean answer different questions
For A’s four post-exercise recorded values 144, 120, 96 and 80, order them as 80, 96, 120, 144. The median is (96 + 120)/2 = 108 beats per minute. The mean is 110. The two values differ because they summarise the observed numbers in different ways. Neither tells us the time average across the five-minute period without assumptions about unobserved values.
If an unusual extra reading of 240 is appended, the mean changes dramatically, while the median of the five sorted values becomes 120. This demonstrates sensitivity to extreme observations, not a reason to prefer median in every question. Choose a summary to answer a defined purpose and state exactly which observations are included.
Variation: range, quartiles and standard deviation
If several students repeat a standard protocol, a table may compare distributions rather than one time series. The range describes maximum minus minimum, quartiles describe positions in the ordered data, and standard deviation measures spread about a mean. The 2027 K310 syllabus includes these forms of summary. Compare groups using both typical value and spread when the question warrants it.
Do not calculate standard deviation across time points for one pupil and then automatically interpret it as variation between pupils. The observational units matter. One set contains measurements at different times; another contains different people under a matched condition. They answer different questions. A formula only becomes useful once you decide which data are meaningfully comparable.
A simple standard-deviation comparison
Consider two fictional matched-time datasets: Group C has rates 90, 90, 90 and 90; Group D has 75, 85, 95 and 105. Both have mean 90. Group C has zero standard deviation because all readings match. Group D has greater spread because its values vary around the same mean. Knowing the mean alone hides that difference.
The calculation is not a diagnosis. In a school statistical question, explain what variability means in that chosen set. Before transferring it to a practical claim, ask whether the groups were measured fairly and whether four readings are enough for the proposed generalisation. A statistic measures a feature of the supplied data, not the truth of every story told around it.
Estimate missing values only when asked
If a question explicitly instructs linear interpolation between A’s readings at times one and three, at two minutes the model predicts halfway between 120 and 96, namely 108 beats per minute. Label 108 as an estimate and name the assumption: a straight-line change between the two sampled points. Without that assumption, the two endpoint observations do not uniquely determine the intervening value.
Students sometimes write the estimated point into the original table without marking it. This changes the evidential status of the dataset. Keep original observations and estimates distinct. In an examination, you can draw an open marker or annotate ‘estimated’ if the format allows, then use it for the specified model calculation.
The first complete worked problem
Question: A student records a pulse rate of 156 beats/min immediately after activity and 114 beats/min after two minutes. Find the average rate of change during this interval. Working: (114 − 156)/(2 − 0) = −42/2 = −21 (beats/min)/min. Interpretation: the recorded pulse rate fell by an average of 21 beats per minute in each minute of that two-minute interval, assuming the comparison is of the values at the specified times.
Do not describe this as ’21 beats fewer overall’. The change in the reported rate is 42 beats/min, while the average rate of change of that rate is 21 (beats/min)/min. These have different dimensions. Writing the negative sign in the calculation and then ‘a decline of 21’ in the sentence makes the sign’s meaning clear.
The second complete worked problem
Question: One student’s resting rate is 68 beats/min and the immediate post-activity rate is 136. Another’s resting rate is 80 and immediate post-activity rate is 152. Which has the greater percentage rise above its own baseline? For the first: (136 − 68)/68 × 100% = 100%. For the second: (152 − 80)/80 × 100% = 90%. So the first has the greater relative rise although the absolute increases are 68 and 72 respectively.
This example makes a vital distinction. The student with the larger number of additional beats per minute did not have the larger percentage increase. When the starting quantities differ, absolute and relative comparisons can reverse rankings. That is why a conclusion must say which metric it uses.
The third complete worked problem
Question: A recovery rate drops by 20 beats/min in the first minute and 30 beats/min in the next three minutes. Find the average drop per minute across the four minutes. The total decrease is 20 + 30 = 50 beats/min. The time is 1 + 3 = 4 minutes. So the average decrease in pulse rate per minute is 50/4 = 12.5 (beats/min)/min. The signed overall gradient is −12.5.
The error to avoid is averaging 20 and 10 as if their intervals had the same duration, or averaging 20 and 30 directly. The first quantity, 20, is a rate change over one minute; the second, 30, is a total rate change over three minutes. Convert to consistent objects before combining them.
The fourth complete worked problem
Question: A line model for pulse rate between minutes two and four is y = −9x + 126. Find its predicted rate at x = 3 and the predicted time when y = 99. At x = 3, y = −27 + 126 = 99 beats/min. Solving 99 = −9x +126 gives x = 3. Both methods agree. The consistency is a useful reverse check.
Now test whether the same model predicts sensible values at x = 20: y = −180 + 126 = −54, an impossible pulse rate. The equation is not a law of physiology, only a local mathematical fit. The context determines the acceptable domain of use.
The fifth complete worked problem
Question: Four spot readings at times 0, 1, 3 and 5 are 150, 130, 100 and 90. Find their arithmetic mean and explain one limitation of using it as a full-period average. Mean = (150 + 130 + 100 + 90)/4 = 117.5 beats/min. Since the times are uneven and values between readings are unknown, that figure is the mean of the four collected observations rather than an exact time average over five minutes.
A high-quality answer does not reject 117.5 as a wrong calculation. It identifies exactly what that correct number means and the claim it cannot support. Many examination errors arise from giving a correct arithmetic answer the wrong label.
The sixth complete worked problem
Question: A class plots two graphs with identical data but different vertical scales. In Chart 1 the axis runs from 0 to 200. In Chart 2 it runs from 130 to 160. The second chart makes a fifteen-beat-per-minute difference look enormous. Which chart is fraudulent? Not necessarily either, if axes are clear, but the presentation changes the visual impression.
Compare the actual labelled values rather than the apparent height of the lines. If choosing a public chart, consider whether a reader could misinterpret the scale and provide enough context to prevent it. The mathematical task is not just creating a graph; it is communicating quantities without distortion.
The seventh complete worked problem
Question: A report claims that one student returned to baseline in two minutes. The only measurements were at one minute, when rate was 90, and three minutes, when rate was 70; the pre-exercise baseline was 74. The student may have reached 74 between the recorded times, but the exact moment is not observed. A straight-line interpolation between (1,90) and (3,70) gives y = −10x +100. Setting y =74 gives x =2.6 minutes.
The correct conclusion is ‘the linear model estimates 2.6 minutes’, not ‘the student definitely returned to baseline at 2.6 minutes’. The real pulse pattern between readings may not be linear. The distinction between estimate and observation matters especially when converting analysis into advice.
The eighth complete worked problem
Question: A pupil finds two equal average gradients over intervals of different lengths and says the total decrease must be the same. Explain why this is false. If the gradient is −10 (beats/min)/min for one minute, the change is −10 beats/min. Over three minutes, the same gradient gives −30 beats/min. The total change equals gradient multiplied by duration.
Use an everyday analogy: a vehicle travelling at a constant speed for longer covers a greater distance. Here the vertical quantity is a rate rather than distance, but the relationship between constant gradient, interval duration and total vertical change is the same.
The ninth complete worked problem
Question: In a comparison, Person A’s rate falls from 160 to 100 and B’s from 140 to 90 over the same three minutes. The absolute declines are 60 and 50 beats/min. The relative declines from their respective starting values are 60/160 ×100% =37.5% and 50/140 ×100% ≈35.7%. A has a slightly larger relative decline in this window, but neither calculation by itself establishes superior health or fitness.
Finish by writing the chosen metric in the conclusion. An examination answer that says only ‘A is better’ has lost the object of comparison. What is better? The larger absolute fall, the larger relative fall, the smaller reading, or the smaller distance to baseline? The problem’s wording decides.
The tenth complete worked problem
Question: One sample has readings 82, 86, 90, 94 and 98 beats/min; another has 88, 89, 90, 91 and 92. Both means are 90. The ranges are 16 and 4 respectively. The second has less spread as measured by range. With further calculation, standard deviations can offer a fuller comparison. Do not say the second sample is ‘more accurate’; spread concerns consistency, while accuracy concerns closeness to a trustworthy reference.
This distinction transfers to every practical subject. Repeated readings that are similar may still be systematically wrong because of calibration or method. Less variation is not proof that a result is closer to the truth.
Time pressure: a ten-minute solution routine
First minute: underline the time origin, units and reference baseline. Second minute: write the relevant values in a small table. Third and fourth: determine whether the task asks for a change, a gradient, a mean, a percentage or a comparison. Fifth through seventh: perform the operation with working and units. Eighth: reverse-check a value or test whether the answer lies in a plausible range. Ninth and tenth: write a precise sentence addressing the requested claim.
The routine prevents a common failure: investing all the time in arithmetic and leaving no conclusion. In the official Paper 2 real-world task, using a number to support an explanation is part of mathematical application. The answer should remain concise, but it cannot be context-free.
Study repair for students coming from PSLE
If fractions and ratios are still unreliable, practise one question at a time using the PSLE Mathematics word-problem guide and the Primary 6 Mathematics hub. Recover the meaning of ‘part divided by whole’ before attempting repeated percentage comparisons. If signed numbers cause errors, revisit why final minus initial may be negative. If graph scales are fragile, practise reading coordinates before interpreting line shape.
Earlier Volume 0664 dealt with respirometer calibration and rate normalisation. Use it if you cannot distinguish observed movement from computed volume. The current chapter instead asks whether an uneven sampling schedule changes how you calculate gradients and averages. Repair the missing layer rather than restarting the entire series.
A final examination-style challenge
Use the fictional A and B dataset from the beginning. Calculate A’s whole-period average gradient, B’s gradient during one to three minutes, each person’s gap above baseline at five minutes and the percentage that A’s immediate reading exceeds its resting value. Then explain why the unweighted mean of four spot rates is not necessarily a time average. Answers: A’s whole-period gradient is −12.8 (beats/min)/min; B’s one-to-three-minute gradient is (94−130)/2=−18; five-minute gaps are eight and four beats/min; A’s immediate value is 100% above baseline.
For the explanation, say that the samples are unequally spaced and do not measure all values between the samples. Finish with a caution: numerical evidence supports a mathematical description of the supplied readings. It cannot alone identify a physiological cause or diagnose an individual’s health. This is how the next English chapter will treat claims attached to graphs.
Transfer task: evaluate a public infographic
A fictional poster announces, ‘Pulse recovery improves by 50% in five minutes.’ Before accepting the headline, ask what ‘recovery’ means: reduction from peak, distance remaining above baseline or fraction of a fixed recovery range? A percentage without a named numerator and denominator can sound impressive while hiding the calculation. Ask whose measurements were used and whether every participant was observed on the same schedule.
A precise replacement might be: ‘In our four recorded post-activity readings, the mean pulse rate at five minutes was lower than the immediate reading for each participant shown.’ Even that statement should be checked against the actual numbers. In the final EMS chapter, this translation from arithmetic to public wording will decide whether the poster is released.
Why correct working matters more than a shortcut
Written work acts as a visible argument. A formula such as gradient = change in rate / elapsed time shows what is divided by what. A percentage formula shows the base of comparison. A labelled graph shows which variable is measured at each point. Without that structure, a correct calculator result can be misused or cannot be checked if a copied number is wrong.
The official Mathematics syllabus notes that omission of essential working can result in lost marks. Preserve the key setup, substitution, units and conclusion, but do not turn the solution into a story. The purpose of working is to make the method inspectable.
Sources, sequence and limits
Examination facts are drawn from the SEC overview, 2027 G3 subject list and G3 Mathematics K310 document. English Language is K300 and Biology is K325. English, Mathematics and Biology remain separate assessed subjects; ‘EMS’ in this series names a learning connection, not a combined SEC paper.
All calculations, graphs, participants and scenarios in this chapter are original teaching examples. For the scientific meaning of recovery, return to Volume 0667. For the risk of overclaiming from a mathematical output, revisit Volume 0666. The next chapter reads a health-advice passage critically, then the final chapter will bring all three subjects into one evidence decision.
