A timeline can show that one event preceded another. It cannot, by its arrangement alone, show that the first event caused the second. Historical explanation begins when we ask what changed, through which mechanism, under what conditions and with what evidence beyond the order of dates.
This workshop separates three tasks that are often compressed into one diagram: chronology, periodisation and causation. We use selected milestones from Singapore’s early MRT development to practise labelling events, then an entirely fictional reading-room case to examine competing causal explanations in detail. The historical and invented evidence are clearly separated.
You will build a timeline with honest date precision, compare two ways of dividing a history into periods, repair an overconfident causal paragraph and complete a changed-condition exercise. No specialised historical knowledge is needed. The main prerequisite is a willingness to preserve uncertainty instead of making the story look tidier than the evidence allows.
This guide belongs to the History and Heritage Library. Begin with What Is History? for the wider discipline. Here the reader’s job is to turn a sequence of events into an explanation whose strengths and limits remain visible.
Three questions hidden inside one timeline
Chronology asks what happened and in what order. Periodisation asks how to divide a span of time into meaningful stages. Causation asks how and why an outcome occurred. A timeline can help with all three, but each question needs a different kind of judgement.
Suppose an institution discusses a proposal, approves it, builds a facility and opens it. Those events can be placed in order. Calling the first stage “preparation” and the last “operation” is an interpretation of the process. Claiming that a particular debate caused the approval requires evidence about how the decision was reached.
The distinctions matter because a beautifully designed timeline can make interpretation look like a property of the dates themselves. An arrow may mean only “later than,” but readers may understand it as “therefore.” A large divider may suggest a complete break even when many practices continued across it.
Before drawing anything, write the question the timeline is meant to answer. A chronology of construction milestones will select different events from a history of passenger experience. Neither selection is automatically wrong, but each needs a stated purpose so that omissions can be assessed.
Start with an event record
For every entry, record a date or interval, an event description, the source and the kind of event. A discussion, recommendation, approval, start of construction, first public service and ceremonial launch are different events. Giving them all the label “opened” creates apparent contradictions that better wording may resolve.
Separate the date of an event from the date of its record. A speech delivered on one day may be catalogued much later. A retrospective institutional history may describe a decision made decades earlier. The later source date does not move the historical event forward in time.
Preserve the source’s precision. If you know only a year, enter a year. If the record gives an approximate interval, retain the interval. Do not convert “during the early 1980s” into a precise day merely to make a timeline application accept the entry. A sorting convention should not become a historical claim.
An event record can also include a confidence or verification note, but explain what the note means. “Date supported by an official event record” is more useful than an unexplained score of nine out of ten. It identifies the basis on which another reader can check your judgement.
A small Singapore MRT chronology
The following is a selected teaching chronology, not a complete history of Singapore rail transport. LTA’s The LTA Story, under its rail discussion, describes an early vision in 1967, a 1981 study and debate, and construction beginning in 1983. It is an institutional retrospective, rather than a contemporary transcript of every decision.
SMRT’s Group Review 2022/23, in its anniversary discussion, dates the first service between Toa Payoh and Yio Chu Kang to 7 November 1987. The National Archives of Singapore’s record of the official-opening speech identifies a distinct official-opening occasion on 12 March 1988.
| Date | Selected event | Event type and source |
|---|---|---|
| 1967 | Early vision for a rail system | Planning context; LTA retrospective |
| 1981 | Study and public debate about the transport choice | Analysis and debate; LTA retrospective |
| 1983 | Construction of the first rail line began | Implementation milestone; LTA retrospective |
| 7 November 1987 | First MRT service on the initial section | Operational milestone; SMRT retrospective |
| 12 March 1988 | Official-opening occasion | Ceremonial milestone; NAS audiovisual catalogue record |
The source types are part of the table’s meaning. An operator’s later review and an archive’s description of a recording serve different evidential roles. This guide uses the archive record’s identification of the occasion; it does not reproduce the recording or claim to have analysed every statement in the speech.
Why two opening dates need not conflict
The operational and ceremonial milestones refer to different events. A first service can precede an official launch. If two summaries use the word “opening” without explaining which event they mean, readers may infer a contradiction that is not actually present.
Repair the labels before choosing a winner. Ask whether one date concerns the first passenger service, another a formal ceremony and another the completion of a later phase. A precise description may allow all the dates to coexist in the same chronology.
This does not mean every disagreement can be solved by relabelling. Two sources may genuinely assign different dates to the same event. In that case, identify their source relationships, inspect the relevant records and preserve the disagreement until the evidence supports a resolution.
The useful habit is to compare complete claims: event, date, place and scope. Comparing isolated dates strips away the information that tells you whether they are answers to the same question. The same method applies to a school’s founding, a law’s approval and commencement, or a building’s completion and public opening.
A sequence is not yet an explanation
The selected rail milestones establish an order of planning, discussion, construction and public operation. They do not, on their own, measure the contribution of every argument, institution or person involved. A causal history would need more detailed evidence about decisions, resources, alternatives and implementation.
LTA’s account describes a choice between transport options and disagreement about cost. That helps identify a question for further inquiry: why did one option prevail? It does not justify replacing the historical investigation with “rail was inevitable because it eventually happened.” The outcome should not erase the alternatives considered before it.
Avoid describing every earlier event as a necessary step unless the evidence supports that claim. Some developments may have enabled the outcome, some accelerated it and some may have been incidental. Their position before the opening date does not tell us which relationship applies.
A careful timeline therefore uses modest labels such as “preceded,” “during” and “followed” until a stronger relationship is established. When you add a causal arrow, explain the proposed mechanism and the evidence that supports it. The arrow should summarise an argument rather than substitute for one.
Periodisation makes an argument about change
One way to organise the early rail chronology is by project stage: planning and debate, construction, then operation. This highlights a change in the main activity required to bring a system into use. It can help a reader see that a proposal, a building process and a running service require different kinds of work.
Another way is by the public’s relationship to the system: a discussed possibility, a visible construction project, then an available transport service. This highlights experience and access rather than the internal sequence of project delivery. The same dates support a different organising question.
Neither division is an official periodisation supplied by the sources here. They are teaching interpretations. Their value depends on what they help explain and what they leave out. A period boundary can be useful without implying that all activities changed at once or that every person experienced the transition similarly.
State your criterion explicitly. “I divide these events by the dominant project task” is clearer than “history naturally falls into three stages.” The first gives readers a basis for assessment. The second conceals the author’s organising choice behind a claim of inevitability.
Keep continuity visible across a boundary
Construction does not necessarily eliminate planning. Operation does not necessarily end construction, maintenance or debate. When a timeline moves into a new period, some earlier activities may continue and some later activities may already have begun. A single divider can hide this overlap.
Use a short continuity note beside each boundary. For a fictional institution, the move from a temporary room to a permanent building might change location while retaining staff, collections and users. Describing the move as a complete institutional rebirth would overstate the break unless the evidence supports it.
Likewise, a dramatic event may matter differently to different groups. An opening can immediately change access for nearby users while having little direct effect on people elsewhere. A national label should not erase uneven timing or experience within the period it describes.
The best periodisation is useful for a stated purpose and open to revision. If your question changes from infrastructure delivery to labour conditions or neighbourhood access, reconsider the boundaries. Reusing the same stages without checking their fit can make the new history answer the old question instead.
A fictional case with enough detail to test causes
We now leave the historical rail example. Every reading-room record, date and count in the following case is invented for teaching. The case is designed so that you can compare explanations without needing access to a real institution’s private records.
A small reading room reports rising visits during a proposed extension of its evening hours. Its organiser writes, “Our new hours caused the increase.” You receive a proposal minute, a publicity note, routine attendance records and a later review. Your task is to decide how much of that claim the packet supports.
The measure is visits, counted consistently at entry during the stated weeks. One person can make more than one visit. The records do not identify unique visitors, their reasons for attending or the time each spent in the room. These limits matter when interpreting a change in the count.
The organiser’s conclusion may be plausible, but plausibility is not the same as demonstration. Begin by reconstructing what happened and when. Then ask which part of the increase could be explained by the actual additional hours and which part requires another explanation.
Build the fictional chronology
| Event or observation date | Record | What it establishes in the teaching packet |
|---|---|---|
| 3 March 2024 | Meeting minute proposes an evening extension from 1 April | A plan existed; operation had not begun |
| 10 March 2024 | Publicity note records circulation of a leaflet | The proposed change was publicised |
| Week ending 17 March | Routine count records 40 visits | A weekly observation before the extension |
| 18 March 2024 | Partner newsletter mentions the reading room | Another publicity route appeared |
| Week ending 24 March | Routine count records 60 visits | Visits rose before the additional hours began |
| 1 April 2024 | Session log records the first extended evening | The operational change began |
| Week ending 7 April | Routine count records 80 visits | A higher weekly count after the change |
| 20 April 2024 | Receipt records a donation of books | A later collection change occurred |
| 5 May 2024 | Review attributes the increase to longer hours | A later interpretation was recorded |
The review date is not the date on which the increase began. The donation date is not evidence that the donated books caused an earlier rise. The proposed start date and the actual session log perform different roles, even though they align in this fictional packet.
The table also separates a count’s observation period from the day someone might enter it into a spreadsheet. A weekly total should retain its week. If you place it at an arbitrary single day without explanation, you can create a false impression of when the underlying visits occurred.
What the timing rules out
The additional evening had not begun when visits rose from 40 to 60. Its actual operation therefore cannot directly explain that initial increase through people attending the new session. This is a meaningful causal constraint supplied by chronology.
However, the announcement of the future hours could have influenced awareness or expectations before the first extended session. The publicity leaflet and partner newsletter also offer candidate explanations. These possibilities need evidence; they should not be treated as proven merely because they fit the order of events.
The later rise to 80 visits is compatible with the additional hours contributing. It is also compatible with continuing publicity effects, changes in the existing sessions or other unmeasured influences. The chronology narrows the claim but does not allocate the increase among those possibilities.
A better interim conclusion is: “Visits increased before and after the extension. The new hours may have contributed to the later count, but they cannot explain the earlier rise through their actual operation.” This preserves useful evidence while rejecting the organiser’s overly broad attribution.
Name the mechanism you propose
A mechanism explains how a cause could produce an outcome. For longer hours, one proposed mechanism is that people with daytime commitments gain a usable time to attend. For publicity, the mechanism is that more people learn the room exists or remember to visit. The two mechanisms make different evidence relevant.
If the new session attracts people who could not attend earlier, attendance by session and voluntary accounts of availability might help evaluate that explanation. If publicity increases visits to the original sessions, the pattern would support a different pathway. Neither observation alone necessarily settles the whole causal question.
Avoid replacing a mechanism with a label. “The programme worked” names a judgement but does not explain what changed. “Awareness increased” is a candidate mechanism only if you have a reason to think people encountered and acted on the information. Specify the steps that connect the proposed cause to the observed outcome.
The mechanism should also fit the timing. A book donation received later cannot explain an earlier increase through readers using those donated books. If someone proposes an anticipation effect, such as publicity about an expected donation, that is a new mechanism requiring its own evidence.
Compare explanations with a prediction table
| Candidate explanation | Pattern it would make relevant | Additional evidence to seek |
|---|---|---|
| Extra hours create usable access | Visits in the added session, including users unavailable at earlier times | Session-level counts and appropriately collected accounts |
| Publicity attracts attention | Increased visits following communication, possibly in existing sessions | Publicity timing and how visitors heard about the room |
| Visits shift from old sessions to the new one | New-session attendance rises while old-session attendance falls | Comparable counts across sessions |
| A counting change inflates the total | A break coinciding with a new counting rule | Definitions, collection instructions and duplicate handling |
The table does not pretend that each explanation is equally likely. It shows what evidence could distinguish them. You might eventually support several mechanisms at once, but listing them is not proof that each operated.
Also ask whether the evidence would be independent of the claim being tested. A review that repeats the organiser’s explanation is not the same as a separate attendance record or participant account. Preserve source relationships so that repetition does not appear to strengthen the causal case automatically.
Use comparison without inventing a perfect control
Suppose you obtain weekly counts from a second fictional reading room that did not extend its hours. A comparison might help identify changes affecting both places. If visits rose similarly in both, a wider influence becomes worth considering. If they diverged, you still need to examine whether the rooms were comparable.
The second room might serve different users, have different publicity or operate under different constraints. Calling it a “control” does not remove those differences. A comparison needs a reasoned account of what is shared, what differs and how those differences affect the inference.
Before-and-after evidence has a similar limitation. The same institution at two times is not automatically identical in every relevant condition. A timetable change can coincide with a new collection, a holiday or a different counting procedure. Preserve those possibilities rather than letting the calendar do all the explanatory work.
Historical inquiry often works with incomplete comparisons. The appropriate response is to make the inference proportionate and identify the evidence that would strengthen or weaken it. You can reach a useful qualified judgement without claiming that the surviving records simulate a controlled experiment.
Counterfactuals clarify the claim
A causal claim often implies a counterfactual: what would have happened if the proposed cause had been absent while relevant conditions were otherwise appropriately considered? In the reading-room case, the question is whether visits would have risen as much without the extended session.
We cannot directly observe that alternative history for the same room at the same time. We can use evidence and comparison to reason about it, but the counterfactual should not be written as an established event. “Without the extension, no one would have visited” is much stronger than the packet supports.
Distinguish necessity from contribution. A factor can contribute to an outcome without being necessary for any increase at all. Publicity and extended hours might each make some difference. Showing that visits rose before the extension undermines the claim that the extension was necessary for every increase, but it does not prove it had no later effect.
This distinction helps avoid all-or-nothing arguments. Evidence against a sole-cause claim may support a revised multi-cause explanation. The goal is to improve the causal account, not simply to move from “this caused everything” to “this caused nothing.”
Repair an overconfident historical paragraph
Here is an original flawed paragraph: “The reading room became successful when the hours changed. Visits rose from 40 to 80, proving that the extension doubled demand. The donated books also caused the rise. The May review confirms that every new visitor came because the organiser made the right decision.”
The paragraph changes visits into demand and then into unique new visitors. It compresses several observation periods into one before-and-after comparison. It assigns causal importance to a later donation without matching the timing. Finally, it treats the review’s attribution as independent confirmation of its own explanation.
A stronger paragraph is: “The recorded weekly visits increased across the period surrounding the extension, with an initial rise already visible before the extra session began. Publicity and other conditions therefore need consideration alongside the new hours. The later count is compatible with the extension contributing, but the packet does not show how visits were distributed across sessions or identify each visitor’s reason. The May review records the organiser’s interpretation; it does not by itself establish the size of the extension’s effect.”
The revised paragraph is not weaker because it has fewer dramatic claims. It is more useful because it distinguishes observation, candidate explanation and missing evidence. A reader can see what has been established and what the next inquiry should examine.
Draw a timeline that does not mislead
If events are placed at equal visual intervals, say that the display shows order rather than elapsed time. Otherwise, a long period of planning and a short interval between openings may look equally long. A scaled timeline should place dates according to a consistent time axis.
Use an interval or a labelled uncertainty band when an event is only broadly dated. Do not represent an approximate span as a single exact point without explanation. If two events occurred in the same year and their internal order is unknown, keep that uncertainty visible rather than arranging them for narrative convenience.
Separate event tracks when that helps. A record-creation track can sit beside an event track, or planning and operation can be distinguished in a table. The purpose is to show a relationship clearly, not to introduce so many colours and symbols that the explanation becomes difficult to follow.
Every visual choice should answer a question. A divider marks an analytical period; a line shows duration; an arrow indicates a specified relationship. If the symbol’s meaning is uncertain, replace it with a short label or use a table until the underlying claim is clear.
Practice questions
Try these before reading the commentary. Use the historical table for the first two questions and the fictional packet for the rest.
- Why can a first-service date and an official-opening date both be correct?
- What does the sequence from planning to construction establish, and what does it leave unresolved?
- Which fictional observation prevents actual operation of the new hours from directly explaining the entire rise?
- Does that observation prove that announcing the new hours had no effect?
- Why is the book-donation receipt insufficient to explain the earlier March increase?
- What is lost when a writer replaces visits with visitors?
- Propose one periodisation of the fictional case and state its organising criterion.
- Name one additional record that could help distinguish added access from redistribution between sessions.
For each causal answer, identify the claim’s scope. A response that disproves “sole cause of all growth” does not automatically disprove “contributed to later growth.” Precision about the claim prevents a correct observation from being used to support an excessive conclusion.
Answers with commentary
Question 1: The dates can describe different events: the start of public service and a later formal occasion. Compare the event definitions before treating the dates as contradictory. If two sources genuinely refer to the same event, further source comparison is needed.
Question 2: The sequence establishes temporal order within the selected milestones. It does not by itself establish the contribution of each decision, debate or organisation. A causal account requires evidence about mechanisms, alternatives and implementation.
Question 3: The weekly count had already risen before the first extended evening. People could not yet have attended an extra session that had not operated. This constrains the direct operational explanation of the initial rise.
Question 4: No. Publicity or anticipation could act before operation. That possibility needs evidence about communication and response. Distinguishing an announced proposal from an operating service keeps the mechanisms separate.
Question 5: The receipt records a later event. The donated books cannot directly explain earlier visits through their use before arrival. A separate claim about advance publicity would require evidence that such publicity existed and affected behaviour.
Question 6: Visits count occasions of attendance; visitors count people. One person may generate multiple visits. Replacing one measure with the other changes the claim and can exaggerate how many individuals were reached.
Question 7: One defensible division is proposal and publicity, initial operation, then review. The criterion is the dominant stage of the trial process. The answer should also acknowledge overlap, such as publicity continuing during operation.
Question 8: Comparable session-level counts before and after the change would help show whether the additional session added visits or shifted them from existing sessions. They would still not identify every person’s motive or remove all other influences.
Change the evidence and reconsider
Add a new fictional record: after the extension, total weekly visits are 80, but 30 occur in the new session and 50 in the original sessions. This tells us more about distribution. It does not establish that all 30 visits are additional, because some people may have attended an original session otherwise.
Now suppose the organiser discovers that the counting rule changed on 1 April from counting entries once per session to counting every re-entry. The before-and-after totals are no longer directly comparable without examining how much that change affected them. The apparent increase could combine behavioural change and measurement change.
Revise the conclusion rather than merely appending a footnote. The new evidence affects the meaning of the comparison itself. A responsible next step is to reconstruct comparable counts where possible and retain an explicit limitation where the records do not permit it.
This exercise shows why chronology should include changes in measurement as well as events that make an attractive story. A quiet administrative change can matter greatly to the interpretation of a dramatic-looking trend.
A delayed task and the next inquiry
Return later with a new fictional sequence: a school garden is proposed in January, announced in February, planted in April and featured in a June newsletter. Attendance at a gardening club rises in March. Write one causal explanation the timing rules out, one it leaves possible and one additional source you would seek.
A worked response rules out participation in the planted garden as a direct explanation of the March rise, because planting had not occurred. It leaves open an effect of the announcement or another influence. Club records and appropriately collected accounts of why members joined could help distinguish those possibilities. The June newsletter is a later account, not the date of the original change.
For source identity and record history, continue through How Archives Catalogues and Finding Aids Work. For the reasoning behind a claim, use Claims, Reasons and Counterexamples. Keep your timeline as an inspectable account of order, and let the causal explanation carry the additional evidence it needs.