Two accounts may share an underlying source; a missing record is not automatically a missing event.
archival silence
A gap in surviving or collected records that limits historical reconstruction.
independence
Whether evidence originates separately rather than repeating the same observation.
01 · TRY
Do three printed reports give three independent witnesses?
Draw arrows showing which account depends on another. State a narrow conclusion, an unresolved numerical question, and the next independent evidence you would seek.
Color-code shared source origins and explain one omission in a sentence.
Make a provisional claim
Loading saved evidence…
Inspect evidence and test an explanation
The investigation controls appear when your local evidence is ready.
04 · UNDERSTAND
Connect evidence and mechanism
Corroboration without counting copies
Do three printed reports give three independent witnesses?
Context and mechanism
Corroboration compares sources, but independence matters more than a simple count. Three newspapers that reproduce the same dispatch supply evidence that a claim circulated widely. They do not supply three independent observations of the event. Trace where each statement originated, and distinguish common wording from agreement reached independently.
Archives are shaped by power, literacy, preservation, collection decisions and chance. An official register may record taxable property while excluding unpaid work. Its silence about a kind of labor tells us about its scope before it tells us that the labor did not exist. Absence can become evidence when we know a record-making system normally documented the activity and survives adequately.
Seek sources with different positions and purposes, without assuming that every perspective has equal factual support. A private letter may challenge a public report, but it also has a limited viewpoint. An honest conclusion may identify a credible event, uncertainty about its scale, and groups whose experience cannot be reconstructed from this packet.
Work through the evidence
Fictional reports A, B and C describe 80 participants. B identifies A as its source; C reproduces B’s wording. A separate gatekeeper notebook records 50 entries but excludes a second entrance. There is one report chain and one partial count. Neither ‘three witnesses prove 80’ nor ‘the notebook proves only 50’ follows.
Your investigation
Draw arrows showing which account depends on another. State a narrow conclusion, an unresolved numerical question, and the next independent evidence you would seek.
Sources and their limits
Report chain
A reports 80 participants; B cites A; C copies B.
Fictional dependency structure: repeated publication is not independent observation.
Gate notebook
50 entries were counted at the north gate; the south gate was unobserved.
Fictional partial count; people may use another gate or re-enter.
Extend and revise
High school: Color-code shared source origins and explain one omission in a sentence.
AP practice: Use corroboration and a sourcing limitation to qualify a historical argument.
Introductory college: Create a provenance graph and discuss survival bias in the collection.
Describe an archival silence without inventing testimony to fill it.
Investigate preservation before estimating an outcome
What survived is not always what happened
A complete fictional intake register lists twelve distinct petitions. Each eventually received exactly one grant or refusal. Only some outcome records survive. The register establishes the twelve-case denominator; it does not reveal missing outcomes. These are invented records, not a medieval dataset.
Predict the surviving-record rate and the widest possible full-register range before changing the scenario. Each scenario filters the same twelve invented cases; it does not change the underlying events.
Petition 1granted
Petition 2refused
Petition 3granted
Petition 4refused
Petition 5granted
Petition 6Outcome missing
Petition 7granted
Petition 8Outcome missing
Petition 9granted
Petition 10Outcome missing
Petition 11granted
Petition 12Outcome missing
8 outcome records survive; 6 show a grant; 4 outcomes are missing.
Among surviving records: 6/8 = 75.0% granted.
Using only the displayed outcomes, among all twelve registered petitions: at least 6/12 (50.0%) and at most 10/12 (83.3%) could have been granted.
The lower bound treats every missing outcome as refused; the upper bound treats every missing outcome as granted. These are logical bounds under the stated exhaustive register and binary-outcome assumptions, not a statistical confidence interval or a probability distribution.
Inspect how this scenario was constructed
These notes reveal additional teaching information. The displayed bounds use the outcome cards alone, before this extra information is supplied.
The display was constructed to retain all six granted outcomes and only two refusals. Its 75% rate therefore overstates the constructed full-register rate. Real preservation mechanisms need evidence; a researcher cannot assume this rule merely because records are missing.
Controls are temporary and reset on reload. Save your scenario, calculation and qualified claim in the source notebook; captures do not save this selector. There are no timers or background work.
Keep three questions separate
What was registered? The complete intake register fixes twelve petitions in this constructed cohort. It does not count every possible grievance or every person who might have wanted to petition.
What survives? The selected display contains eight outcome records, including six grants. Its denominator is eight, not twelve.
What can be inferred? Four unobserved outcomes leave the full-cohort rate between 6/12 and 10/12. A point estimate of 75% requires assumptions beyond simply counting what survives.
Write a caption for each scenario before opening the complete register. Include the surviving count, the cohort denominator and a limitation. Do not treat the width of the range as an estimate of how likely an endpoint is. All assignments of missing outcomes consistent with the stated record are possible here; they are not necessarily equally probable.
Transfer to a different archive
A fictional intake list registers ten distinct petitions. Seven outcome records survive: two grants and five refusals. Calculate the surviving-record rate and full-register bounds. Then consider a new complication: the intake list itself has missing pages, so ten is only the known minimum number of petitions. Explain which earlier denominator-based statement can no longer be defended.
Review the unfamiliar archive
With a complete ten-case intake list, the observed grant rate is 2/7, about 28.6%. The full-register bounds are 2/10 to 5/10, or 20–50%. Missing outcomes are not automatically refusals.
If the intake list is incomplete and the total number is unbounded by the supplied evidence, the 20–50% statement no longer bounds the entire historical cohort. It still describes the ten known cases under the original assumptions. Additional unknown petitions could change both numerator and denominator; seek another intake source or defensible population bound. Do not simply reuse ten because it is the only count available.
Connect a rule to practice
For a charter or statute, compare three sources: the written rule, a complete intake list, and individual decisions with outcomes. A favorable decision is not automatically proof that the requested remedy was delivered. Add an implementation receipt or later correspondence if your claim concerns actual redress. Neither a grant rate nor a surviving complaint alone establishes equal access, fair treatment or the rule’s causal effect.
Save a claim that names the population it covers. For AP practice, explain how the archive’s purpose and survival affect the argument. For college work, specify how you would search for contradictory records and test completeness without inventing missing testimony. The charter investigation supplies a separate historical example of why written restraints and enforcement must be distinguished.
Worked example
Fictional reports A, B and C describe 80 participants. B identifies A as its source; C reproduces B's wording. A separate gatekeeper notebook records 50 entries but excludes a second entrance. There is one report chain and one partial count. Neither 'three witnesses prove 80' nor 'the notebook proves only 50' follows.
