Begin with a question you want to answer. “What happened?” is different from “Why did it happen?” and “Will it happen again?” A record may help with one question while leaving another open. A good word helps you keep that boundary visible.

For example, a readout tells you what an instrument displayed. Calibration concerns how measurement relates to a reference. Knowing the first does not supply the second. Likewise, three copies of one account are different from three accounts with independent evidence.

The table gives the senses used in this guide. Primary and secondary have other meanings, including in schooling and colour descriptions. Here they describe sources. Use the surrounding sentence to identify the intended relationship.

Try a useful sentence instead of an impressive one: “There is a discrepancy between these readings; we need to check how each was obtained.” It names an observation and a next question without inventing an explanation. You can speak plainly while being precise.

These words also help you describe your own limits. If you have seen only a short extract, say so. If you cannot establish who created a record, keep its authorship uncertain. The aim is an account another reader can follow and check.

Imagine you are writing a short history of a neighbourhood mural. Three local websites say that one artist painted it alone. At first, three matching accounts look persuasive. Then you find that all three cite the same exhibition brochure.

The websites may accurately repeat the brochure. However, they do not provide three separate observations of the painting process. Their agreement has a shared origin. Ask about the independence of the evidence before treating the repetition as fresh corroboration.

In this invented example, you later find a work diary written during the project and a photograph showing several people painting. These are relevant primary materials for investigating the work. A later article interpreting them is a secondary account. Neither label settles every question about reliability.

The Library of Congress guidance on primary sources asks readers to consider point of view, historical context and comparison with other sources. Those questions give a useful starting point for examining a record. They do not make an original document automatically complete or correct.

Check authorship and purpose. Was the brochure celebrating the designer, documenting every painter or advertising an exhibition? Its omission of assistants may reflect its purpose, an error or something else. You need evidence before assigning a motive.

Now repair the claim: “The brochure names the artist, while a project photograph shows other people painting too. We still need to establish their roles.” That account keeps the useful information and exposes the unresolved question. It is more informative than either “the brochure is useless” or “three websites prove it”.

For a small practice task, draw arrows from each later account to the material it uses. If every arrow reaches the same document, look for a genuinely additional record rather than another repetition of the sentence.

In a fictional makers’ group, two scales weigh the same box of materials. One readout shows 500 grams and the other shows 520 grams. The immediate fact is a 20-gram discrepancy. The readings alone do not tell you which scale is closer to a dependable reference value.

First check whether the comparison was made in the same way. Was the same box used? Was packaging included both times? Did someone accidentally count a tray? Were both readings in grams? A difference in procedure can matter before you investigate the instruments themselves.

If the conditions match, a calibration record may help. NIST’s measurement handbook describes calibration in relation to reference standards or a designated measurement process. This is technical measurement guidance. The everyday question is simpler: “What reference supports this reading, and under what conditions?”

Do not settle the disagreement by choosing the more attractive display. Extra digits do not supply a missing check. Nor does repeating the same reading establish that it is correct: a device could return a consistent result while differing from the reference.

Imagine that a supervisor checks both scales using suitable reference masses and records the results. That gives you more information than appearance or confidence in a brand. Read what was actually checked. A check at one value does not, by itself, describe performance at every possible load.

Your role in this vocabulary exercise is to ask and explain, not to carry out a professional calibration. Write the strongest statement the available information permits: “These scales disagree under the stated conditions; we have not yet established the reason.” Replace it when the checks give you a better account.

Consider this invented record of a large jigsaw being assembled during five club sessions. The cumulative number of pieces joined by the end of each session is 10, 18, 24, 24 and 30. The record counts pieces correctly joined and assumes none were removed.

The running total ends higher than it began. Yet the additions in sessions two to five are 8, 6, 0 and 6 pieces. A rising total does not mean the group is adding more pieces at every session. Cumulative names what the numbers represent, so you can ask the right next question.

The trajectory describes the course across the sessions. The unchanged total between sessions three and four creates a short plateau in this record. It does not establish that progress has stopped permanently, or that the group did no useful work during session four.

Perhaps they sorted pieces, studied the box illustration, discussed a difficult area or had less time. Those are possible explanations, not facts supplied by the totals. Check a session note before turning the graph into a story about effort.

The time scale matters in other settings too. NOAA distinguishes short-term weather from climate patterns over much longer periods, commonly described using records spanning decades. A single day cannot describe that full pattern. This illustrates why the observation period must fit the question; a five-session jigsaw record is not a climate study.

Ask two questions when you see a chart: “What does each point count?” and “What period does the claim cover?” If a caption says the club is becoming steadily faster, the invented totals are insufficient. Session lengths and comparable work would matter as well as the pieces added.

Imagine an optional questionnaire for a new drawing club. Twelve members answer after the first session. Later, the organiser asks the same twelve people again, but only seven reply. A respondent is someone who provides an answer; the invitation list and the response list are different things.

The five missing follow-up responses create attrition in the group being followed. We do not yet know why those people did not reply. Some may be busy, some may have stopped attending, and others may simply have missed the message. Do not replace missing answers with a convenient guess.

Suppose the seven later respondents all say the drawing instructions are clear. You can report exactly that. You cannot conclude from those seven answers that all twelve original respondents now find the instructions clear, or that each person has improved their view.

The change in who answers matters. To describe changes in individual views, you need to connect each person’s earlier and later answer appropriately. To describe the whole membership, you need to consider whose views remain absent. These are different questions, even when both use the same questionnaire.

This example is invented to expose the reasoning. It does not provide a general rule that every smaller follow-up group gives a false result. The honest response is to identify the missing information and narrow the conclusion accordingly.

Try repairing this sentence: “Everyone now understands the instructions.” A supported version is: “All seven people who answered the follow-up said the instructions were clear; five earlier respondents did not answer.” The revised sentence is longer because the missing group changes what the reader can infer.

Read this original fictional passage:

A notice says a small museum’s model railway was “rebuilt in 2018”. An old maintenance notebook records new rails and wiring that year. A volunteer recalls that the buildings on the model were made decades earlier. Sam writes, “Apparently the records contradict one another: the entire railway was created in 2018.”

Sam has widened the notice’s claim. The word “rebuilt” does not necessarily mean every component was newly created. The notebook describes particular work, while the volunteer describes the buildings. Both accounts could be compatible with a substantial rebuilding project.

Use apparently carefully. It signals how something seems from available information, but it cannot repair a claim that ignores what the records actually say. A cautious-sounding word still needs a clear basis.

To triangulate the question, compare relevant evidence from different sources or methods. Here you might examine the dated work notebook, ask the volunteer about the basis of the recollection and look for project photographs. The task is to investigate which components changed, not to count which version receives the most mentions.

A better account is: “The railway was described as rebuilt in 2018. The maintenance record identifies new rails and wiring, while the buildings may be older. We need further evidence about what the rebuilding included.” It preserves agreement where possible and uncertainty where necessary.

Now change the question. If you only need to know when the rails were replaced, the notebook may be particularly relevant. If you need to identify the designer of the original buildings, it may be insufficient. Evidence quality depends partly on its connection to the question you are asking.

Notice the movement from knowledge to understanding: you began with three separate statements and ended with a more precise relationship between them. The new vocabulary helps organise that relationship. It does not remove the need to inspect the actual material.

Choose three or four words from one part of the bank. Start with a real, low-stakes question about a hobby, a description or a set of observations. You can also reuse one of the fictional examples. The following routine is an editorial suggestion for practice.

1. State the question. Keep it narrow enough to answer with the available material.
2. Name what you have. Identify the record, observation or measurement.
3. Make the connection. Explain how it supports your account.
4. Mark the limit. Say what is missing or what the information cannot establish.
5. Choose a useful next check. Look for evidence that could distinguish the remaining explanations.

For the scales, the next useful check concerns measurement conditions and references. For the museum, it concerns which components were replaced. The same word discrepancy can begin both inquiries, but it does not imply that the same check will resolve them.

Return later with a new case. Explain the terms without looking at the table, then check your wording and repair any confusion. If cumulative becomes “the amount added today”, use two consecutive totals to recover the distinction. If corroboration becomes “complete proof”, look again at what additional evidence does and does not support.

You can speak, annotate a short extract, sketch a sequence or write a paragraph. Keep the format useful to the question. A parent or friend can ask, “Which part can I check?” without supplying the answer for you.

Do not turn every everyday remark into a lengthy investigation. Match the depth of checking to the importance and uncertainty of the claim. Sometimes the useful outcome is simply a clear sentence saying what you know and what remains unresolved.

Must a fourteen-year-old master every entry?

No. The bank provides possible targets within this editorial sequence. Familiar words can support richer explanations, and unfamiliar technical words can wait until a useful example makes them meaningful.

Are original sources always better than later accounts?

No. Ask what each source can contribute to your question. An original record may be incomplete, while a careful later account may compare several records. Check the evidence and reasoning rather than relying on a label alone.

Does uncertainty mean we know nothing?

No. Preserve what the evidence supports. You can know that two readings differ without knowing why, or know which people answered without knowing the views of those who did not.

Can I use the words in ordinary conversation?

Yes, when they make the point clearer. “These are cumulative totals” may prevent confusion about a chart. If the listener does not know the term, explain it in a short familiar phrase.

How can someone help me practise?

Ask them to check one connection: the term’s meaning, the evidence behind a sentence or the limit of a conclusion. Revise that part together. Understanding is easier to discuss when the example is concrete.

Why are earlier words still used here?

Only the opening banks avoid repeated headwords. Explanations depend on familiar language, and practice should revisit earlier words whenever they help you make a new connection.