A Primary 6 Science pupil in Clementi opens an online worksheet about plant growth. Across the top are three neat headings: dark cupboard, window and lamp. Down the side are days one, two and three. The numbers appear perfectly aligned. A sighted classmate can glance at the grid and compare the values. A pupil using a screen reader moves through the same table and hears a series of numbers without dependable information about which day or light condition each belongs to. The worksheet exists, but part of its meaning has gone missing.
Why do screen readers sometimes read tables incorrectly? What makes a table accessible for students? Can merged cells, copied spreadsheets and image-only results tables make a correct answer harder to reach? These questions matter in Mathematics, Science, geography, school reports and everyday learning. Tables compress relationships into position. The challenge is to make those relationships available to learners who do not receive, or cannot efficiently use, the same visual layout.
A row of numbers is not yet an explanation. A table cell gets its meaning from its row header, column header, units, grouping and sometimes a footnote. If the relationships are visible but not represented in a form that assistive technology can discover, the learner may receive the right digits attached to the wrong categories. The repair is not to make the table merely prettier; it is to preserve its structure and give students a reliable path from value to meaning.
The 50-second answer
An accessible data table makes the relationship between each value and its relevant row and column headings understandable to the learner. On the web, this normally means real table structure with header cells identified programmatically, rather than spacing text into columns or presenting a screenshot of a table. Simple tables can use clear row and column headers; complex tables with grouped or merged headings may need more explicit associations. W3C’s Web Accessibility Initiative provides detailed guidance, but good teaching also asks whether the table is unnecessarily complicated in the first place.
Students should read a value with its complete address: category, condition, time and unit. Parents can help by asking what a number describes instead of treating the child as weak at calculations. Teachers should test representative tables using keyboard and assistive technology, offer accessible alternate representations when needed, and check the final exported version—not just the original spreadsheet. A correct calculation cannot rescue a value taken from the wrong row or column.
A table is a miniature coordinate system
On paper, the position of a value is meaningful. In a simple table with days as rows and experimental conditions as columns, the cell where Day 2 meets Window refers to the window group’s Day 2 measurement. The reader silently performs two lookups: which row, and which column? Sometimes a third lookup is needed for units or a group heading. The grid makes that work appear effortless for someone who can see the spatial layout.
A screen reader does not automatically inherit all those visual shortcuts. The underlying document must expose an appropriate reading sequence and the semantic relationships among headers and data. In an HTML table, a properly defined heading can help an assistive technology user hear or inspect the relevant label while moving among cells. In a badly constructed table, the same user may hear isolated values or misleading labels. The experience depends on the browser, assistive technology and document construction, which is why no single screenshot or automated check can certify every reader’s experience.
The learner’s problem is not “reading more slowly”. It may be that the relationship was never encoded in the version they received. This is a design and accessibility issue. Once the source provides the right associations, the student can spend attention on the intended Science or Mathematics rather than reconstructing layout from unrelated fragments.
Begin with a four-value example
Imagine this fictional results table for a classroom investigation. The rows are Day 1 and Day 2; the columns are No Added Light and Lamp. The measured quantity is height of seedlings in centimetres. The values are 2 and 3 on Day 1, then 2.5 and 4 on Day 2. Visually, a teacher may arrange them as a tidy grid. The question is, “How much did the lamp group’s height increase?” The intended operation is 4 minus 3, producing 1 centimetre in this invented dataset.
But what if the learner hears only “2, 3, 2.5, 4” with no labels? They must remember an earlier description of the table layout and infer the position of each value. This introduces a second task beyond interpreting the experiment. If the reader assumes the first two numbers belong to one column rather than one row, they may calculate 2.5 minus 2 or 4 minus 2.5 instead. The arithmetic can be performed correctly on the wrong pair.
A strong teaching response is to read each cell as a complete relationship: “Day 1, lamp, height 3 centimetres”; “Day 2, lamp, height 4 centimetres.” Now the comparison is inspectable. That verbal format is also a useful check for sighted pupils who routinely jump between rows and columns too quickly.
A visual border is not a semantic heading
Teachers often make worksheets in spreadsheets or word processors. They merge a few cells, fill the header row with colour and centre the text. The result may look excellent. Yet visual emphasis is not the same as a programmatically identified header. On the web, a table built from plain paragraphs separated by tabs can look grid-like while offering no machine-readable table relationships. A header written in bold is still just bold text unless the document structure represents its role.
W3C’s tables tutorial explains that accessible HTML tables distinguish header cells from data cells and convey their relationship. For simple tables, a marked-up header row and labelled row headers may be enough. For more complex groupings, authors may need scope or explicit id–headers associations. These are technical mechanisms, but the teaching principle is approachable: the student should be able to ask of any number, “What is this value a value of?” and receive a reliable answer.
Designers should not convert a simple table into a complex arrangement merely to show off formatting. The first accessibility improvement may be removing a merged cell, making units explicit in each relevant heading and keeping each row about one observation. Structural simplicity can help every reader, not only those using assistive technology.
The merged-heading trap
Consider a table showing two years of exam practice results, each year divided into English, Mathematics and Science. A designer places “2025” across three merged columns and “2026” across another three, with subject names beneath. Sighted readers can use the large group headings to interpret an English score as belonging to a particular year. A screen reader may need an explicit association between each score, its subject and its year.
Suppose the table contains 68 under English 2025 and 74 under English 2026. If the year association is lost, the learner may hear “English, 68” followed later by “English, 74” with no clue about chronology. That changes the task. The intended question is whether performance changed between years. The available information may look like two unlabeled samples instead.
An accessible design could present two simple tables, one for each year, or use properly structured multi-level headers. Splitting the data is not always best; some comparisons benefit from one table. The decision should be made by the meaning students need to extract. The goal is not one universally correct HTML pattern. The goal is preserving associations under the intended access modes.
The difference between reading order and header association
Two separate problems are sometimes confused. Reading order concerns the sequence in which content is encountered. Header association concerns the labels that give a cell meaning. A table may be traversed in a sensible row-by-row sequence yet still omit the headings required to interpret values. Another table may expose correct headings but contain footnotes or captions placed somewhere confusing in the reading sequence.
To test reading order, a teacher can navigate the exported resource without using the mouse and listen to how content is announced through a suitable screen reader. To test associations, select representative cells: first data row, last data row, a cell under a group header, a cell with a footnote and a cell near a blank. Can the reader identify the cell’s complete meaning without memorising the whole visual grid? If not, something important may be missing.
This check is more informative than a broad claim that “the PDF is accessible”. Accessibility includes many requirements, and one can pass while another fails. A page may have selectable text but poor table structure. A table may be well tagged but require inaccessible colour cues. The actual access task should guide the diagnosis.
Science: when a number belongs to the wrong condition
In a science lesson about heat, imagine a table where each material is tested twice. Columns show initial temperature, final temperature and change. A student is asked which covering reduced temperature loss most in the example. If the header “Final temperature” is accidentally associated with the “Change” column, the student may interpret a small change value as a final temperature. The resulting conclusion can be perfectly logical from the incorrect label and still scientifically wrong.
Teachers should protect the conditions and units as part of the table meaning. “Temperature” is less informative than “Final water temperature (°C) after 10 minutes.” The wording need not be that long in every cell, but the relationship should be recoverable. Students can practise reading a complete record aloud: “Material B, trial 2, final temperature 42 degrees Celsius.” Then ask whether the proposed conclusion uses initial, final or changed values. That check connects data reading to scientific reasoning.
For accessibility, a textual version can be provided as one record per line if the table structure cannot be repaired immediately. This is a temporary equivalent route when accurately written; it should not be a pretext to leave important digital learning material permanently inaccessible.
Mathematics: the hidden denominator in the header
Imagine a percentage table comparing attendance for two groups. The value 80% may mean 80% of enrolled pupils attended, or 80% of pupils surveyed preferred an activity. The denominator is not contained in the digits 80. It lives in the heading, method and perhaps a note under the table. A screen reader that announces only “80 per cent” has left the student without the population definition required to interpret the result.
An apparently simple comparison—80% against 75%—may therefore be ill-posed if the denominators differ. Students should ask: eighty per cent of what, collected when and under which condition? This skill matters for data literacy even on paper. The accessibility perspective reveals why the labels must travel with the values rather than sitting somewhere the student may not be able to discover.
A useful calculation activity begins only after everyone writes the units and denominator in words. If 24 of 30 students completed a task, the completion percentage is 80%. If 20 of 25 completed a second task, it is also 80%, although the counts differ. The table’s context determines which conclusion is warranted. The correct use of a percentage sign does not replace the need for a complete row-and-column address.
English: tables are read as arguments too
Reading comprehension tasks sometimes include schedules, survey summaries, library borrowing records or comparison charts. Students may be asked to infer a pattern from information shown in a table. If the table’s row and column labels are obscured, the pupil’s apparent comprehension difficulty may be a format problem rather than a language problem.
Consider a school-library table listing genres by term. A question asks which genre became more popular in Term 2. The words fiction, biography and poetry are easy enough to recognise. But unless the values are securely associated with Term 1 and Term 2, the learner cannot make the comparison. A written summary such as “Biography loans rose from 12 in Term 1 to 19 in Term 2” may be accessible where an irregular grid is not. Yet the student must still interpret whether the change reflects popularity, loan availability or some other factor; the table alone cannot settle every explanation.
The point is not to spoon-feed an inference. It is to make the source information available so the inference is genuinely the learner’s task. A comprehension assessment should not accidentally become a test of whether a child can reverse-engineer a broken table export.
What about a scanned image of a table?
A scanned worksheet may display a perfectly readable table to some sighted users while providing no text or cell semantics to assistive technology. Optical character recognition may recover individual words and digits, but correct character recognition does not guarantee that each value remains attached to its intended header. A row may be read as a column; footnotes may be inserted mid-table; a multiplication sign may be lost.
The remedy depends on the source. Where possible, return to the editable original and export an accessible format with semantic table structure. If only a scan exists, specialist remediation may be required. A teacher can also provide an accurately transcribed alternative that includes row, column, units and notes. The essential test is whether the learner can reach the same task information, not whether the file technically contains a text layer.
This is why photographs and accessible tables are distinct topics. A homework photo can lose information through glare, blur and cropping. A neatly exported digital table can be visually pristine while losing structural meaning for assistive technology. Different failure mechanisms demand different repairs.
When a blank cell is not zero
A table may contain blank cells for many reasons: not measured, not applicable, unavailable, deliberately withheld, or a value yet to be entered. Zero is an actual numerical value. A dash may be a shorthand for no data, no event, or not applicable depending on the table’s legend. If an accessible version replaces blank entries with zero or omits their position entirely, it has changed the dataset.
Imagine a plant-growth table where the Day 3 observation for one group is missing because the sample was damaged. The blank is not evidence that the plant’s height became zero. For an accessible row description, say “Day 3, lamp group: measurement unavailable,” if that is the true record. Do not invent a number to maintain symmetry. Scientific interpretation depends on distinguishing absence of observation from an observation of absence.
This is a small but powerful data habit for all pupils. Before averaging, plotting or drawing conclusions, classify the values. Which are measured? Which are calculated? Which are missing? Which are labels? The table cannot do that reasoning for the student, but its representation should not make the classification impossible.
Keep captions and units attached to the correct table
Some documents contain several tables on one page. A short caption such as “Table 2: Plant height over three days” helps identify which data belong together. But when an exported document reorders the caption or detaches a unit note, a learner can mistake the note from Table 1 as applying to Table 2. The same issue can affect sighted students when print pagination splits a table across pages.
A strong resource gives each table a clear purpose, accessible caption and units in appropriate headers. It avoids abbreviations that are never defined. Where a group of columns shares a unit, the reader should still be able to discover that unit while inspecting an individual value. A heading “Temperature (°C)” should not become a hidden visual decoration when the table is exported or read through another mode.
Teachers preparing printable and digital versions can check both. The digital semantic structure and the printed pagination are not identical problems. A resource that works in one mode may need adjustments in the other. Accessibility is therefore a final-artifact property, not a promise made by the original design file.
A simple table-building rule for teachers
Start with the question students must answer. If it asks how an outcome changes across days under one condition, make time and condition explicit. Keep each row about one observation and each column about one consistent variable or attribute. Define units where the learner will need them. Give the table a descriptive caption. Mark header cells structurally, not just visually. If there are more than two levels of grouping, consider whether smaller tables or a textual explanation would be clearer.
When a complex table is necessary, follow W3C guidance for irregular or multi-level headers. This may require deliberate scope, id and headers relationships in HTML; in documents and spreadsheets, use the relevant application’s accessibility features and test the exported file. Do not assume a merged cell creates an accessible grouped heading on its own. Also remember that browser and assistive-technology combinations can behave differently, so representative user testing remains valuable.
The authoring rule is not “make every resource simple enough for a machine”. It is “make the intended relationships explicit enough that humans using different access modes can follow them”. Complexity belongs where the concept requires it, not where formatting habits introduced it.
Learner route: read every number with its address
A student using any access mode can develop a three-part table habit. First name the variable and unit. Second identify the row and column conditions. Third read the selected value as a complete sentence before calculating. “Day 2, lamp, 4 centimetres” is safer than “the fourth number”. Then check whether the proposed calculation compares values from the same variable and appropriate conditions.
If an assistive format hides the header, students should report the access problem clearly: “I can reach the value 4 but the reader does not announce whether it belongs to the window or lamp group.” This gives the teacher or accessibility team a concrete repair target. It is more useful than “the table is confusing”, although the latter is an understandable first reaction.
The learner’s job is not to become responsible for fixing inaccessible source code. They should receive an equivalent usable resource. Meanwhile, the table-address method remains valuable for independent checking once the information is accessible.
Parent route: diagnose the presentation before drilling arithmetic
Suppose a child can subtract accurately on a plain-number worksheet but repeatedly selects the wrong values from a table. More subtraction drills may not solve that problem. Ask the child to explain what a chosen value refers to. If they cannot identify the row, column, group and unit, investigate the table representation and the reading process. If the child uses assistive technology, request an accessible version instead of relying on the parent to read every table aloud indefinitely.
Parents can teach a useful everyday application at home. Compare two bus timetables, a grocery receipt or a weather forecast table. Ask what each number represents and what heading makes that meaning clear. Keep the activity light and accessible; the goal is to develop a portable habit, not to turn every family errand into an exam.
For schoolwork, parents should report a demonstrable access barrier with the file name, page, table caption and a concrete example of a misplaced or missing header. This allows a teacher to correct the actual resource rather than guessing at the child’s ability.
Teacher route: the five-cell audit
Choose five representative cells from a table: one in the first data row, one at a group boundary, one near a merged header, one with a unit or footnote and one from the final row. In the final digital format, navigate to each cell using the intended assistive technology where possible. Can a reader identify its data value, row label, column label, relevant group and unit? If the answer is no, adjust the structure or offer an accessible alternative.
Next, conduct a meaning audit. Ask a colleague to read the table without seeing your original spreadsheet. Can they answer the intended question? Do the labels imply a false comparison? Are blanks explained? Do captions remain attached? The assistive-technology audit tests the relationship encoding; the meaning audit tests whether those relationships support the educational purpose.
Do not use one success as proof of universal accessibility. Some table techniques are supported differently across tools. The right standard remains the applicable accessibility requirements plus real-world usability for the students who need the material.
A classroom exercise: rebuild the header map
Give students a small fictional table with deliberately removed headings, clearly labelled as a puzzle. Ask whether they can determine which number belongs to which experimental condition. Most will discover that the values alone are insufficient. Add the column headings but hide the row labels. Now students can identify conditions but not days. Add the rows and units. The question becomes solvable without inventing information.
Next, show two renderings of the same complete dataset: one a simple accessible list of statements, the other a well-structured table. Have students compare which is more efficient for identifying a trend, checking an individual observation and describing the experiment. They may prefer different formats for different jobs. That is a useful result: accessibility need not imply a single prescribed visual appearance.
A final transfer challenge uses a school timetable. Ask students to identify a lesson from day, period and subject. The same header-to-cell principle travels from Science to everyday planning. Students see that table literacy is not a special accommodation only; it is a general skill in interpreting structured information.
A caution about automatic accessibility checkers
Automated checks can detect certain structural problems: missing table headers, invalid associations or suspicious layout use. They cannot always determine whether a header is conceptually correct, whether a complex table should have been split, or whether a unit footnote remains intelligible in context. A technical pass can coexist with a confusing learning task.
The reverse is also possible. A table may look unusual but be semantically coherent in the intended technology. That is why authors should combine standards-based checks with task-based reading and human testing. A teacher’s understanding of the learning objective is essential: which comparison matters, which distinction could be lost and what false conclusion would that loss produce?
W3C’s guidance is a strong technical authority for web tables; it does not itself establish that a specific table layout causes a particular mark reduction in Singapore classrooms. The classroom examples here are reasoned illustrations designed to make possible barriers inspectable, not measurements from a research experiment.
Common failure modes and repairs
Failure one: a row of bold text is treated as a header. Repair by using structural header markup. Failure two: a screenshot replaces a usable table. Repair by providing semantic data or an accurately transcribed equivalent. Failure three: merged cells look clear but have no explicit scope. Repair by making group associations discoverable. Failure four: units exist only in a detached note. Repair by connecting them to the relevant values. Failure five: blanks silently become zeros. Repair by preserving missing-data meanings.
Failure six: the table reads in sequence but lacks context. Repair the header mapping, not only the reading order. Failure seven: the accessible version contains copied numbers with one row misaligned. Compare representative values with the authoritative source. Failure eight: only the edit file is tested. Inspect the exported and delivered version. Failure nine: a software audit replaces student testing. Add task-based checks with intended users. Failure ten: students are drilled in arithmetic when the data address is inaccessible. Repair access and then evaluate the intended academic skill.
Each failure has a distinctive symptom. The useful question is not merely whether the table passes a checklist. It is whether the learner can reliably reconstruct what a particular number is measuring.
The experiment that exposes a hidden table problem
A teacher need not begin with software terminology. Give three pupils the same compact Science table: one sees the original layout, one receives a plain text version in reading order, and one hears a partner read the cell values without announcing their headers. Ask a simple question, such as “Which material gained the most mass after soaking?” Then ask each pupil to explain how they identified the correct value. If the table is well designed, the header, unit and observation should remain connected in every version. If the listener hears “12, 18, 4, 9” without material names or units, the calculation may be trivial but the answer cannot be justified. That failure is in the representation, not necessarily the pupil’s Science knowledge.
For a concrete example, write a table with three columns: Material, Mass before soaking (g) and Mass after soaking (g). The first row reads Sponge, 12, 18; the second Cloth, 9, 13; the third Paper, 4, 7. The increase is six grams for sponge, four for cloth and three for paper. A pupil who hears “Sponge, twelve, eighteen” in the correct row context can compute six. A pupil who receives all the first-column names followed by all the masses, with no programmatic relationships, may have to memorise a long sequence and reconstruct which number belongs where. The task has changed from comparing mass to decoding a damaged map.
The lesson for a teacher is to test the route, not only the appearance. A screen reader might offer a way to move between headings and cells, but successful navigation depends on the document structure, the tool and the reader’s familiarity with it. A visual inspection cannot establish that every cell is associated with its label. Likewise, a validator that flags no errors cannot prove that the headings are meaningful. The strongest practical check combines a visual review, a structural check and, where possible, use with an assistive technology workflow familiar to the learner.
Multiple levels of headings: useful precision, avoidable maze
More complex school tables sometimes group measurements under shared headings. Imagine two experiments, Day 1 and Day 2, each with columns for Temperature and Time. A sighted reader can infer the hierarchy from merged header cells. A student hearing the table needs the equivalent relationship: “Day 2 — Temperature,” not merely “Temperature,” because the same child heading appears twice. In structured HTML, properly defined groups, scopes or header associations can supply that context. In a PDF or spreadsheet, the accessible route depends on how the document was authored and exported. The aim is not to decorate every table with extra markup. It is to ensure that a value’s full address survives different reading methods.
A useful simplification is to flatten a small complex table when doing so preserves meaning. Instead of a two-tier matrix, use columns named Day, Measurement, Value and Unit, then give each reading a separate row. This may make navigation easier, but it changes how comparisons are presented, so it should be tested against the intended question. The best layout is the one that lets the learner find the relevant comparison without carrying a reconstruction puzzle in working memory.
A parent-teacher conversation that starts with evidence
Suppose a child consistently answers verbally when an adult reads the worksheet but freezes when using the digital version alone. The first instinct may be to recommend more independent practice. That advice could help, but it could also miss an accessibility fault. Ask a narrower question: Can the child identify each column’s meaning without another person supplying it? Does the same problem occur in a plain three-column table? Is the confusing document a scanned picture, a structured page or a spreadsheet? Does the problem appear only when the screen is magnified or read aloud?
A parent can bring one example rather than a general accusation. “In this file, the reader announces 15 but does not make clear whether it means minutes or degrees Celsius” gives the teacher something to inspect. The teacher can compare the digital source, export and alternative presentation, then try a corrected version with the student. If the child’s answer improves, that is useful evidence that access was a contributing factor. It is not proof that every academic difficulty had the same cause. Classroom practice should continue to build content knowledge while making the presentation route dependable.
Where the accessibility evidence stops
Standards such as the W3C’s table tutorials and WCAG techniques describe how to encode relationships between headings and data. They establish a defensible technical expectation, not a guaranteed academic score increase for each corrected table. Evidence about working-memory demands makes the educational mechanism plausible: if a learner must reconstruct labels before reasoning, some attention is spent on navigation rather than the subject. But the size of that effect will differ by age, familiarity, task, device and access needs. It would be misleading to turn a design recommendation into a universal claim that one table format raises marks by a specific amount.
The practical test is proportionate. For a two-column table, make the headers explicit and test that they are announced. For a complex timetable or results grid, check more than one cell in different regions, including repeated headings and blank entries. If a student needs a personalised format, ask what actually works for that learner. Accessibility is not an ornamental feature added after content has been finished; it is one of the conditions that lets the content mean what its author intended.
FAQs
What makes an online data table accessible?
It needs understandable relationships between header cells and data cells, plus usable reading order, labels, units and any necessary notes. On the web, semantic table markup helps assistive technology provide this context. Complex groupings may need explicit associations.
Can a visually beautiful table be inaccessible?
Yes. Borders, colours and bold headings help some readers but do not necessarily encode semantic relationships for assistive technology. A screenshot can look perfect while containing no navigable cell structure at all.
Does every table need complicated HTML?
No. A simple table with one row of column headers may need only straightforward semantic markup. Complex grouped headings require more care. Simplifying the table is often preferable when it preserves the educational purpose.
Are merged cells always bad?
No. They can convey meaningful group relationships. Problems arise when those groups are not exposed accurately to the intended access methods or when the resulting table becomes unnecessarily difficult to understand.
Why not just read all the numbers aloud?
Because the student still needs to know which category, condition and unit belongs to each number. A sequence of digits without a header map is not an equivalent representation of the dataset.
Do screen readers all behave the same way?
No. Browser, document format, screen-reader settings and version can affect the experience. Standards-based markup improves the foundation, but testing representative tools and learner tasks remains useful.
What if the worksheet is a scanned PDF?
Ask for an accessible version, ideally from the editable source. OCR may recover words but not dependable table relationships. An accurately transcribed alternative can help while proper remediation is arranged.
Is this only relevant to students with visual impairments?
No. Explicit row and column labels also support students reading dense tables, using small screens, working with unfamiliar data or checking calculations. Accessibility improvements can benefit wider audiences, though different learners still have different needs.
Can a wrong header change a correct Science conclusion?
Yes. If a temperature change is labelled as final temperature, or a value is assigned to the wrong experimental condition, a student can reason correctly from incorrect information and reach the wrong conclusion.
Sources and research notes
W3C Web Accessibility Initiative, Tables Tutorial, explains table header and data cell relationships: https://www.w3.org/WAI/tutorials/tables/ . W3C, Tables with Multi-Level Headers, documents explicit associations for complex table designs: https://www.w3.org/WAI/tutorials/tables/multi-level/ . W3C, WCAG 2.2 Technique H43, gives an example method for id and headers mappings, while noting techniques are not the only possible route to conformance: https://www.w3.org/WAI/WCAG21/Techniques/html/H43 . W3C, WCAG 2.2 Failure F90, describes consequences of incorrect header-to-cell relationships: https://www.w3.org/WAI/WCAG22/Techniques/failures/F90 . CAST Universal Design for Learning Guidelines 3.0 (2024) provides principles for accessible perception and representation: https://udlguidelines.cast.org/representation/perception/ . The classroom datasets in this article are invented teaching examples, not reported experimental results.
For a separate issue concerning physical room for answers, read Why the Answer Box Changes the Answer. For a separate issue concerning camera capture quality, see Why a Homework Photo Can Be Harder to Read Than the Page. This article owns only the semantic mapping of values to table headers across access modes; it does not re-own table sorting, photographic legibility or worksheet answer-space design.
Takeaway: numbers need their addresses
The Clementi Science table looked ready to teach. Its grid was neat, its colours calm and its numbers correct. Yet a learner moving through it differently received values without the relationships that made them useful. Once those relationships are made explicit, the task returns to its rightful owner: the learner can compare, calculate, explain and question the experiment. A genuinely accessible table is not a concession that weakens academic challenge. It protects the challenge by delivering the data in a form the student can actually interpret. Every number deserves its correct address.