To complete a function table from a graph, read the x-coordinate, then the y-coordinate, at each point on the plotted line or curve, and enter the pair in the row that matches that input. That sounds simple, and for a straight line it is. Two conditions decide how the work actually goes: whether the graph shows discrete plotted points or a continuous line, and whether you need only the marked points or the value of the function at inputs the graph never explicitly labels.
For a straight line through the origin like y = 2x, every row behaves identically. For a parabola such as y = xΒ², the same input can produce a different output on each side of the vertex, so symmetry becomes your main check. This guide walks through both, plus the failure cases β a table whose points all read the same sign, a graph where the function is undefined at a listed input, and the vertical line test that tells you whether a graph is a function at all.
Read a Point as an Ordered Pair Before Filling Anything
A coordinate pair is written (x, y), always in that order, and the order is the single most common source of wrong answers. The x-value is how far you move along the horizontal axis; the y-value is the height on the vertical axis. If a table lists 3 under the input column and your point sits at (3, 6), the output is 6 β never 3 flipped into position because the graph happens to pass near a gridline labeled 6 on the x-axis elsewhere.
Most textbook graphs plot the independent variable on the horizontal axis and the dependent variable on the vertical one, but you cannot assume it. Confirm the axis labels first. A graph that plots time on the y-axis and distance on the x-axis flips the convention, and filling the table with x and y swapped will produce a table that looks plausible and is entirely wrong.
Grid scale matters just as much as axis order. Count the subdivisions between labeled ticks before you read a single point. If one labeled tick spans two grid squares, a point sitting one square above a tick has a y-value of half a unit past that tick, not a full unit. State the scale explicitly in your own notes before starting.
Label each point with its coordinates as you find it, on the graph itself, using a pencil. Writing (2, 4) next to a dot takes three seconds and removes the risk of re-reading the wrong point later. This habit pays off most on continuous-line problems where you are generating points rather than reading dots that were drawn for you.
Decide Whether the Graph Wants Exact Points or Interpolated Values
A graph built from dots connected by segments is usually a discrete or sampled relationship, and the table should use the plotted dots. A graph drawn as a smooth curve represents a function defined everywhere on the interval, and the table row inputs are wherever they say β not just the labeled features. Read the instruction line and the axis labels, because the difference changes every cell.
When the table already lists its inputs, do not re-read the graph at whichever x-values seem convenient. Work the listed inputs in order. If the table says x = 0, 1, 2, 3, 4 and the graph labels only 0, 2, and 4, you are expected to interpolate the odd values off the curve itself.
For discrete graphs, the domain is exactly the set of plotted points, so an input with no dot attached is genuinely outside the table's scope. For continuous graphs, the domain is normally all real numbers unless the context restricts it, and a listed input with no visible dot still gets a value read from the curve. Mark the distinction at the top of your work.
Curves that flatten near the vertex or near an asymptote are the places where interpolation gets risky. Read those inputs off the line at higher zoom, or compute them from the rule you extracted, rather than eyeballing a shallow slope across several grid squares.
Produce a Value for Every Listed Input in Order
Use this sequence whenever the table supplies the inputs and leaves the outputs blank. The order matters because it catches scale and axis mistakes before they contaminate the whole table.
1: Write the input column from the table into the left column of your answer sheet, in the given order. 2: Write out the axis scale for both axes β units per grid square β before reading any point. 3: For each input, locate that x-value on the horizontal axis. 4: Follow a vertical line from that x-value to the graph; if several heights exist, the graph fails the function test and the table cannot be completed. 5: Read the height at the horizontal axis and convert grid squares to units using the scale you wrote in step 2. 6: Write the ordered pair (x, y) beside the graph and copy y into the matching output cell. 7: Re-check the finished table against the graph left to right, one row at a time.
Failure handling matters here. If an input produces two heights, stop and write "not a function" rather than forcing a number. If an input lands on a vertical asymptote β common in rational functions like y = 1/x where x = 0 β the function is undefined at that input, and the correct entry is "undefined," not a large number read off the steep part of the curve.
If the input falls outside the plotted domain entirely, mark it off the graph and say so. Inventing a plausible-looking output for an out-of-range input is a worse error than leaving the cell flagged, because a flagged cell can be corrected.
Worked Straight-Line Example: y = 2x Across Five Inputs
Take the line y = 2x passing through (0, 0), (1, 2), (2, 4), (3, 6), and (4, 8). The graph rises two units for every one unit it runs right, so the grid reads cleanly at every integer input, and no interpolation is needed for the table below.
| Input (x) | Graph reading | Output (y) | Pair |
|---|---|---|---|
| 0 | on the x-axis | 0 | (0, 0) |
| 1 | two units up | 2 | (1, 2) |
| 2 | four units up | 4 | (2, 4) |
| 3 | six units up | 6 | (3, 6) |
| 4 | eight units up | 8 | (4, 8) |
Two checks confirm the table. Each output is exactly twice its input, and the rise over run computed from any consecutive pair β (2 β 0) over (1 β 0), for instance β equals 2 throughout. When a table passes both checks, the rule is consistent and the reading is sound.
Now suppose the table asked for x = 2.5 instead of x = 2. The line passes through that x at a height of 5, exactly halfway between gridlines 4 and 6, so the output for that row would be 5, not 4 and not 6.
This is where a scale error shows up loudly. On a graph where every labeled tick spans two grid squares instead of one, reading the height at x = 2 as "four grid squares" would yield 8 units, doubling every output in the table. Counting units, not squares, is the fix.
Worked Curved Example: y = xΒ² and the Symmetry Trap
Take y = xΒ² over the inputs from β3 to 3. The curve is a parabola with its lowest point at the origin, and it rises the same amount on both sides of the y-axis. A table that lists only non-negative inputs will never show the trap. A table that lists both signs will punish careless reading.
| Input (x) | Graph reading | Output (y) | Pair |
|---|---|---|---|
| β3 | nine units up | 9 | (β3, 9) |
| β2 | four units up | 4 | (β2, 4) |
| β1 | one unit up | 1 | (β1, 1) |
| 0 | vertex, on the x-axis | 0 | (0, 0) |
| 1 | one unit up | 1 | (1, 1) |
| 2 | four units up | 4 | (2, 4) |
| 3 | nine units up | 9 | (3, 9) |
The symmetry is a built-in answer check you should use deliberately: if the graph is symmetric about the y-axis and the inputs include both n and βn, both rows must carry the same output. If they don't, you have misread a point or the scale changes between quadrants.
The subtle failure here is reading (β2, 4) as (4, β2) because your eyes travel the wrong axis first. That single swap makes the table look reasonable while every pair is reversed. Check one point against the symmetry rule and the swap surfaces immediately.
For a graph shifted sideways, such as y = (x β 2)Β² + 1, the vertex moves to (2, 1) and the symmetry check must be applied about x = 2 rather than about the y-axis. State the axis of symmetry in your notes before filling the table, or the check will reject correct work.
Run the Vertical Line Test to Rule Out Non-Functions
Before trusting any completed table, confirm the graph passes the vertical line test: every vertical line drawn over the graph's domain meets it at most once. Fail that test and no function table exists, because a function assigns exactly one output to each input by definition.
The usual example is a circle of radius 3 centered on the origin. Its equation is xΒ² + yΒ² = 9, so the input x = 0 yields two outputs, y = 3 and y = β3. A table listing 0 in the input column cannot be completed as a function. The honest entry is that the relationship is not a function, with the two values noted.
Sideways parabolas are the second common case. A curve of the form x = yΒ² passes through (4, 2) and (4, β2) at the same time, and the same verdict applies. Recognizing the shape of the equation before looking at the graph saves the whole exercise: anything with yΒ² and no xΒ² isolates on the same side of the equals sign is a candidate failure.
Vertical-line misconduct can also be local rather than global. A graph that is a function everywhere except at one input, often an asymptote, still produces a completable table for the rest of its rows. Mark the single undefined input and proceed with the others.
Trace the Rule to Verify the Table and Predict Unlisted Rows
A completed table is evidence, not proof. Turn each row into a test of the rule: compute the difference between consecutive outputs for a straight-line graph, or compare the output to the input's square for a parabola. If any row contradicts the rule, revisit that point on the graph before accepting the table.
For straight lines, the rule is y = mx + b, where m is rise over run and b is where the line crosses the vertical axis. Measure both off the grid and rebuild each output from the formula. On a line through (0, β1) rising one unit per two units right, the rule is y = 0.5x β 1, and the table should show (β4, β3), (β2, β2), (0, β1), (2, 0), and (4, 1). If your graph reading gave (2, 1) instead of (2, 0), the formula catches it at once.
Use the rule to extend the table past the plotted range. A table asking for x = 6 on the y = 0.5x β 1 line yields 2 even though the graph may stop at x = 4, because the rule is what defines the function. The graph is a drawing of the rule, not a fence around it.
Predicting unlisted rows is also the fastest way to check a family of graphs. Two parallel lines share the slope and differ only in the intercept, so their tables differ by a constant across every row. If the tables differ by varying amounts, one of the intercepts was read wrong or the lines are not parallel after all.
Spot the Six Mistakes That Invalidate a Function Table
Nearly every wrong function table traces back to a short list of errors, and knowing them lets you audit your own work in under a minute.
| Mistake | What it looks like | How to catch it |
|---|---|---|
| Swapped pair order | output written in the input column | first pair fails an x-then-y re-read |
| Grid square read as a unit | every output doubled or tripled | count subdivisions against labeled ticks |
| Mirror confusion | (β2, 4) entered as (4, β2) | symmetry rule disagrees between opposite inputs |
| Two outputs given for one input | circle or sideways parabola treated as a function | vertical line test fails |
| Misfit rule accepted | one row contradicts the formula | recompute each output from m and b |
| Undefined input forced | asymptote height copied as a number | vertical line test fails locally only |
The scale mistake deserves special attention because it survives every other check. A table filled from a graph whose labeled ticks span two grid squares per unit can have curve shape, symmetry, and rule consistency all correct while every value is off by a factor of two. Only the direct comparison of grid squares to labeled units catches it.
The misfit rule has a specific remedy. When one row disagrees with the formula while all the others agree, the problem is almost always that point's reading off the graph, not the formula. Re-locate the point, re-check which gridline it sits on, and re-read its height before adjusting anything else.
Mirror confusion is the error most likely to be judged "close enough." It is not close at all; a reversed pair is a wrong point, and a table built from it will fail the next section's questions about the rule.
Audit the Table Against the Graph One Row at a Time
Finish by working backward. Cover the output column, take each finished output, and return to the graph to confirm a point exists there at exactly that input. Row by row, this takes about a minute for a five-row table and catches anything missed during the forward pass.
Confirm three things in each row. The point (x, y) lies on the graph. The output follows the rule you extracted. The sign agrees with the quadrant the graph occupies. A positive input producing a negative output on a curve that never dips below the horizontal axis is a contradiction the graph itself rules out.
If a checked row fails, mark it and re-read the point rather than patching the number to fit. Patching hides which of the earlier mistakes occurred and the next row will fail the same way. Re-reading finds the cause.
Close the loop by writing a one-line statement of the rule beneath the table β "y = 0.5x β 1 over inputs β4 to 4" or "y = xΒ², symmetric about x = 0" β so anyone reading the table later, including you, can verify its rows without the graph at hand.
FAQ
Does the table always start from the marked points on the graph?
Not always. Discrete graphs made of plotted dots limit the table to those dots, but continuous curves accept any listed input, including ones without a visible marking. The table's instruction line decides. When it says "use the points shown," restrict yourself to the dots; when it gives its own input column, you must interpolate.
What do I write when the graph has no point at a listed input?
Write "undefined" and note why. A vertical asymptote, as in y = 1/x at x = 0, is the standard cause: the function has no finite value there, and the curve's steep approach is not the value. Placing a large number in that cell because the line shoots upward is the most frequent version of this error.
Is a completed table a proof that the graph is a function?
No. A table can be filled row by row and still be built on a non-function, so run the vertical line test first. Any vertical line crossing the graph twice voids the table entirely. The test takes seconds and settles the question before any cells are written.
How does a linear table differ from a quadratic one?
A linear table's consecutive outputs change by the same fixed amount, while a quadratic's changes grow steadily larger. For y = 2x the outputs step 2, 2, 2; for y = xΒ² they step 1, 3, 5, 7 as the inputs rise by one. Check the step pattern and you can name the family without seeing the graph.
Can I complete the table if the graph is cut off before the last input?
Yes, by using the function's rule rather than the drawing. A line through (0, β1) with slope 0.5 gives an output of 3 at x = 8 even if the axis stops at 4. The graph illustrates the rule over a window; the rule itself has no such window.
What if two rows of the table read the opposite signs from what the graph shows?
That is a reading error, not a graph error. A curve sitting entirely above the horizontal axis cannot produce negative outputs, so any such row was read from the wrong quadrant or with swapped coordinates. Re-locate that point on the grid before touching the others.
Does the order of the input column change the answers?
The values are pairs, so order by itself does not change what each input produces, but keeping the table's order exposes errors sooner. Re-ordering can hide a swapped pair or a scale slippage by breaking the pattern the checks rely on. Fill in the given sequence.
How do I handle a graph with different scales on the two axes?
Note both scales separately before reading any point, then convert each coordinate using its own axis. Graphs of real data routinely compress the vertical axis, so a height of two grid squares may represent 20 units. Reading squares in place of units is the failure mode to guard against.