When Representations Disagree
A graph may appear to approach one height while a table of nearby values seems to approach another. When that happens, the right response is not to choose whichever representation looks more convincing. First check that both observations concern the same function, target, and direction of approach. Then ask whether the data are sufficiently informative to justify either estimate.
Graph and table evidence is useful for estimating a limit, but finite observations do not usually prove what happens arbitrarily close to the target. A disagreement is a reason to investigate the evidence, not an automatic proof that a limit does not exist.
What Counts as a Limit Estimate?
Suppose \(f\) is defined on a domain \(D\), and we are interested in the behavior of \(f(x)\) as \(x\) approaches \(a\). A graph estimate comes from tracing the curve toward \(x=a\), without treating the value at \(a\) as the approached value. A table estimate comes from looking at outputs for inputs in \(D\) that get closer to \(a\).
For a two-sided limit, evidence must address inputs approaching from both the left and the right. A left-hand observation alone cannot settle the two-sided behavior, and a right-hand observation alone cannot settle it either. When comparing representations, keep the direction explicit.
The condition \(0<|x-a|\) excludes the input \(x=a\). Therefore, the function’s value at the target, if it is defined, does not by itself determine the limit. The accumulation-point condition matters too: without nearby domain inputs, the definition cannot describe an approached value in the usual way.
A Careful Comparison
Before deciding whether observations conflict, record what each one actually says. A graph may only suggest an approximate height because its scale, thickness, or resolution limits what can be read. A table may contain only a few inputs, or may list inputs from just one side. These are different kinds of limitations.
| Question | Why it matters |
|---|---|
| Are both observations about the same function? | Evidence about different functions cannot be combined to estimate one limit. |
| Are they about the same target input? | Behavior near different targets may differ. |
| Are they approaching from the same side? | Left-hand and right-hand behavior may differ. |
| How close are the inputs or graph trace to the target? | Observations farther away may not represent the behavior very near the target. |
| Is the evidence exact or only visual and approximate? | A visual estimate can be misleading or too imprecise to distinguish nearby heights. |
If those checks reveal that one source is about the left side and the other about the right, the observations do not directly disagree about the same one-sided behavior. If they do concern the same side, but suggest different heights, the evidence is apparently inconsistent. That still does not tell us which source is wrong, or whether the actual limit exists.
Worked Example: Compatible Graph and Table Evidence
Worked Example: Compatible Graph and Table Evidence
A graph of \(f\) near \(x=1\) appears to approach \(y=2\) from both sides. A table gives the following values, all with inputs to the right of \(1\). Do the observations conflict?
First compare each output with the proposed height \(2\):
The input distances from the target are:
Thus the tabulated outputs get closer to \(2\) as the inputs get closer to \(1\) from the right. That is compatible with the graph’s estimate of \(2\). The table supplies right-side evidence; the graph appears to show both sides.
Check: Each table output is \(2+(x-1)\), so its distance from \(2\) equals the input’s distance from \(1\). This confirms the arithmetic and the observed pattern. The finite table and visual graph are consistent, but they do not by themselves prove that the limit is \(2\).
Worked Example: Evidence That Appears to Conflict
Worked Example: Evidence That Appears to Conflict
Near \(x=0\), a graph appears to approach height \(1\) from the right. A table for the same function gives:
The input distances from \(0\) are \(0.1\), \(0.01\), and \(0.001\). The table outputs’ distances from \(3\) are:
Those values support a right-side estimate of \(3\). Compared with the graph’s proposed height \(1\), the table outputs have distances:
The table and graph therefore give apparently conflicting estimates for the same side of the same target. The table entries have the exact pattern \(f(x)=3+x\) at those listed inputs, but three samples cannot establish the behavior at every sufficiently close input. Likewise, the graph’s visual trace is an estimate, not exact proof.
Conclusion: Do not average the two heights or silently select one. Recheck the graph scale, plotted function, target, and side of approach; then gather more accurate observations closer to \(0\). If independent, exact information truly established that the same function approached both \(1\) and \(3\) from the right, that would be impossible, as the uniqueness argument below shows. These finite observations alone do not establish that situation.
Check: The table’s distances from \(3\) equal the corresponding input distances from \(0\), while its distances from \(1\) remain greater than \(2\). The arithmetic confirms why the two estimates look different.
Why Two Genuine Limit Values Cannot Conflict
There is an important difference between conflicting observations and conflicting proven limits. A function cannot have two different limits at the same accumulation point. The accumulation-point condition is necessary for this uniqueness result: it ensures there are domain inputs, other than the target, close enough to test both proposed limits.
To prove the theorem, suppose instead that the two proposed values differ. Let \(d=|L_1-L_2|>0\), and set \(\varepsilon=d/3\). Since both are limits, there are positive numbers \(\delta_1\) and \(\delta_2\) such that domain inputs within the respective punctured neighborhoods satisfy \(|f(x)-L_1|<d/3\) and \(|f(x)-L_2|<d/3\).
Let \(\delta=\min(\delta_1,\delta_2)\). Any domain input \(x\) with \(0<|x-a|<\delta\) meets both conditions.
Because \(a\) is an accumulation point of \(D\), there is an \(x\in D\) with \(0<|x-a|<\delta\).
For this input, \(d=|L_1-L_2|\leq |L_1-f(x)|+|f(x)-L_2|<d/3+d/3=2d/3\), which contradicts \(d>0\).
The contradiction proves \(L_1=L_2\). Notice why step 2 cannot be skipped. If \(a\) were not an accumulation point, there might be no allowed \(x\) in the punctured neighborhood at all. The limit conditions could then hold vacuously for different values, and the uniqueness argument would fail.
Worked Example: Sparse Data Do Not Settle a Limit
Worked Example: Sparse Data Do Not Settle a Limit
Suppose a table near \(x=0\) contains only these observations:
The input distances are \(0.1\) and \(0.01\), while the outputs are \(2\) and \(0\). These two points do not show a consistent trend toward one height. In fact, finite data of this kind cannot determine the limit on their own.
To see why, consider a function on the punctured domain \(x\ne 0\) that equals the stated outputs at \(x=0.1\) and \(x=0.01\), and equals \(1\) at every other input. Its limit at \(0\) is \(1\): choose \(\delta=0.005\); then every \(x\) satisfying \(0<|x|<\delta\) is neither listed input, so \(f(x)=1\).
Now make a second function with the same two listed outputs, but let it equal \(4\) at every other input. Its limit at \(0\) is \(4\), because the same choice \(\delta=0.005\) ensures that every allowed nearby input has output \(4\).
Check: Both functions give \(f(0.1)=2\) and \(f(0.01)=0\), yet their limits are different. In each case, \(\delta=0.005\) excludes both exceptional inputs because \(0.005<0.01<0.1\). Therefore the two table entries alone cannot justify a particular limit.
How to Report the Evidence
When sources disagree, a precise conclusion separates what is observed from what is established. For example: “The table suggests a right-hand limit of \(3\), while the graph appears to suggest \(1\); these estimates conflict, so the available representations do not justify a single estimate.” That statement does not claim the limit fails to exist. It accurately describes the evidence and its limitation.
A useful check is to ask whether more nearby observations resolve the disagreement. For a table, choose inputs closer to the target on the relevant side. For a graph, inspect the scale and trace the curve toward the target rather than relying on a distant segment. If the sources still disagree, state that uncertainty. If left- and right-side evidence differs, keep the one-sided conclusions separate rather than blending them into one two-sided estimate.
Why the Distinction Matters
Limit estimates are based on behavior near a target, not on a vote between representations. A graph can be hard to read, a table can be sparse, and either source can omit the side needed for a two-sided conclusion. Checking the target, direction, and closeness of observations helps prevent an unsupported answer.
The main habit is simple: describe each source’s evidence accurately, compare only observations addressing the same behavior, and match the strength of your conclusion to the strength of the evidence. Compatible observations support an estimate; conflicting observations call for investigation; sparse observations may leave the limit unresolved.
Check Your Understanding
For each question, distinguish what the observations suggest from what they prove.
- A graph appears to approach \(y=4\) from both sides of \(x=2\), while a table of right-side outputs gets closer to \(4\). Are the representations compatible, and does that finite evidence prove the limit?
- A table suggests a right-hand value of \(1\), while a graph appears to approach \(3\) from the left. Do these observations directly conflict about the same one-sided behavior?
- Why does a table with only two inputs near a target generally fail to determine the limit?
- In the uniqueness proof, what role does the assumption that \(a\) is an accumulation point of the domain play?
- If a graph and table appear to give different heights for the same side, name two checks to make before drawing a conclusion.
