Reading Figures and Supporting Information

Introduction

A figure can communicate a result faster than a page of text, but it can also hide essential choices. Axes may begin above zero, curves may be normalized to different masses, or error bars may represent technical repeats rather than independent samples. Supporting information often holds detailed synthesis, calibration, control plots and raw spectra. Reading both the figure and its underlying method is necessary before accepting a chemical comparison.

Core explanation

Read the caption before interpreting the plot. Identify the measured quantity, axis units, scale, treatment groups and sample count. A logarithmic axis changes visual distances; a truncated vertical axis can magnify a modest difference. If a curve is “normalized,” find the denominator: per gram of active material, per electrode area, per whole-cell mass or relative to initial value. Two battery curves that look similar after normalization may correspond to very different absolute capacities or loadings.

Check what the plotted points represent. Are they single devices, group means or model predictions? Do shaded bands show standard deviation, standard error, a confidence interval or a fit range? Without that label, the apparent precision cannot be evaluated. A narrow error bar from repeated reads of one sample does not show between-batch reproducibility. Some figures omit failed samples or select a representative image; ask how that selection was made. Independent replicate counts and exclusion rules should be in methods or supporting material.

For spectra, inspect baseline, peak assignment and controls. A small peak may be an impurity, overlapping transition or background artifact. Compare the raw or minimally processed spectrum with reference and blank spectra. For microscopy, ask about image scale, contrast settings, field selection and preparation artifacts. A micrograph of one particle cannot by itself establish the frequency of that feature across a batch. For kinetics, see whether data are initial rates, endpoint yields or transformed axes; fitting a straight line to logarithms changes the error structure.

Supporting information may provide experimental procedures, instrument settings, additional replicates, computational parameters and calibration plots. ACS author guidance permits detailed procedures in supporting information while emphasizing important methods in the main manuscript. Treat supplementary material as part of the paper's evidential record, not as optional decoration. A key claim should still be understandable from the main article, but its validity may depend on controls shown only in supplementary figures.

Cross-check figure, caption, main text and table. If a caption says n = 3 , determine whether that is three independent synthesis batches or three instrument scans. If a main-text efficiency differs from the plotted value, perhaps one is stabilized power and another is a scan maximum. If a unit conversion or axis label is missing, do not guess. Record the precise panel and conditions supporting a claim so it can be verified later.

Visual uncertainty is especially important near a decision threshold. Overlapping error bars do not by themselves prove no difference; nonoverlap does not automatically establish a meaningful mechanism. Interpretation depends on what the bars mean, dependence between groups and the appropriate comparison model. Use the raw sample values or a stated statistical analysis when possible, and consider practical magnitude as well as statistical evidence.

Step-by-step reasoning

Read title and caption, then list every axis with units and scale. Identify raw versus normalized response and the experimental unit. Find the methods that generated the plotted points, including sample selection and processing. Inspect blanks, standards, references and replicate panels, often in supporting information. Compare the figure's visible pattern with the authors' textual claim and note any gap. Cite the specific figure or supplementary section that actually supports the conclusion.

Visual explanation

Imagine the same data plotted three ways: absolute capacity on a zero-based axis, capacity normalized to the first cycle, and a vertical axis truncated near the top. The underlying measurements are unchanged, but visual emphasis shifts. Place a callout on the caption defining each denominator and error bar. A side panel lists the supplementary control graph needed to rule out instrument drift.

Real-world analogy

A map without a scale bar can make a short walk look like a long journey. A chemistry graph without clear units, normalization and sample count can similarly exaggerate or obscure a difference. The legend and supporting methods are the scale bar for scientific figures; one must read them before trusting visual distance.

Real-world example

A solar-cell paper shows two current–voltage curves and claims a contact layer improves efficiency. The main plot uses a selected best device for each condition. Supplementary data show 20 devices per group, with the new layer improving median fill factor only slightly but reducing manufacturing failures substantially. The practical benefit is yield, not necessarily the record efficiency implied by the main figure. Reading the full distribution changes the interpretation.

Why?

Why can a supporting-information control be decisive? A striking main-panel signal may also appear in a reagent blank or dark control. A supplementary control can show whether the claimed chemistry is specific to the new treatment. Omitting that check from one's reading leaves a mechanism claim resting on an ambiguous picture.

Common misconception

“The graph speaks for itself.” Axis choices, preprocessing and experimental-unit definitions shape what it appears to say. “Supporting information is less important because it is outside the main text” is false for method and control evaluation. “Overlapping error bars prove treatments are equivalent” confuses a visual heuristic with a formal equivalence test and ignores the type of bars.

Worked example

A battery figure reports 90% retention after 100 cycles for material A and 85% for B. The supplement reveals A began at 100 mAh/g and B at 150 mAh/g. Final specific capacities are 0.90 × 100 = 90 mAh/g and 0.85 × 150 = 127.5 mAh/g . A has better retention percentage, but B still delivers more capacity under those conditions. Neither number alone captures full-cell energy, rate or lifetime. The example shows why normalized curves and absolute starting values must be read together.

Quick check

1. What should be checked when a graph shows tiny error bars but the caption does not define them? Answer: Find whether bars represent standard deviation, standard error, confidence intervals or instrument repeats, and identify the number of independent experimental units before interpreting precision.

Exam focus

Extract axis units, scale, denominator and sample count before discussing trends. Distinguish a representative image from a measured population. Look for controls and detailed procedures in supporting information. Compare numeric values with normalized visual claims, and state what the figure does and does not establish about a mechanism.

Advanced insight

Figure selection can create a form of publication bias within one article: only the most visually persuasive panels may appear in the main text. Distribution plots and raw-data availability help readers assess variability. Data extraction for meta-analysis should use exact reported values and uncertainty definitions; digitizing a curve is a last resort when tables are unavailable and introduces its own error.

Summary

Chemistry figures must be read with their captions, methods and supporting information. Axes, units, normalization, error-bar definitions and sample selection determine the meaning of a visual pattern. Controls and raw or additional data may appear outside the main text. A careful reader reconstructs the measured comparison before repeating the authors' interpretation.

Practice questions

1. A capacity plot shows only percent retention. What additional value is needed to compare final absolute capacities? Answer: Each material's initial absolute capacity under comparable units and conditions is needed; multiply it by retention fraction to obtain final capacity.

2. A micrograph shows one cracked electrode particle. What can and cannot be inferred? Answer: The feature exists in the imaged region, but its prevalence across the electrode or causal contribution to performance requires representative sampling and complementary measurements.

3. Why should one inspect the supporting-information blank for a spectroscopic peak claim? Answer: The peak may arise from solvent, reagent, instrument or preparation background. A matched blank helps determine whether it is specific to the claimed species.