How do you translate a chromatogram into plain language or another language without mistaking a peak for proof of a chemical’s identity? Start with the horizontal axis, the detector response, the peak labels and the analytical method. A chromatogram is not a picture of molecules. It is a record of a separation experiment, showing when a detector responded and, with appropriate calibration, how that response can contribute to an estimate of amount. An accurate translation must preserve that distinction.
People looking for how to read a chromatogram, explain HPLC results, interpret a gas chromatography graph or translate chromatography peak tables often encounter the same difficulty: the graphic looks simple, but its labels are doing several different jobs. Retention time locates a response in the run. Peak area integrates a response over a time interval. A compound name can be a proposed assignment rather than an independently established fact. The translator must not turn any one of those into another.
This guide shows students, educators, multilingual laboratory readers and technical communicators how to convert a chromatogram into a careful verbal description. It treats translation as a transfer of evidence, not merely of words. Every example below is hypothetical and educational; laboratory identification, validation and release decisions require the applicable analytical method and competent specialist review.
What the chart actually reports
A chromatographic instrument separates parts of a sample under specified conditions and produces a detector signal over time or, in some methods, another separation coordinate. In common gas and liquid chromatography displays, time advances along the horizontal axis. The vertical axis shows a detector response such as absorbance or ion signal, not necessarily concentration in any direct sense. A cluster of readings rises above a baseline to form a peak. The visual height makes a dramatic impression, but the recorded signal depends on detector type, sensitivity, method settings and how the sample was prepared. To translate the chart, identify what was measured rather than describe only what the line resembles.
The reader needs to know whether the image shows a complete run, an enlarged section, a comparison of two runs, a stack of detector channels or a processed peak table. A screenshot might omit the start of the run, the final elution or the method note that explains the signal. A laboratory report might display a processed trace with integration boundaries already chosen. These details change what can be concluded. An adequate translation is therefore an account of a particular record with its limits, not a general story about ‘what the chemical did.’
The first pass: title, sample, instrument and method
Before interpreting a peak, locate the figure title, sample identifier, analytical method, injection number, acquisition date when present, detector name and any channel or wavelength setting. These are not decoration. Two traces that look comparable may have been collected with different methods. A translated heading should preserve whether the page describes a standard solution, a blank, a control or an unknown. ‘Sample 7’ must remain Sample 7; it must not become a named substance just because the adjacent paragraph discusses that substance. Keep identifiers as identifiers.
Method context governs equivalence. A detector configured at one ultraviolet wavelength may register a substance differently from a detector set at another. A mass-selective channel may display a particular ion signal rather than a general response. Column conditions, mobile phase, temperature and gradient programme affect where peaks appear. Translate labels, but do not silently normalise methods into one convenient narrative. A useful description says, for example, ‘the report shows a UV detector trace for the stated sample under the stated method’ before interpreting any peak assignment.
How to preserve both axes and every unit
The horizontal axis in an ordinary chromatogram is frequently labelled time, minutes or seconds. The point corresponding to 4.5 minutes does not mean a compound has existed for 4.5 minutes, nor that the experiment took only 4.5 minutes overall. It indicates the detector response associated with that position in the separation run. The vertical axis may use arbitrary units, millivolts, milli-absorbance units or counts. These units matter because one laboratory’s ‘200’ may be incomparable with another laboratory’s ‘200.’ A translated figure should reproduce units exactly.
When the axes are unlabelled, identify them as unlabelled rather than inventing likely units. A logarithmic scale, a baseline-corrected display and a magnified inset can change the reader’s perception. If a chart begins at a nonzero value, a peak may look taller than it would on an origin-based display. Translators sometimes shorten ‘detector response’ to ‘amount of chemical’ because it sounds more intuitive. That is an inference requiring additional method information. Keep the observable signal separate from any quantitative interpretation.
Retention time is a coordinate, not an identity card
Retention time is the time associated with an analyte’s passage to the detector under the specified chromatographic conditions. It is commonly reported at a peak’s apex or by the method’s definition. If two peaks appear at 2.1 and 5.8 minutes, the second arrived later in that run. That statement does not prove which substances produced them. The same substance can move to a different retention time when the separation conditions change; different substances can sometimes elute at similar times. A target-language phrase that implies unique identification would overstate the evidence.
Comparing retention times to reference standards can support an assignment when conditions match. Some methods use relative retention time or additional detection information. Preserve these distinctions in captions and tables. ‘Consistent with’ is different from ‘confirmed as.’ ‘Peak assigned to compound A’ is different from ‘compound A definitively present.’ A translator working from a report should maintain the original level of certainty even if a simpler sentence would be easier to read. When identity is material, document what independent confirmation the report actually supplies.
Peak height and peak area are not interchangeable
Height describes how far the signal rises above its defined baseline at a particular peak. Area describes the integrated detector response across a chosen interval. A narrow tall peak can have less area than a shorter broad peak. That is why changing ‘area’ to ‘height’ creates a substantive measurement error. If the table reports Area 250 and Height 90, those values identify different calculated features. Avoid translating both as a generic ‘peak size.’ Preserve the column headings and units, and explain which quantity is being used in the original conclusion.
Peak area may be related to analyte amount through a calibration model and suitable conditions, but raw area does not automatically equal concentration, mass, purity or percentage of the sample. Some detectors respond differently to different compounds. An area-normalised percentage is a ratio of integrated detector responses under defined rules; it is not necessarily mass fraction. If the source explicitly uses ‘area percent’, translate that term rather than replacing it with ‘purity percent.’ This may sound pedantic, but the difference can change a student’s answer or a reader’s understanding of a technical report.
Baseline, noise and genuine evidence
Between peaks there is often a baseline: the detector’s background response while no distinct eluting component is recorded above its normal variation. The baseline may drift gradually, curve under a broad feature or contain noise. The existence of a wiggle does not guarantee a chemically meaningful peak. Software may apply smoothing or baseline correction, and those processing choices influence which features are detected and integrated. A plain-language rendering should distinguish ‘a small rise in detector signal’ from ‘an identified impurity’ unless the original evidence supports the stronger conclusion.
A translated description can say that a low-amplitude feature appears near a certain time, that its status is uncertain, and that the supplied record does not show how it was classified. This is more useful than guessing. The same discipline applies to blank runs. A peak also appearing in the blank could be associated with a solvent, carryover or another source of background, depending on method investigation. It should not be erased from the description simply because the narrative is about the sample. Context decides significance.
Read the peak table as data, not prose
A peak table can contain retention time, start and end time, area, height, width, area percentage, compound assignment, signal-to-noise ratio and other method-specific fields. Translate the column headers accurately, preserving their units and the relationship between a row and a peak. If the report uses Peak 1, Peak 2 and Peak 3, keep that numbering even when their order appears obvious. Changing the order to sound more elegant may break correspondence with the image. A numeric row is not just a string of facts; it belongs to a specific feature.
Tables are vulnerable to decimal and separator errors when moving between language conventions. A value of 1.250 may use a decimal point, while 1,250 may be a thousands grouping or a decimal comma depending on locale. Confirm conventions from the source before localising punctuation. Do not round 0.049 to 0.05 without permission if the original precision matters. Preserve scientific notation, inequality signs, omitted values, flags and footnotes. When a target table cannot preserve layout, label each value with its own field name to prevent column drift.
Why integration boundaries change the story
Peak integration is the process of defining which signal belongs to a feature and estimating an area relative to a selected baseline. Boundaries can become difficult when two peaks overlap, when a shoulder appears, or when a drifting baseline cuts beneath a broad feature. A report may mark start and end times or show a software-generated integration line. The resulting area depends on those choices, which is why a screenshot without integration details cannot fully explain the peak table. Translate what the record shows, not what you imagine the software should have done.
Suppose a broad peak begins near 6 minutes and has a subtle shoulder at 6.6 minutes. One processing method may report a single combined feature; another may split it into two peaks. Without the method and evidence, neither narrative should be asserted as unquestionably correct. A careful description can say ‘the trace contains an apparent shoulder; the accompanying table integrates the region as one peak.’ The wording protects the difference between observed shape and analytical classification. That difference is a cornerstone of trustworthy technical communication.
Overlap, co-elution and the false comfort of a clean peak
Two compounds may elute close together or at essentially the same apparent time, producing partially overlapping or unresolved responses. A single smooth feature on a general-purpose detector does not always correspond to a single chemical constituent. Conversely, a complex-looking feature may arise from acquisition or processing effects rather than multiple analytes. Accurate translation cannot turn a visually tidy peak into proof of exclusivity. Describe its appearance, identification method and any uncertainty specified by the source.
Words such as ‘resolved’, ‘separated’ and ‘co-eluted’ have technical meanings that are narrower than ordinary ‘clear’ or ‘mixed.’ If a source reports insufficient resolution, the target should not present two peak assignments as secure merely because they are printed on separate lines. If it reports a shoulder, preserve ‘shoulder’ rather than calling it a second confirmed substance. Technical readers rely on these distinctions because interpretation may guide the next analytical step, such as changing a separation method or using orthogonal detection.
The difference between paper chromatography and instrument traces
A classroom paper chromatogram often shows coloured spots that moved different distances along paper, while an HPLC or GC chromatogram often shows a detector trace with peaks versus time. Both belong to chromatography, but they require different explanations. Paper or thin-layer chromatography uses movement relative to a solvent front under stated conditions and can involve retention factor, commonly written Rf. An instrumental peak’s retention time is not an Rf value. Translate the representation you actually have.
In a paper example, identify the origin line, solvent front, separated spots and distances measured from the appropriate reference. Rf is ordinarily the distance travelled by a spot divided by the distance travelled by the solvent front under the stated procedure. It is dimensionless. If a student measures from the edge of the paper rather than the origin, the result changes. An explanation of a gas chromatogram should not introduce Rf unless the original task explicitly makes that comparison. Similar subject names do not authorise importing another method’s variables.
Standards, blanks and controls as separate evidentiary roles
A standard contains known or characterised material used for calibration or comparison under a method. A blank helps identify responses not attributable to the target sample. A control can test whether an analytical system is performing as expected. These roles are not synonyms. If the original report shows traces marked Blank, Standard and Sample, a translated version must keep them separate, even when their peaks partly align. Otherwise the reader may attribute a known reference signal to the unknown sample or treat an artefact as a result.
A clear account gives the order and purpose of comparisons: a reference was run under appropriate conditions, a sample trace was recorded, and the report compared specified features. If a peak appears at the same nominal time as a reference, say that it corresponds under those conditions rather than claiming more than the method permits. A system suitability result, calibration check or quality-control result may constrain interpretation. Include the existence of such checks when shown, but do not invent pass/fail decisions from a figure alone.
Different detectors produce different kinds of signals
UV absorbance, fluorescence, refractive-index detection, flame ionisation and mass-selective detection all generate chromatographic records, but their response mechanisms differ. Translators need not become instrument engineers to preserve meaning. They do need to keep the detector named correctly and avoid treating every vertical axis as a universal measure of quantity. A mass chromatogram may refer to selected ions or total ion current. A UV trace may state the wavelength used. Both details determine which observations are represented.
When the instrument label is unfamiliar, retain the original abbreviation with an accurate expansion when verified. Avoid replacing ‘UV 254 nm’ with ‘visible light’ or ‘mass-to-charge ratio’ with ‘mass’ if those replacements distort the method. The goal is to make the reader more informed without flattening science into approximations that are false. Glossaries should hold recurring terms such as retention time, peak area, response factor, baseline and co-elution. Choose a consistent target-language equivalent after checking reputable field usage.
Preserving numbers and significant figures
Translating technical charts requires more than swapping decimal separators. Measurement values have precision, uncertainties and qualifiers. ‘Less than 0.1’ must not become ‘0.1’; ‘approximately 4.0’ is not the same as ‘exactly 4.0.’ If a table reports a retention time to three decimal places, preserving the reported digits is usually safer than informal rounding. The reader should also know whether a value is measured, calculated, estimated or below a stated threshold. Those are different epistemic statuses.
Take a hypothetical note: ‘Peak B: tR 3.452 min; area 12,845 units; estimated amount below the quantitation limit.’ A faithful version keeps the time, unit and thousands convention while preserving the limitation on amount. It must not turn the area into 12,845 milligrams or treat ‘below limit’ as zero. If a localisation convention changes punctuation, perform a separate numeric check against the source. That review catches mistakes that ordinary spellcheckers do not understand.
Uncertainty, detection limits and non-detection
A non-detected substance is not necessarily absent from the sample. The method may not respond to it, the amount may be below the detection limit, or the chosen conditions may not reveal it. Likewise, a detected signal is not always quantifiable with the needed reliability. Method reports distinguish the limit of detection and limit of quantitation for good reasons. A translator should retain ‘not detected’, ‘below limit’, ‘not quantified’ and ‘not tested’ as distinct states.
When a chromatogram shows no obvious peak in one region, write ‘no distinct peak is visible in the supplied trace at this display scale’ rather than ‘the sample contains none of this compound.’ That description is honest about what was inspected. If the source reports a validated non-detect, preserve the source wording and test context. Translation cannot upgrade instrument visibility into certainty about the entire specimen. The same caution applies when a translated caption will circulate independently from the full report.
An accessible verbal order for describing a trace
For readers who cannot use the graphic directly, spatial descriptions should still carry the scientific sequence. Start with the type of chart and axes, followed by the run interval. Then describe the number of prominent features, their approximate locations and relative sizes only when these are visible and reliable. Follow with the table’s exact peak assignments and numerical values, distinguishing what the instrument recorded from what the analyst inferred. End with limitations the report states or the image imposes.
Consider: ‘The chart displays detector response against elapsed minutes from zero to ten. Three integrated peaks are listed. Peak 2 occurs at 4.30 minutes and has the largest integrated area. The accompanying table labels it as the reference compound; the figure alone does not independently confirm identity.’ That wording is usable in an accessibility description, a translation or a classroom explanation. It avoids hiding the method behind visual adjectives such as ‘big spike’ or ‘smooth mountain.’
A language-learning approach to technical terms
A bilingual student may know the everyday word ‘retention’ as keeping something, yet misunderstand chromatographic retention as storage in a container. In chromatography, the term describes behaviour during separation, not the instrument holding a sample forever. The everyday word ‘area’ may suggest a physical surface, while peak area is an integrated signal. Teaching both meanings and their contexts prevents false friends even when the two languages share a familiar cognate.
Build a small two-column terminology note, but add a third column for scientific function. Under ‘peak’, write ‘detector signal feature’, not merely a translated noun. Under ‘baseline’, write ‘reference level for estimating response’; under ‘standard’, write ‘known comparison material’; under ‘impurity’, distinguish a reported assignment from an unexplained small signal. Learners can then produce sentences that connect words to evidence. Vocabulary becomes useful when a student can explain why a particular term is correct here and misleading elsewhere.
What machine translation and OCR can safely do
Optical character recognition can recover labels and table cells from a legible image, while machine translation can suggest target-language terminology. Neither establishes whether a small feature is a valid peak or whether a compound assignment is justified. OCR is particularly vulnerable to negative signs, superscripts, cramped decimal places and similar-looking characters. It may move a value into a neighbouring column. Check the recovered text against the image before asking an automated system to explain it.
AI tools can help draft a plain-language description when the source includes enough context, but reviewers should demand traceable support for every numerical and identification claim. A useful prompt asks for a separation between ‘observed features’, ‘table-reported values’ and ‘interpretations requiring validation’. A dangerous prompt asks the system to identify every compound from the picture alone. Humans remain responsible for confirming scientific meaning, and consequential laboratory decisions remain with qualified analysts.
Translating for children, students and specialists
The level of explanation can change while the scientific boundary stays fixed. For a beginner: ‘The instrument records a line that rises when something reaches its detector; each rise is called a peak.’ For a more advanced student: ‘Peak location is expressed as retention time, while integrated peak area may support quantitation after suitable calibration.’ For a specialist, retain detector channel, calibration model and integration parameters. These are not three different facts; they are three layers of the same evidence.
Plain language does not mean deleting important uncertainty. A child can understand ‘this peak might match the reference, but we need more evidence.’ A professional can understand a precise note about method specificity. The right translation provides sufficient explanatory scaffolding without pretending that the reader’s age or language removes constraints. A good test is to ask readers to restate what the chart directly shows, what the accompanying report claims and what cannot yet be concluded.
How to build a verifiable translated caption
Write a caption in four sentences when possible. Sentence one identifies the sample and chromatographic technique as labelled. Sentence two defines horizontal and vertical axes with units. Sentence three reports the peaks and any peak-table observations relevant to the reader’s question. Sentence four notes the evidence limits or method-dependent assignments. A caption need not include every number, but it must not conceal a qualification that changes the meaning of the result.
Review the caption against the original. Can a reader find every statement on the chart or in the accompanying documentation? Have any compounds been added, amounts invented or peaks silently merged? Have area and height been kept distinct? Does ‘later’ refer to retention time rather than later collection of the sample? If the caption is translated into several languages, compare the final scientific assertions across versions rather than assuming identical-looking sentence structures carry identical confidence.
An end-to-end verification workflow
First, capture the source title, axes, units, legend, sample identifier and method notes. Second, transcribe the peak table exactly, including blanks and flags. Third, classify every statement as observed, calculated, assigned or inferred. Fourth, translate terms according to the relevant scientific domain. Fifth, assemble a description that follows the same peak order as the source. Sixth, compare all numbers, qualifiers and compound labels. Finally, obtain domain review if the translation supports reporting, compliance, research or health-related decisions.
Keep a record of unresolved ambiguities. A source might not define an abbreviation or may show an area value that disagrees with its printed percentage. The proper action is to note the inconsistency, not to fix the figure silently. Technical translation has an editorial temptation: make the account cleaner than the evidence. Resist it. Readers benefit when the translated record preserves both its information and its limitations.
Ten worked translations: what the evidence lets us say
Worked example: the tall narrow peak
A hypothetical run contains Peak A at 1.9 minutes, height 600 units and area 900 unit-minutes. Peak B occurs at 3.8 minutes, height 410 units and area 1,500 unit-minutes. Someone says A is ‘the biggest peak’ because it is taller. If the question asks which response has the greater integrated area, B is the appropriate answer. A faithful translated caption says A has the greater height and B has the greater area. It does not assert that B contains more mass without detector-response and calibration information. This miniature example shows why the same image can support different correct comparisons, depending on which measurement the reader has been asked to preserve.
Worked example: a peak shifts between two runs
Imagine a reference standard that produces a peak near 4.0 minutes under Method X. A later run uses a changed temperature programme and reports a feature at 4.6 minutes. It is tempting to write that the sample chemistry has changed. But changes in the separation method can move retention times. The translation should identify both methods and avoid comparing coordinates as if all conditions were identical. If an analyst provides a method-specific assignment, preserve its wording and evidence. Readers should learn to ask ‘under which conditions?’ before interpreting apparent movement. The experiment’s coordinate system is part of the meaning.
Worked example: an unassigned small peak
A trace displays a small feature after the main peak, but the table labels it ‘Unknown 2’. Calling it ‘impurity X’ introduces a chemical identity not present in the source. Calling it ‘noise’ may erase a feature the laboratory chose to integrate. The safer translation states that the report integrates a small additional peak marked Unknown 2. If its significance matters, the analyst’s method and identification evidence must be reviewed. This is particularly important when a screenshot is shared without accompanying commentary: readers may otherwise mistake confident prose for a conclusion the instrument never made.
Worked example: area percentage versus purity
Suppose the table lists integrated areas of 80, 15 and 5 percent after software normalisation. A hurried translation may say the sample is ‘80% pure’. That conclusion requires assumptions about detector response, method selectivity and which components were detected. If the report itself labels the value as area percentage, the translated account should retain ‘80% of the integrated signal area within the stated method’. The result may be useful for particular validated comparisons, but it is not universally interchangeable with mass purity. Accurate words prevent the most visually appealing number from acquiring more authority than it deserves.
Worked example: the solvent-front confusion
In a classroom paper-chromatography diagram, a blue spot travels 3 centimetres from the pencil baseline while the solvent front travels 6 centimetres. Its Rf is 0.5 under the stated procedure. In the translation, neither value should be described as retention time. A reader may be studying both paper and gas chromatography in the same course, which makes the risk of term mixing real. The explanation should locate the baseline and solvent front first, then define the ratio. If the paper was tilted or the solvent front marked incorrectly, that procedural issue must remain a separate concern.
Worked example: several samples on a shared graph
An overlay shows three runs in different colours, each with a legend and offset vertical position. A translator writes ‘three peaks were detected’, but the figure actually shows one peak in each of three samples. The correct description distinguishes traces from features. It also identifies whether offsets were introduced for display and whether traces share a scale. Colour should not be the only means of identifying a sample; include names and line styles when present. The error matters because readers may otherwise infer a three-component mixture from what was in fact a comparison of three single-feature runs.
Worked example: a report and its peak table disagree
A report’s narrative calls the main feature Peak 3, while the printed table assigns the largest area to Peak 2. Without more information, the translator cannot know whether the paragraph, table or graphics contain the error. Silently harmonising them would conceal a source problem. Reproduce the labels accurately and flag the discrepancy for the responsible editor or analyst. A bilingual quality check should compare the translated narrative against both the original narrative and the original table. Translation is not a licence to resolve analytical contradictions by intuition.
Worked example: detection below a threshold
A method report states that the target was ‘not detected above the method detection limit’. A summary written for non-specialists reduces that to ‘the target is absent’. That change turns a bounded observation into an absolute claim. The accurate rendering explains that no qualifying signal was detected under the method’s stated threshold. If the threshold value is provided, preserve it with its units and relation sign. This reasoning is transferable to medical, environmental and food-testing reports: a missing displayed peak and the absence of a substance are not logically identical.
Worked example: translated units and decimal punctuation
A table prepared with decimal commas reports 3,25 minutes and 1.250 units, with the thousands point used in the second number. In a target locale using decimal points and commas for thousands, the corresponding values may be 3.25 minutes and 1,250 units. Mechanical punctuation replacement could turn one or both into a value hundreds of times wrong. A responsible workflow identifies the source numbering convention from the document, transforms values using an explicit rule and checks each result. Technical translation treats a digit as part of meaning, not as typography that can be rearranged casually.
Worked example: a chromatogram shown without its legend
A student receives an image with a horizontal scale marked 0 to 12 but no unit, and one conspicuous peak. The student can describe a single prominent feature near the centre of the horizontal range. They cannot safely state that it appeared at six minutes unless a caption elsewhere establishes minutes. They also cannot identify the compound from shape alone. This deliberately modest description is more educationally useful than a polished but invented analysis. Readers learn to recognise missing metadata as missing evidence and to ask for the original legend, method and associated table.
Questions a careful reader can answer
Ask yourself whether you can state the axis units without guessing; distinguish peak height from integrated area; point to the evidence behind a chemical name; recognise a blank and a standard; and explain why an unlabelled small feature should remain unassigned. These are better comprehension questions than asking students to memorise a picture’s visual appearance. Strong learning means being able to apply the same questions to a new chart with unfamiliar labels.
Next, practise translating a caption into a second language and then checking it against the original table. The target wording may change because technical register and word order differ. The numerical and evidentiary relationships should not change. If a bilingual learner can explain those relationships clearly in both languages, the exercise has moved from memorised vocabulary to genuine scientific communication.
Where this guide fits in the learning route
Chromatograms present a distinctive translation problem: separation coordinates, detector response and analytical inference coexist in one compact graphic. A general guide to translating graphs cannot fully teach method-dependent peak identity, while a chemistry lesson may not focus on what happens to certainty when scientific prose crosses languages. This article owns that intersection. It is intended to complement, not replace, laboratory education or professional analytical validation.
For a wider approach to translating meaning, follow eduKateSG’s Master Art of Translation. For the words that make precise explanation possible, continue to the vocabulary and English-learning resources. Technical readers can also compare this method with the related guides on scientific waveforms and chemical reaction schemes. The recurring principle is that a well-formed target sentence should never become more certain than the source evidence allows.