To translate logarithmic scales, decibels, pH values and orders of magnitude accurately, you must preserve the multiplicative relationship hidden behind an additive-looking scale. A difference of 1 on a base-10 logarithmic scale can represent a tenfold ratio in an underlying quantity. A change of 3 decibels does not simply mean “three more units of sound.” A fall in pH does not mean acidity changed by the same additive amount as the printed numbers. Logarithmic scales compress wide ranges, and translation fails when the compressed scale is read like an ordinary linear ruler.
This guide explains how to translate log scale, logarithmic values, decibels (dB), pH, log-transformed data, powers of ten and orders of magnitude in scientific reports, technical manuals, charts, research papers, educational material and data dashboards. It solves a specific search intent: how do you move a logarithmic statement between languages without turning a ratio into a difference, losing the reference quantity, reversing a pH direction, confusing power and amplitude decibels, or treating “log” as though it always meant base 10?
The safest method is to record four elements before translating: the underlying quantity, the logarithm base or defined scale, the reference value and the direction of interpretation. Then reconstruct the sentence in the target language while preserving those elements. NIST’s work on the bel and decibel emphasises that logarithmic quantities depend on defining equations and reference relationships, while NIST’s pH metrology materials show that pH is a specialised measurement system with its own traceability and conventions. This article also connects conceptually to eduKateSG’s guides on scientific notation and powers of ten and measurement scales.
A sixty-second orientation: logs turn multiplication into addition
A logarithm answers a question about powers. If 10²=100, then log10(100)=2. Multiplying an underlying quantity by 10 adds 1 to its base-10 logarithm. Multiplying by 100 adds 2. This is why logarithmic scales can represent enormous ranges compactly.
Translation must therefore distinguish a difference on the log scale from a difference in the original quantity. If log10(x) rises from 2 to 3, x rises from 100 to 1,000: a tenfold multiplication, not an increase of one original unit. If the target sentence says “increased by one” without naming the log scale, readers can easily misinterpret the result.
1. “Log” is incomplete unless the base or convention is known
The notation log can mean different bases in different fields. In some mathematical writing, log means the natural logarithm, base e. In other contexts, especially elementary science or engineering, log may mean base 10. Computing systems may label functions explicitly as ln, log10 or log2, but conventions vary.
Do not add a base in translation unless the source or field convention supports it. A sentence “values were log transformed” may be insufficiently specific; the methods section or software code may identify the actual transformation. Translating it as “converted to base-10 logarithms” would invent information if the authors used natural logs.
If the base matters to interpretation and is not specified, raise a focused query. “Does log in this analysis mean natural log or base 10?” is better than guessing from typography. Base choice affects coefficients, axis labels and back-transformation even when model fit may be mathematically related.
2. Natural log, common log and binary log belong to different conventions
The natural logarithm ln uses base e. The common logarithm uses base 10. Binary logarithms use base 2 and appear often in information theory and computing. A unit increase means multiplication by a different factor in each case: e for ln, 10 for log10, and 2 for log2.
Suppose a fictional variable has ln(x)=2 and later ln(x)=3. The underlying ratio is e³/e²=e, approximately 2.718, not 10. If the source instead reports log10(x) changing from 2 to 3, the ratio is 10. Translating both as “one order higher” would be wrong in the natural-log case because order-of-magnitude language usually refers to powers of ten.
Preserve subscripts, function names and software notation. A typesetting system that drops the 10 from log10 can create an ambiguity the source did not have. Mathematical typography is part of translation accuracy.
3. Order of magnitude is a power-of-ten idea, not a synonym for “a lot”
In technical writing, an order of magnitude usually refers to a factor of about ten. Two orders of magnitude correspond to about a factor of one hundred. The phrase is sometimes used loosely in everyday speech to mean much larger, but a scientific translation should preserve the technical meaning when that is what the source intends.
If a concentration increases from 2×10⁻⁶ to 2×10⁻⁴, it increases by two powers of ten, or a factor of 100. Calling that “two times larger” would understate the change dramatically. Conversely, a value rising from 2 to 4 doubles but does not increase by an order of magnitude.
Use explicit factors when readers may not know the phrase. “About two orders of magnitude, or roughly one hundred times” can clarify an educational text. Do not add a precise factor if the source itself uses order-of-magnitude language approximately.
4. A log axis is not a linear axis with unusual tick spacing
On a base-10 logarithmic axis, equal visual distances can represent equal ratios rather than equal absolute differences. The interval from 1 to 10 can occupy the same axis length as 10 to 100 or 100 to 1,000. A reader who assumes linear spacing can misread the size of changes.
Translate axis titles and captions explicitly when needed: logarithmic scale, log10 scale, log-transformed axis. Do not remove “log” to shorten a chart heading. That one word tells the reader how to interpret distances and slopes.
If the source uses a semilog plot, one axis is logarithmic and the other linear. A log-log plot uses logarithmic scaling on both axes. Those chart types can reveal exponential or power-law relationships. Translating both simply as “log chart” loses structural information.
5. Zero and negative values need special treatment on ordinary log scales
The ordinary real logarithm is undefined for zero and negative inputs. Therefore a standard log-transformed variable requires positive values unless the analysis uses a different transformation or convention. A graph with a log axis cannot place an ordinary zero at a finite position.
A target editor should not “fix” an axis by replacing missing zeros with a small positive number unless the source method defines that procedure. Adding a pseudocount such as 1 before taking logs is a substantive transformation that can affect small values strongly.
Some visualisations use symmetric log scales or other transformations that accommodate zero and negative values. Preserve the exact transformation name. “Logarithmic” alone may be insufficient when the scale is modified.
6. Back-transforming a log value restores the original scale
If y=log10(x), then x=10ʸ. If y=ln(x), then x=eʸ. Back-transformation matters when translating model output because a coefficient on a log scale may correspond to a multiplicative effect on the original scale.
Suppose ln(x) differs by 0.693 between two conditions. Since e0.693 is approximately 2, the original quantity differs by a factor near two. Translating “a 0.693-unit increase” as if it were a small additive change in x would miss the transformation.
Keep the source scale visible when reporting transformed statistics. “Mean log concentration” is not the same quantity as “log of the mean concentration,” and neither is automatically the arithmetic mean on the original scale. Word order and parentheses can carry mathematical meaning.
7. Geometric means often appear naturally with log-transformed data
For positive values, averaging logs and back-transforming produces a geometric mean. The geometric mean is useful for multiplicative data and ratios, but it is not the same as the arithmetic mean. A translation that drops geometric changes the statistic.
Consider values 1 and 100. Their arithmetic mean is 50.5, while their geometric mean is 10. The difference is enormous. If a report summarises concentrations with a geometric mean because the distribution is log-normal or multiplicative, translating it as “average” without qualification can mislead readers.
Use the established target-language term for geometric mean and preserve confidence intervals or back-transformed limits accordingly. Do not assume that every reader will infer the statistic from the fact that logs were used.
8. Decibels express a logarithmic ratio relative to a reference
The decibel is used to express logarithmic levels or ratios. Its meaning depends on what physical quantity is compared and which reference value is used. NIST discusses the bel and decibel in relation to logarithmic decay or gain and emphasises the defining equation behind the quantity.
A translation should therefore preserve both dB and the reference context. “60 dB” by itself is incomplete in many technical settings because different decibel quantities can use different references: sound pressure level, power level, voltage gain or another defined level.
Do not translate dB as if it were an ordinary unit of absolute sound amount. The same numerical dB value can refer to different physical quantities depending on the stated reference and measurement convention.
9. Power ratios commonly use 10 log10
For a power ratio P₂/P₁, a common decibel expression is 10 log10(P₂/P₁). A tenfold power ratio gives 10 dB. A hundredfold ratio gives 20 dB. Doubling power gives approximately 3.01 dB.
This means “3 dB more” is not “three times the power.” It is close to twice the power under that relationship. A translation that replaces dB difference with a direct multiplier should calculate the correct ratio and name the physical quantity.
When the source says “gain of 10 dB,” preserve whether it is a power gain, system gain or another defined level. Do not automatically interpret it as perceived loudness. Engineering decibels and human perception belong to related but distinct discussions.
10. Amplitude-like ratios may use 20 log10 under stated conditions
When power is proportional to the square of an amplitude quantity under the relevant conditions, decibel relationships may be written with 20 log10 of the amplitude ratio. Examples can involve voltage or sound pressure under appropriate reference and impedance assumptions.
A doubling of amplitude then gives approximately 6.02 dB, while a doubling of power gives approximately 3.01 dB. Translating “doubling” without naming what doubled can therefore produce the wrong decibel interpretation.
Do not memorise 10-versus-20 as a language rule divorced from physics. Read the source equation and quantity. A translator should preserve the stated formula or reference standard rather than infer it from the word amplitude alone.
11. dB, dBm, dBW and dB SPL are not interchangeable labels
Plain dB often expresses a ratio, while dBm and dBW use specific power references, and dB SPL uses a sound-pressure reference. These suffixes and qualifiers matter. Removing them can convert an absolute referenced level into an unspecified ratio or vice versa.
If a technical table lists 10 dBm, do not shorten it to 10 dB for visual consistency. dBm references a defined power level. Similarly, a sound-pressure level in dB SPL should not be rewritten as generic dB if the reference matters to interpretation.
Keep unit strings protected in data and formulas. Translate surrounding explanations—such as “sound pressure level”—but preserve internationally recognised technical symbols unless the applicable standard specifies another presentation.
12. Adding decibels can correspond to multiplying ratios
Because logarithms convert multiplication into addition, cascaded gains expressed in decibels can often be added when they refer to compatible quantities and definitions. Two stages with gains of 10 dB and 20 dB can combine to 30 dB under the appropriate system model.
This does not mean two independent sound sources at 60 dB simply produce 120 dB. Combining levels from independent sources requires working with underlying intensities or powers under the relevant assumptions. Equal independent sources can produce only a few decibels of increase, not an arithmetic sum of the level numbers.
Translation should distinguish “gain stages add in dB” from “sound levels add.” The same verb add can describe different operations. Preserve the physical setup and calculation rule supplied by the source.
13. pH direction is easy to reverse in translation
pH is related logarithmically to hydrogen-ion activity under its measurement convention. Lower pH generally corresponds to greater acidity in aqueous contexts; higher pH corresponds toward greater basicity. Because of the negative logarithmic relationship, a numerical decrease in pH can represent an increase in hydrogen-ion activity.
A target sentence saying “pH increased, so acidity increased” can reverse the source interpretation. Preserve the actual variable: pH increased or decreased, acidity increased or decreased, hydrogen-ion activity increased or decreased. Do not substitute one phrase for another without applying the correct direction.
NIST’s pH metrology work emphasises traceable measurement and reference materials. In technical translation, keep pH values, calibration conditions and method descriptions intact rather than treating pH as an informal taste or colour scale.
14. A one-unit pH difference represents a multiplicative change in hydrogen-ion activity
Under the simplified base-10 relationship commonly taught for pH, a difference of one pH unit corresponds to a factor of ten in hydrogen-ion activity, with direction reversed because of the negative logarithm. A solution at pH 4 has about ten times the hydrogen-ion activity represented by pH 5 under the simple comparison.
Therefore pH 4 is not “one degree more acidic” in a linear sense. Nor should pH 4 be described as “25% more acidic” than pH 5 based on the printed numbers. The underlying ratio is multiplicative.
For precise chemical work, pH measurement depends on activity, temperature, calibration and conventions beyond the classroom simplification. A translation should preserve the source’s technical level rather than replacing a laboratory method with an elementary concentration formula.
15. pH 6 is not “twice” pH 3
Printed pH numbers tempt ratio language. Six is twice three, but that arithmetic relationship does not mean a pH-6 solution has twice the pH-property quantity represented by a pH-3 solution. The scale is logarithmic and reversed relative to hydrogen-ion activity.
Under the simplified base-10 model, the difference of three pH units corresponds to a thousandfold ratio in hydrogen-ion activity. The pH-3 solution has much greater hydrogen-ion activity than the pH-6 solution, not half as much.
A translator should therefore avoid ordinary ratio adjectives applied directly to pH numbers. Use higher pH, lower pH, more acidic, less acidic or a stated activity ratio according to the source.
16. pH uncertainty and rounding matter because the scale is logarithmic
A change from pH 7.00 to 7.10 may look small, yet on the simplified logarithmic relationship it represents a multiplicative change in the underlying hydrogen-ion activity. Rounding pH too coarsely can therefore hide meaningful ratios in sensitive applications.
Preserve decimal places according to the source measurement and reporting method. Do not add digits that the instrument or calibration does not justify, and do not remove them solely for stylistic consistency with non-logarithmic variables.
If a source gives pH with uncertainty or calibration information, keep it. Logarithmic scales do not exempt a measurement from ordinary rules about precision, traceability and significant figures.
17. Earthquake magnitude requires modern terminology, not a generic “Richter scale” label
Popular writing often calls any earthquake magnitude the Richter scale, but modern seismology uses several magnitude measures, including moment magnitude for many reported large earthquakes. A translation should preserve the magnitude type used by the source rather than replacing it with the familiar word Richter.
Historical local magnitude associated with Richter is logarithmic, and magnitude differences represent multiplicative changes in measured wave amplitudes under the defined scale. But translating every “magnitude 7” as “7 on the Richter scale” can be technically wrong if the source reports moment magnitude Mw.
Keep magnitude, intensity and energy distinct. Earthquake intensity describes observed effects at locations under intensity scales; magnitude estimates event size. A target that calls magnitude “intensity” can change the scientific quantity even without changing the number.
18. Logarithmic scientific scales may use different definitions
Not every logarithmic scale uses base 10 in the same formula, and not every unit increase corresponds to the same underlying ratio. The definition of the scale determines the interpretation. Decibels include a multiplier and reference ratio; pH includes a negative logarithmic relationship; astronomical magnitude uses its own historical convention.
Do not generalise “one unit means ten times” to every logarithmic scale. Identify the defining equation or authoritative scale description. Translation should carry the name and direction of the scale, not an oversimplified rule learned from another domain.
A useful glossary records both term and transformation: “decibel power ratio: 10 log10(P2/P1)” or “pH: logarithmic acidity-related scale; lower value corresponds to greater hydrogen-ion activity under the defined convention.” This protects later translators from cross-domain assumptions.
19. Log-transformed regression coefficients need the model form preserved
Regression models can use logged outcomes, logged predictors or both. The interpretation of a coefficient depends on which variables are transformed. A log-linear model differs from a linear-log model and a log-log model. Translation should preserve the formula or model description accurately.
If the outcome is ln(y) and x is unlogged, a one-unit change in x corresponds to a multiplicative change in y after exponentiating the coefficient, with exact interpretation depending on model assumptions. If both x and y are logged, the coefficient can often be interpreted as an elasticity in appropriate contexts.
Do not simplify “log-transformed outcome” to “adjusted outcome.” Transformation and adjustment are different operations. Likewise, “exponentiated coefficient” should not become “exponential model” unless the source actually specifies that model form.
20. Percent change after a log transformation needs careful wording
Researchers sometimes translate log coefficients into approximate percentage changes. For small coefficients in natural-log models, 100β% may be used as an approximation in certain settings, while the exact multiplicative change is based on eβ. The article or discipline may specify which interpretation is used.
A translator should not create percentage language from a log coefficient unless the source already interprets it or the editorial task explicitly includes mathematical explanation. Approximate and exact transformations can differ more as coefficients grow.
Preserve words such as approximately, associated with and holding other variables constant. Dropping those qualifiers can turn a model coefficient into a direct causal percentage claim.
21. Fold change and log fold change are different quantities
In biology and other data-heavy fields, fold change expresses a ratio, while log fold change expresses the logarithm of that ratio. A twofold increase corresponds to fold change 2. On a base-2 log scale, log2 fold change is 1. A halving gives fold change 0.5 and log2 fold change −1.
Translation must preserve the word log and the base when stated. “Log2 fold change = 2” means a fourfold ratio, not a twofold ratio. A target table that drops the subscript can double the interpretation error.
Also distinguish fold increase from “times more,” which can be linguistically ambiguous. When exact quantitative interpretation matters, state the ratio directly or use the discipline’s established terminology.
22. Negative log values do not mean negative original quantities
A positive quantity smaller than 1 can have a negative logarithm. For example, log10(0.01)=−2. The original value is positive. The negative sign belongs to the transformed scale, not to the physical quantity itself.
A translator unfamiliar with logarithms may describe a negative log concentration as a “negative concentration.” That is wrong. The source variable is log concentration, and negative transformed values can be perfectly valid.
Always name the transformed variable when the sign could confuse readers. “The log10 value was −2, corresponding to 0.01 on the original scale” is unambiguous in educational material.
23. Ratios below one produce negative log ratios
If a ratio is less than one, its logarithm is negative. This is useful because log ratios become symmetric around zero: a doubling and a halving can appear as equal-magnitude positive and negative values on a base-2 log scale.
For example, log2(2)=1 while log2(0.5)=−1. A translation that calls −1 “one unit less” without naming the log ratio may hide the underlying halving relationship. Use multiplicative language when the source provides it.
This symmetry is one reason log ratios are common in scientific graphics. Preserve zero lines, legends and direction labels so readers know whether positive values indicate increase, enrichment, gain or another domain-specific condition.
24. Worked example: reading a logarithmic chart
Consider a fictional chart whose y-axis is labelled “Concentration (µg/L, log10 scale)” with ticks at 1, 10, 100 and 1,000. Sample A plots at 10 and Sample B at 100. On the graph, the vertical gap between A and B equals the gap between 100 and 1,000 because both represent a tenfold ratio.
A poor translation might relabel the axis simply “Concentration (µg/L)” and leave the tick spacing unchanged. The chart would visually imply a linear scale even though its geometry is logarithmic. Another might say B is “90 units higher,” which is arithmetically true on the original concentration scale but ignores the chart’s ratio-based visual interpretation.
A reliable translation preserves log10 scale in the axis title and, if the audience needs help, explains that equal vertical steps correspond to tenfold changes. The data points themselves remain 10 and 100 µg/L; the axis transformation changes visual spacing, not the measured units.
25. Worked example: decibel gain
Use a fictional two-stage system. Stage A provides a 10 dB power gain and Stage B provides a 20 dB power gain under compatible definitions. The combined gain is 30 dB. In underlying power-ratio terms, 10 dB corresponds to a factor of 10 and 20 dB to a factor of 100; the combined ratio is 1,000, corresponding to 30 dB.
A faulty target might say the gains multiply to 200 dB because 10×20=200. Another might convert 30 dB into “thirty times the power.” Both misread the logarithmic representation. The correct relation is additive in dB because the underlying ratios multiply.
This example assumes the same compatible power-gain convention. Real systems may include losses, impedance conditions and frequency dependence. Translation should preserve the source’s engineering model rather than use this simplified teaching case as a universal calculation rule.
26. Worked example: pH movement
Consider a simplified educational source: “The pH decreased from 7 to 5. Under the simplified hydrogen-ion relationship, this corresponds to about a hundredfold increase in hydrogen-ion activity.” The numerical pH change is −2, but the underlying multiplicative change goes in the opposite direction because pH uses a negative logarithmic relationship.
A flawed translation might say “acidity decreased by two units” because the pH number decreased. Another might say “hydrogen-ion activity doubled.” Both are wrong under the simplified model. Two base-10 log units correspond to a factor of 100.
For a laboratory report, preserve the actual measurement method and avoid adding the simplified ratio unless the source states it. Educational explanation and metrological reporting require different levels of precision.
27. Worked example: log2 fold change
A fictional gene-expression table reports log2 fold change = 1.5. The underlying fold ratio is 21.5, approximately 2.83. This means the quantity is about 2.83 times the reference under the defined calculation.
If log2 fold change = −1.5, the underlying ratio is about 0.354, meaning the quantity is roughly 35.4% of the reference, or about 2.83-fold lower under one common way of describing the reciprocal relationship. The exact wording “fold lower” can be ambiguous, so many technical publications prefer direct ratios or percentage reductions.
A translator should preserve log2, sign and reference condition. Changing the reference reverses the sign. A positive value is not universally “good” or “upregulated” unless the source defines which condition is numerator and which is denominator.
28. Practice clinic: recover the underlying relationship
Practice one. log10(x) changes from 1 to 2. What happens to x? It increases from 10 to 100, a tenfold change, not an increase of one original unit.
Practice two. ln(x) changes by 1. Does x multiply by ten? No. It multiplies by e, approximately 2.718.
Practice three. A power level rises by 10 dB under the standard power-ratio relationship. Does power rise tenfold? Yes. Do not call it a 10% increase.
Practice four. A power ratio doubles. What is the dB difference? About 3.01 dB. Do not say 2 dB merely because the ratio is two.
Practice five. An amplitude ratio doubles under a 20 log10 relationship. What is the change? About 6.02 dB. Preserve the fact that amplitude, not power, is being compared.
Practice six. pH falls from 6 to 5. Under the simplified base-10 relationship, hydrogen-ion activity increases by about a factor of ten. Do not describe it as a one-unit decrease in acidity.
Practice seven. A base-2 log fold change equals 2. What is the fold ratio? Four. Preserve the log base.
Practice eight. A log-transformed variable has value −3. Is the original quantity negative? Not necessarily. For log10, it could correspond to a positive value of 0.001.
29. Frequently asked translation questions
Does “log” always mean base 10? No. Conventions vary. Check the field, notation, methods or software.
Is a 3 dB increase three times as much? Not under ordinary power-ratio interpretation. About 3 dB corresponds to roughly double the power. The physical quantity and reference must be known.
Can I add two sound levels directly? Not as ordinary numbers unless the source’s system definition justifies it. Independent levels are generally combined through underlying intensities or powers, not simple arithmetic addition of dB values.
Does lower pH mean less acidic? In ordinary aqueous interpretation, lower pH corresponds to greater acidity and higher hydrogen-ion activity, subject to the scale’s technical conventions.
Does one order of magnitude mean ten percent? No. It refers to roughly a factor of ten.
Can log values be negative? Yes. Positive original values between zero and one have negative logarithms for bases greater than one.
What is the strongest release check? Ask: “What multiplicative ratio in the original quantity does this target statement imply?” If that ratio differs from the source, the logarithmic translation is wrong even if every printed log value was copied exactly.
30. Connect this specialist guide to the wider eduKateSG translation architecture
This article is a specialist branch of Master Art of Translation, not a replacement broad hub. It connects to Translate | Scientific Notation, Powers of Ten and SI Prefixes, Translate | Percentages, Percentage Points and Basis Points, Translate | Standard Deviation, Variance and Z-Scores, and the protected Vocabulary Learning Hub. For comparative grammar, quantifiers and scope, use How English Works.
Logarithmic translation is reliable when the target preserves the transformation, base, reference, sign and underlying ratio. A reader should be able to recover the same multiplicative relationship from the target that the source encoded. That is the deeper rule connecting decibels, pH, orders of magnitude, log axes and log-transformed models: the printed numbers are compressed representations. Translate the relationship they represent, not just the symbols on the page.