To translate scientific notation accurately, you must preserve magnitude, exponent sign, unit scale and the relationship between the coefficient and the power of ten. A value written as 3.2 × 106 is three million two hundred thousand, while 3.2 × 10−6 is three millionths. The visible difference may be a single minus sign, yet the quantities differ by a factor of one trillion. That makes scientific notation, powers of ten and exponent signs high-risk translation details rather than decorative mathematical formatting.
This guide explains how to translate scientific notation, engineering notation, E notation and SI prefixes without changing the size of a number. It is designed for students, translators, editors, researchers and technical writers working with values such as 6.02 × 1023, 4.7E−3, 250 µm, 3.3 kΩ and 12 MW. The central translation question is not only “What does this symbol mean?” but “What physical or numerical scale does this symbol create?”
A reliable translation separates four layers before rewriting the sentence: the numerical coefficient, the power of ten, the unit and the source precision. Once those are stable, you can adapt decimal separators, spacing, explanatory wording or unit names for the target audience. This prevents common errors such as turning milli into micro, changing E−6 into E+6, rewriting 0.00045 as 4.5 × 10−3, or adding decimal places that suggest measurement precision the source never had.
A fifty-second orientation
Scientific notation usually expresses a number as a coefficient multiplied by a power of ten. NIST’s conversion guidance also uses E notation, where a form such as 3.523907E−02 means 3.523907 × 10−2. NIST’s SI guidance explains that prefixes such as kilo, mega, milli and micro represent decimal powers of ten, and that currently recognised SI prefixes extend from 1030 to 10−30. These conventions are related, but they are not interchangeable pieces of typography.
Before translating a technical number, ask four questions: What is the coefficient? What is the exponent? What unit or prefix modifies the unit? What precision does the source actually claim? If any answer is unclear, pause before rewriting the expression. A translator can change presentation safely only after the represented quantity is understood.
Useful references include NIST’s Guide to the SI conversion factors, NIST’s SI prefix reference and NIST’s guidance on writing SI units. They provide a stable external check for notation, prefixes and powers of ten while the translation itself remains focused on preserving the source meaning.
1. Scientific notation is a relationship, not a string of symbols
Consider 4.8 × 105. The coefficient is 4.8 and the scale factor is 100,000, producing 480,000. A translation that keeps 4.8 but changes the exponent to 104 produces 48,000. A translation that keeps the exponent but changes the coefficient to 48 produces 4,800,000. Every element participates in the magnitude.
This is why a numerical QA pass should not compare digits in isolation. The reviewer should reconstruct the quantity. Ask whether the target expression expands to the same ordinary number as the source. If the source says 7.5 × 10−4, the decimal form is 0.00075. If the target expands to 0.0075, one power of ten has been lost.
Keep coefficient and exponent visually close during editing. Line breaks, superscript loss or copy-and-paste errors can separate them. In plain text, a form such as 7.5e-4 can be safer than a broken expression if the receiving system defines that notation. In polished prose, use the publication’s approved mathematical style, but preserve the semantic pair before worrying about typography.
2. The sign of the exponent controls direction of scale
A positive exponent multiplies by a power greater than one. A negative exponent divides by that power. Thus 2 × 103 equals 2,000, while 2 × 10−3 equals 0.002. The minus sign does not mean the quantity itself is negative; it belongs to the exponent and changes the scale.
This distinction matters in translation because minus signs can migrate visually. A line of source text might contain −2 × 10−3. The first minus sign makes the quantity negative; the second makes the exponent negative. Removing either one changes a different part of the meaning. A character-level check should therefore label each sign by function.
Do not replace a superscript minus with an ordinary dash if the resulting layout makes the exponent ambiguous. Conversely, do not assume every dash near a number is a negative sign; some may be range separators. The source structure must determine whether the mark belongs to the coefficient, exponent or surrounding sentence.
3. Moving the decimal point and changing the exponent must balance
The value 45,000 can be written as 4.5 × 104. If the coefficient is instead written as 45, the exponent must fall to 103 to preserve the same magnitude. The coefficient and exponent compensate for each other. Changing one without the other changes the value.
For small quantities, 0.000082 equals 8.2 × 10−5. A common error is to count visible zeros rather than count powers carefully. The safest translation workflow is to convert the source expression to a known reference quantity, then check the target expression against that reference.
Do not normalise a source expression purely for elegance if its chosen form serves a local purpose. A table may intentionally keep related numbers on similar exponent scales for comparison. Scientific notation can be mathematically equivalent in several forms, but presentation may encode grouping, instrument output or engineering convention. Preserve that intent where the source makes it visible.
4. Normalised scientific notation and engineering notation are different presentation systems
In normalised scientific notation, the coefficient is typically at least one and less than ten. Thus 12,500 becomes 1.25 × 104. Engineering notation commonly uses exponents that are multiples of three, so the same quantity may appear as 12.5 × 103. Both expressions represent the same value.
The difference matters because SI prefixes also advance in convenient powers, often three at a time for the familiar engineering sequence. 12.5 × 103 ohms can correspond naturally to 12.5 kΩ. A translator who forces every number into coefficient-less-than-ten scientific notation may make a technical table less compatible with its unit prefixes.
When the source clearly uses engineering notation, keep the convention unless the brief requests adaptation. Do not describe a non-normalised but valid engineering expression as a source error. First determine whether the exponent pattern is deliberate. Technical literacy prevents unnecessary “corrections” that make a translation mathematically equivalent but operationally less useful.
5. E notation is a machine-friendly exponent form
NIST documents examples such as 3.386389E+03 to represent 3.386389 × 103. The E introduces a base-ten exponent in that notation. It does not mean the mathematical constant e, and it is not automatically part of a unit name. Context determines which interpretation applies.
A spreadsheet may display 6.25E-07, while a typeset report shows 6.25 × 10−7. If the target publication permits typesetting, the form can change while the quantity remains fixed. If the field is machine-readable data, changing E notation into typographic mathematics may break import or validation. Translation and data transformation are separate jobs.
Protect plus and minus signs after E. E+06 and E-06 are radically different. Also protect leading zeros when the receiving syntax expects them, even though a human reader may understand E+6. A translator should not “clean up” machine notation unless the destination format has been verified.
6. SI prefixes are powers of ten attached to units
The SI prefix kilo means 103; mega means 106; giga means 109. On the smaller side, milli means 10−3, micro means 10−6 and nano means 10−9. The prefix modifies the unit scale, so 1 mg is 0.001 g while 1 µg is 0.000001 g.
Prefix symbols are case-sensitive. M and m do not mean the same thing. In many SI contexts, M denotes mega and m denotes milli. A translator, font substitution or automatic case-normalisation rule that changes capitalisation can therefore alter magnitude by a factor of one billion.
Keep the prefix attached conceptually to the unit. When a line breaks between them, a later editor may misread the prefix as a variable or stray letter. In running prose, the full unit name can sometimes be clearer for non-specialists, but the numerical conversion must be verified if the representation changes.
7. Micro needs typographic care
The micro prefix is represented by µ in SI notation. In some systems, the Greek letter mu and the micro sign may look similar or be normalised differently. A font or encoding problem can replace the symbol with an empty box, question mark or another character, leaving the magnitude unclear.
For a source value of 25 µm, the quantity is twenty-five micrometres, not twenty-five millimetres. Replacing µ with m changes the scale by a factor of one thousand. Replacing the whole unit with “m” changes micrometres into metres, a factor of one million.
When a platform cannot safely display µ, follow the target system’s approved plain-text convention rather than improvising. The key requirement is that the fallback be defined and unambiguous. A reader should not have to infer the prefix from context when the number may drive a technical decision.
8. The newest SI prefixes extend the same power-of-ten logic
NIST lists twenty-four recognised SI prefixes ranging from 1030 to 10−30, including quetta, ronna, ronto and quecto adopted in 2022. A translator encountering these less familiar prefixes should verify them rather than replacing them with a better-known nearby term.
For example, quetta uses Q for 1030, while quecto uses q for 10−30. Their symbols differ only by case, yet their scales lie sixty powers of ten apart. This is an extreme demonstration of the same principle seen with mega and milli: typography can carry enormous mathematical weight.
Do not localise official prefix symbols by translating their first letter. Unit systems rely on shared notation precisely so values remain portable across languages. Translate explanatory names where appropriate, but keep the recognised symbol required by the technical context.
9. Decimal prefixes and binary prefixes should not be merged
NIST notes that SI prefixes refer to powers of ten and should not be used to indicate powers of two. Thus kilo corresponds to 103, while binary prefix systems use forms such as kibi for 210. A storage or memory document may therefore distinguish kB from KiB depending on its chosen convention.
A translator should preserve the source distinction rather than rewriting every familiar-looking prefix into one house style. If a product specification says 512 MiB, changing it to 512 MB without an authorised conversion changes the nominal capacity. The difference may matter in software, hardware and data-transfer contexts.
Where the source uses ambiguous legacy terminology, do not silently resolve it according to your preferred standard. Check the product documentation or ask the owner. A translation should expose uncertainty that affects quantity rather than bury it under familiar-looking notation.
10. Prefix conversion and exponent conversion must agree
Suppose a source gives 4.7 × 10−3 volts. Since milli represents 10−3, the same quantity can be written as 4.7 mV. If the target uses 4.7 µV, it has shifted another factor of one thousand smaller.
The safest conversion writes the prefix as an explicit power first. Replace mV with 10−3 V in the working notes, perform the arithmetic, then apply the target prefix. This intermediate step prevents symbol familiarity from replacing actual calculation.
Reverse-check the result. Convert the target prefix back to the base unit and compare it with the source’s base-unit value. If both resolve to the same magnitude, the conversion is internally coherent. This simple check catches many silent prefix errors before publication.
11. Squared and cubed units change how prefixes scale
A centimetre is 10−2 metre, but a square centimetre is 10−4 square metre because the length factor is squared. A cubic centimetre is 10−6 cubic metre because the factor is cubed. The prefix belongs inside the exponent of the unit expression.
Therefore 1 cm² is not 0.01 m². It is 0.0001 m². A translation that converts the prefix linearly while preserving the square symbol produces an area one hundred times too large. The same dimensional logic applies to volume and other powered units.
Before converting, parse the whole unit: mm², cm³, km² and similar forms are not ordinary linear units with decorative superscripts. The superscript changes the conversion rule. This point connects directly to the existing Translate guide to square metres, square feet, hectares and acres.
12. Compound units can place powers of ten in numerator and denominator
A quantity such as mg/L contains a prefixed mass unit in the numerator and a volume unit in the denominator. Changing mg to g multiplies the numerical relationship by one thousand unless the number is adjusted. The slash is part of the meaning because it tells you which quantity is being divided by which.
For a fictional concentration of 250 mg/L, the equivalent is 0.250 g/L. Writing 250 g/L changes the concentration by a factor of one thousand. A translation can preserve every digit and still be wrong because the prefix changed without the coefficient.
Parse compound units from the inside out. Identify each prefix, unit and exponent, then reconstruct the relationship. This is particularly important when target-language prose replaces a slash with “per,” because the word order can shift. Preserve the mathematical direction even when the sentence order changes.
13. Prefixes do not automatically change when the surrounding language changes
Unit symbols are designed to be internationally recognisable. A target language may have a different word for metre, gram or second, but the technical symbol can remain m, g or s according to the adopted notation. Translating prose does not mean translating every character in the unit system.
Conversely, a document aimed at beginners may write “kilometres” in full rather than “km.” That is a presentation choice, not a numerical conversion. Keep symbol expansion separate from unit conversion so the number does not change unnecessarily.
A terminology glossary should therefore store both the concept and the approved representation. “microgram” may have a target-language word form, while µg remains the technical symbol. This reduces the chance that an editor replaces a shared unit symbol with a translated abbreviation that no longer follows the standard.
14. Decimal separators and exponent notation can interact badly
A source may write 1.25 × 104 using a decimal point, while a target locale convention may use a comma in ordinary numbers. If the publication adapts decimal separators, ensure the multiplication sign and exponent remain visually distinct. The reader must still see one coefficient multiplied by a power of ten.
Do not introduce a thousands separator into an E-notation field unless the receiving system explicitly allows it. Machine-readable numeric syntax is often less flexible than reader-facing prose. Localising a decimal mark can therefore be safe in one layer and invalid in another.
Keep a machine value and a display value as separate fields in the working record where necessary. That prevents a translator from using a visually localised string as though it were application data. Good localisation respects both numerical meaning and the syntax contract of the destination.
15. Significant figures travel with the measurement
A source value of 3.2 × 105 may carry two significant figures, while 3.200 × 105 may communicate more reported precision depending on context. Converting both to 320,000 can hide that distinction because ordinary notation may not show which trailing zeros are significant.
Do not add zeros after conversion merely because a calculator displays them. A value of 4.5 mm should not become 0.004500000 m unless the source truly supports that precision. The numerical quantity may be equivalent, but the target would imply a more exact measurement.
When precision matters, preserve scientific notation or an explicit uncertainty rather than relying on ambiguous trailing zeros. The broader translation issue is covered in the existing Translate guide to measurement tolerances, uncertainty and significant figures.
16. Zero is not the same as an extremely small nonzero quantity
A source value of 2 × 10−12 is very small, but it is not zero. A display rounded to insufficient decimal places may show 0.000000000000 or simply 0, depending on formatting. If the target sentence then says “the value was zero,” it has converted a display limitation into a factual claim.
Translate words such as undetectable, negligible, below threshold and zero according to the source. They describe different conditions. A quantity can be below a detection limit without being physically absent, and a rounded display can show zero without the stored value being exactly zero.
Scientific notation is often used precisely because it preserves small nonzero magnitudes compactly. Removing it for readability can erase important distinctions. Choose a reader-friendly explanation that retains the nonzero quantity when the source does.
17. Overflow, underflow and clipped displays need source awareness
Software may display very large or very small numbers in exponent notation automatically. Another system may round, clip or replace them with special indicators. A translator should recognise that the visible representation may be generated by software rather than manually authored prose.
If a target screenshot or interface changes 9.8E+308 into an error marker, the problem may be technical rather than linguistic. Do not rewrite the accompanying explanation as though the source value itself were invalid without investigating the application behaviour.
The translation workflow should therefore preserve source strings during ingestion and compare them with rendered outputs after localisation. This is especially important when fonts, number formatters or spreadsheet imports are involved. A value can be correct in storage but misleading on screen.
18. Worked case: a measurement table with mixed notation
Consider this fictional source table in prose form: “Sample A: 0.0045 g. Sample B: 4.5 mg. Sample C: 4.5 × 10−3 g. Sample D: 4.5E−3 g.” Under the stated units, all four entries describe the same mass.
A translator who standardises all four as 4.5 mg may improve comparability if the brief permits unit normalisation. But that is an editorial transformation, not merely word translation. If the purpose is to preserve the original reporting forms, each representation should remain recognisably distinct.
The private QA check is simple: convert every entry to grams and confirm 0.0045 g. Then decide separately how the target publication should display them. Verification can be uniform even when presentation remains varied.
19. Worked case: a prefix error hidden by fluent prose
Use this invented source: “The sensor detects changes as small as 8 µV.” A flawed target says “The sensor detects changes as small as 8 mV.” The sentence sounds completely natural. Only one character differs, yet the translated threshold is one thousand times larger.
A correct review expands both units to volts. Eight microvolts is 8 × 10−6 V. Eight millivolts is 8 × 10−3 V. The base-unit comparison exposes the error immediately.
This is why technical translation QA should include magnitude checks rather than rely on fluent reading. The eye tends to recognise familiar unit shapes and move on. Expanding critical values into powers of ten forces the reviewer to inspect the scale explicitly.
20. Worked case: engineering notation and component values
A fictional component specification gives 4.7 × 103 Ω. This is 4.7 kΩ. A second line gives 4700 Ω. A third gives 4.7 kΩ. These representations can describe the same resistance under the stated notation.
A flawed translation might produce 4.7 MΩ by misreading k as an ordinary abbreviation rather than the kilo prefix. The component value then changes from thousands of ohms to millions of ohms. The existing Translate guide to resistance and capacitance values covers that specialist context in more depth.
The general lesson is transferable: when engineering notation aligns naturally with an SI prefix, use that alignment as a cross-check. The exponent and prefix should agree. If 103 becomes mega, or 106 becomes kilo, the translation has broken the scale.
21. Worked case: scientific notation in a sentence, not a table
Suppose an invented research sentence says: “The estimated count is approximately 2.4 × 107, with substantial uncertainty.” The translation must preserve the magnitude, the approximation marker and the uncertainty statement. Converting the number to “24 million” can be valid for a general audience if the editorial brief permits it.
However, “approximately 24 million” should not become “exactly 24 million,” and a target reader should not infer that all seven digits are known. The word approximately may carry more practical meaning than the notation itself.
If the surrounding paper uses scientific notation consistently, retaining the source form may be better than converting one number into words. Translation quality includes document coherence. The best local rewrite should still fit the information system around it.
22. Practice clinic with explained answers
Practice one. Convert 6.4 × 103 to ordinary notation. The answer is 6,400. The decimal point moves three places to the right because the exponent is positive.
Practice two. Convert 6.4 × 10−3 to ordinary notation. The answer is 0.0064. The negative exponent moves the scale in the opposite direction; it does not make the number itself negative.
Practice three. Express 0.000091 in normalised scientific notation. The result is 9.1 × 10−5. The coefficient is between one and ten, and expanding the expression returns the source value.
Practice four. Translate 2.5E+06 into ordinary notation for an explanatory note. It represents 2,500,000. Do not interpret E as the constant e or remove the positive exponent sign before understanding the field.
Practice five. Convert 3.6 × 10−3 metres to millimetres. Since milli is 10−3, the answer is 3.6 mm. A target of 3.6 µm would be one thousand times smaller.
Practice six. Convert 7.2 MW to watts. Mega is 106, so the result is 7.2 × 106 W. Lowercase m would not be an equivalent prefix.
Practice seven. Convert 1 cm² to m². One centimetre is 10−2 metre, so the area is 10−4 m². Applying only the linear factor would produce the wrong area.
Practice eight. A value is written 5.0 × 10−9 A. Expressed with an SI prefix, it is 5.0 nA because nano is 10−9. Preserve the reported 5.0 rather than reducing it to 5 if the trailing zero carries intended precision.
Practice nine. The source says 750 µg. In milligrams, the equivalent is 0.750 mg. Preserve the source precision according to the publication’s rules rather than adding arbitrary digits.
Practice ten. A dataset uses 1.2E-05 while a report uses 1.2 × 10−5. The two can be numerically equivalent. Do not replace the machine-readable field simply because the report form looks more elegant.
23. A release checklist for powers of ten and prefixes
Check exponent signs first. Then check coefficient digits and decimal position. Then expand critical values into ordinary notation or a base unit. After that, verify every SI prefix, including case. Finally, confirm significant figures and any words such as approximately, at least, below or estimated.
For compound units, check numerator and denominator separately. For squared or cubed units, apply the power to the conversion factor. For machine-readable notation, confirm that localisation has not changed syntax. For reader-facing prose, confirm that any adapted notation still represents exactly the same magnitude.
This layered review is faster than trying to notice every possible error while reading for style. Magnitude, units, precision and prose each get their own pass. The method reduces cognitive load and makes the final language more trustworthy.
24. Where this article sits in the eduKate translation system
This specialist guide extends the factual discipline in Translate | Names, Numbers, Dates and Units. The wider architecture remains Master Art of Translation. For exact word distinctions and technical terminology, use the Vocabulary Learning Hub; for how English attaches modifiers, quantities and conditions to one another, use How English Works.
The governing principle is simple: scientific notation compresses magnitude, but translation must not compress away the relationships that create that magnitude. Preserve the coefficient, exponent, unit and precision first. Once those anchors are secure, the surrounding language can become as natural, clear and reader-friendly as the target context requires.
Before release, take one final representative sample from the article or document and convert it all the way back to a base quantity. If the target returns to the same value as the source, the scale has survived. If it does not, stop and repair the numerical layer before polishing another sentence.