Technical Writing

Copyright 1988, 1999 by Ronald B. Standler

Table of Contents

my favorite style manuals
         Strunk & White
         disagreement about rigid rules
use of numbers in sentences
         zero and infinity
         six or 6 ?
use of units with numbers
         units of computer memory
equations in text
common misuse of words in English
         a or an ?
         the conjunction or
         use of pronouns
citations and bibliography
         What kind of literature should be referenced?
         format of citations and references
         citing sources you have not read
verb tense and voice
illogical rules in English


I began this document in 1978, when I was teaching undergraduate classes in electrical engineering and found that my students needed some guidance on grammar and style. I enlarged this document frequently during my ten years as a professor. During 1990-95, I shared this document with my colleagues in Germany who wanted to know more about style in American English. The favorable reaction of my former students and my German colleagues has encouraged me to post this document on the Internet, in the hope that it might be useful to a wider audience.

my favorite style manuals

Because I do not intend this document to be a comprehensive list of everything that one needs to know about grammar and style, I begin with an annotated list of my favorite reference and textbooks in this area. There are many good books — I have deliberately made this list short, because most scientists and engineers are not interested in technical writing.

G.L. Kittredge and F.E. Farley: An Advanced English Grammar

originally published by Ginn in 1913, reprinted by AMS Press in New York City in 1972.
Teaching formal grammar became unfashionable in many schools in the USA in the 1970s, which has had disastrous consequences for the ability of professionals in the USA to produce erudite prose. The lack of modern reference books means that one must rely on old grammar books, like this one. This old book even contains conjugation of the verb to be for four tenses of the subjunctive mood. There are still some judges that use subjunctive mood for contrary-to-fact statements in writing their opinions, but use of the subjunctive seems to be rare in most other writings by American authors.

Strunk & White

Strunk and White: The Elements of Style

first edition privately printed by Prof. Strunk at Cornell University in 1918. His former student, E.B. White, revised this book in 1959, when it was first published by Macmillan. I highly recommend this book for two reasons: (1) it is terse and (2) the rules make sense. To Strunk's rules, I would add one fundamental rule:
One should avoid constructions that are likely to be misunderstood, even if they are arguably "correct" according to a style manual or dictionary. It is not enough to make a document easy to read, it should also be difficult to misunderstand.
For example:
  1. Strunk and White's rule 15 about putting statements in positive form is just a special case of my fundamental rule.
  2. Say "flammable" instead of "inflammable", because the latter word is likely to be misunderstood by someone who thinks the in- prefix always means not.
  3. When writing for an international audience, which includes non-native speakers of English, make all sentences literally true (i.e., avoid idioms and colloquial phrases).

I want to mention here three of Strunk and White's rules that are often violated:
  1. My favorite rule (Strunk and White rule 17) is: "Omit needless words." Many words can be omitted from a first draft without affecting the meaning of sentences. The following italicized words are examples of needless words:
    definitely proved,
    orange in color
    viable alternative (it is not an alternative if it is nonviable!)
    worst-case maximum possible error.
    because of the fact that

  2. Strunk and White say (rule 11) that a "participial phrase at the beginning of a sentence must refer to the grammatical subject. Sentences violating rule 11 are often ludicrous." A simple, although overly cautious, rule is never begin a sentence with a word that has an "-ing" suffix. For example:
    "Assuming a << 1 and substituting in Eq. 12, the relation becomes Eq. 14."
    Equations neither assume nor substitute. It should read:
    "When a << 1, Eq. 12 can be simplified to become Eq. 14."

  3. Put statements in positive form (Strunk and White rule 15). Avoid statements like:
    "None of these integers were non-negative."
    Say instead:
    "All of these integers were negative."

The Chicago Manual of Style

This book, which is published by the University of Chicago Press, is a good general reference book on style in student papers and scholarly publications in the USA.

specific style manuals

One will also need the style manual for the particular journal in which one intends to publish, for example:

style for units of measurement

Everyone doing technical writing should have a copy of an authoritative style manual for units that lists all of the metric prefixes, abbreviations for units, etc. This information is available online from NIST or BIPM (Bureau International des Poids et Mesures).

disagreement about rigid rules

One of the distressing features of modern academic life in the USA is the fragmentation of disciplines, so that style manuals for scholarly publications diverge in what they consider correct form for citation to books and journals.

For example, consider a bibliographic citation for an article that is found in volume 23 of the Journal of Stuff, beginning on page 438.
Science magazine, published by the American Association for the Advancement of Science, and all of the journals published by the American Institute of Physics, would write:
J. Stuff 23, 438
The journals published by the Institute of Electrical and Electronics Engineers would write:
J. Stuff, vol. 23, p. 438
Some scientific journals use:
J. Stuff, 23:438
a legal journal would write:
23 J. STUFF 438

It makes no sense to me that physicists, engineers, physicians, and attorneys should use such radically different styles. Not only are the styles different, but editors often regard their preferred style as holy dogma that may be neither doubted nor disobeyed. Editors often treat anyone who disagrees with their style as an illiterate ignoramus. In contrast to editors, I regard many matters of style as the prerogative of authors. What is important is the information (i.e., volume number and page number), not whether the volume number is boldface or "vol. 23".

use of numbers in sentences

zero and infinity

There is no such thing as zero or infinity in laboratory measurements. When tempted to claim that some value is "zero" or "unmeasurably small", give an estimate of the smallest nonzero magnitude that could be detected. A concise way to phrase this is to write, e.g., "Vout was less than 5 mV", instead of a long-winded discussion of least significant digits, resolution, accuracy, and precision. If special care was taken to measure a small voltage (e.g., use of preamplifier, modulate noise with a signal of a much different frequency, etc.), then these techniques should be mentioned. When using "unmeasurably small" in an analysis and discussion, compute an upper or lower bound. Do not avoid a calculation just because a signal could not be detected.

six or 6 ?

In general, single digit numbers (zero through nine) are spelled out, while numbers of two or more digits (10, 20th Century, ...) use figures. There are several exceptions to this simple rule:
  1. Use figures for all numbers when there are numbers of two or more digits for related quantities in the same sentence , such as "6 of 23 physicians recommend".

  2. Always use figures when a unit of measurement follows (e.g., "5 A").

  3. Always spell approximate values, "round numbers", (see The Chicago Manual of Style, p. 233 of the 1982 edition). If the approximate value is followed by a unit, then the unit is not abbreviated.
    about five years
    two orders of magnitude
    about four times larger
    several kilovolts
    a few tens of megahertz
    a few volts

  4. Use figures when mathematical operations are implied:
    factor of 2
    3×3 matrix

  5. When a number is at the beginning of a sentence it is always spelled-out. If this rule produces a result that looks awkward, it may be better to rewrite the sentence to avoid starting with a number.

use of units with numbers

  1. All numerical values that have dimensions must have their units specified. In general, the units must follow the numerical value every time. However, in a table of numbers, the units may be specified at the top of the column, provided all of the values have the same units. If there are two numbers in a phrase with the same units (e.g., "frequency between 4 and 5 kHz"), then put the units only after the second number.

  2. All units, including those that are named for a person, have a lower-case first letter when written out (not abbreviated). Thus, write "ohm, farad, coulomb, volt, ampere, hertz" for units.

  3. abbreviations for units
    Units that are named for a person have an upper-case first letter when abbreviated; all other units have a lower-case first letter.

    Only metric prefixes for 10+6 or more have an upper-case abbreviation (e.g., M = 10+6, G = 10+9, etc.). In particular, note that the prefix m indicates 10-3 and M indicates 10+6. The difference between an upper-case M and a lower-case m is nine orders of magnitude! One should be warned that American manufacturers of capacitors often use "mF" or "MF" to indicate microfarads, a practice that is both incorrect and misleading.

    The proper abbreviation for "kilohertz" is "kHz": only the "H" is upper case.

    Note that the proper abbreviation for "second" is "s", not "sec".

    The same abbreviation is used for the singular and plural form of a unit.

    A period is not placed after an abbreviated unit, unless it is at the end of the sentence.

  4. spaces with units
    There should always be one blank space between a number and a unit:
    "5 kHz", not "5kHz"
    In modern wordprocessors, the space between number and unit should be a nonbreaking space, so the number will always appear on the same line as its unit.

    There is no space between the metric prefix and the base unit.

  5. Watch out for expressions such as:
    The signal generator had a 15 kHz frequency.
    Units of measurement ("kHz") are nouns and can not be used to modify another noun ("frequency"). The proper phrasing is:
    The signal generator had a frequency of 15 kHz.

    This type of problem often occurs when length or thickness is specified, as in:
    6 cm long
    The proper phrasing could be:
    6 cm in length
    the length is 6 cm
    having a length of 6 cm

  6. Avoid labeling the axis on a graph or a column in a table as, for example,
    volts x 10-3
    This is ambiguous: are the numbers to be multiplied by the reader, or has the multiplication already been done?
    Maybe the number 3 on the axis of the graph or in a column of text means 3000 V, which times 10-3 is just 3. In such a case, the author should have written 3 kV.
    Maybe the number 3 on the axis of the graph or in a column of text means 0.003 V, which is 3 x 10-3. In such a case, the author should have written 3 mV.

  7. In general, choose a metric prefix that will make the numerical value between 0.1 and 1000. However, the value of a parameter or a variable over the range of a few paragraphs or in a table should have the same metric prefix to allow easy comparison of different values.

  8. Do not use metric prefixes when using scientific notation: e.g.,
    "4 x 105 m/s", not "4 x 102 km/s".

  9. Never use a double metric prefix. For example, in older American literature, one will find small capacitances expressed in "µµF" (or, worse, "mmF"), for "micro-micro-farad", instead of the modern preferred unit, "pF" for picofarad.

  10. When writing the full name of a metric unit, in which a metric prefix that ends in a vowel is attached to a base unit that begins in a vowel, retain both vowels, except:
    instead of "megaohm" use "megohm"
    instead of "kiloohm" use "kilohm"
    instead of "hectaare" use "hectare"

  11. The proper abbreviations for "alternating current, direct current, and root-mean-square" are lower case "ac, dc, rms". However, if the term appears in a title or as the first letter in a sentence, then the entire abbreviation is upper case: "AC, DC, RMS".

  12. Some scientists, physicians, and engineers use the word "micron" to denote what is properly called a "micrometer", which is abbreviated µm. Avoid using the word "micron", as it is technical slang.

units of computer memory

Computer memory, both semiconductor memory and hard disk memory, is arranged in integer powers of 2. A binary digit, which as a value of either 0 or 1, is called a bit. A group of eight bits is called a byte. A hard disk drive that can store 1.0737 × 1010 bytes is commonly (but erroneously) specified to have a capacity of 10 Gbyte. The metric prefix Giga indicates 109, so one would expect that a 10 Gbyte drive could hold exactly 1010 bytes. The difficulty arises because the computer industry commonly refers to a factor of 210 or 1024 as "k", 220 or (1024)2 as "M", and 230 or (1024)3 as "G". This practice of the computer industry is acceptable only if one does not call their k, M, and G "kilo-, mega-, and giga-". In December 1998, the International Electrotechnical Commission decided to call the computer industry's k, M, and G, "kibi-, mebi-, and gibi-".

I note in passing that some manufacturers of hard drives specified capacity properly in megabytes (i.e., 106 bytes), probably because that practice would give them a 4.8% larger number — thus making their disk drives appear larger than they actually were. Hence, the ambiguity in the M could be exploited to confuse consumers.

equations in text

Technical writing often contains equations, however the use of equations is not commonly discussed in books on style and composition. Mermin (Physics Today, pp. 9-11, Oct 1989) pleads that equations are not a "grammatically irrelevant blob". There are several simple rules that gracefully include equations in text.
  1. Equations are prose and should be punctuated as such. It is common for an equation to function as a clause or sentence: variables and parameters are nouns; the equal sign is equivalent to the verb "is". Operators serve as conjunctions.

  2. Do not embed equations in a line of text: every equation goes on its own line.

  3. Number each equation at the right-hand margin (use the "flush right" command in WordPerfect). Even if it is not necessary to refer to the equation by number in the text, someone may want to refer to the equation in a letter or future publication.

  4. One must identify each of the variables and parameters by name when they first appear. (One might be excused for not saying that t is the time, since this is such a standard choice of variable.)

  5. If an equation is so long that it requires more than one line, then consider introducing new variables for terms in the equation. If each term has some physical significance, it will also make it easier for the reader to understand the equation. For example:
    D(t) = A [1 - exp(-t/0.001)] exp(-t/0.10) cos(106 t)
    V(t) = D(t) + 170 sin(377 t + 0.03)             for t > 0

    D(t) is a damped oscillation that is superposed on a sinusoidal oscillation to give the total voltage, V(t).
    One normally uses parameters, not numbers, in equations. I used numbers in the above example, only because HTML 3.2 does not include Greek letters.

  6. There are several ways to use equations with text. The brute force way is simply to refer to the equation number in a sentence; the equation is a separate sentence. For example:
    The voltage, V, and current, I, in a resistor are related by Eq. 3.
                        V = I R                                             (3)

    Often it is more elegant to include the equation in the sentence, for example:
    The power, P, dissipated in any two-terminal device is given by
    P = I V                   (4)
    where I is the current in the device and V is the voltage across the device. For the special case of a resistor, the relation between voltage and current, Eq. 3, can be used to express the power as a function of only voltage:
    P = V2/R.               (5)

    Notice the period at the end of Eq. 5, since it is the end of a sentence.

common misuse of words in English

•   "Affect" is usually a verb, which means "to influence" or "to change", as in:
Will bad grammar affect my grade?
"Effect" is usually a noun, which means "result", as in:
cause and effect relationship

•   The word data is plural. Therefore, one must say, for example:
These data are....
If the singular is required, use "datum", or say "this measurement".

It is important to distinguish data, which are always measured, from results of computer simulations or calculations. One is not necessarily more trustworthy than another, although many scientists and engineers seem to have more confidence in measurements than in computer calculations. The computer work may be described as a "model", "calculation", or a "simulation". Do not refer to results of a computer simulation as either "observation(s)" or data.

•   Avoid using nouns as verbs. Francis Straus (Science, Vol. 177, p. 1154, 29 Sept 1972), tongue-in-cheek, calls this "verbing virtuous nouns". Examples:
avoid saying "He authored ...", instead say "He wrote ...".
avoid using "obsoleted", instead say "made obsolete".

•   A meticulous writer distinguishes between that and which. An expert in English grammar would say that properly introduces restrictive clauses, while which introduces nonrestrictive clauses, but if you understand this sentence, you don't need a grammar lesson. <grin> A more helpful statement of the rule is:
  1. use which, which is always preceded by a comma, when introducing nonessential information.
  2. use that, never preceded by a comma, when introducing essential information.
For example:
Resistors, which often have pretty color bands, are common elements in electronic circuits.
In this example, "pretty color bands" are irrelevant to the functioning of the circuit or whether resistors are commonly used.
A resistor that has smoke rising from it is in danger of failing.
Here, "that has smoke rising from it" identifies specific resistor(s) that are essential to the meaning of the sentence.

Sometimes long restrictive clauses get in the way of quickly understanding a sentence. Consider the following example, a letter to the editor that appeared in The Economist, a respected British news magazine (20 Mar 1999, page 8):
John Coale says he would like to introduce those who insist that citizen gun-ownership is a guarantee against overmighty government to the 101st Airborne Division.
The problem with this sentence is that a long restrictive clause ("who insist that citizen gun-ownership is a guarantee against overmighty government") between the verb introduce and the object "to the 101st Airborne Division." The sentence is easier to understand if the object is moved earlier:
... he would like to introduce to the 101st Airborne Division those who insist that citizen gun-ownership is a guarantee against overmighty government.

•   The word "resonant" can not be used to describe frequency. Instead, say "resonance frequency". The circuit or system, but not the frequency, is resonant. (See F. V. Hunt, J. Acous. Soc. Am., 50:435, 1971)

•   The word "factor", when properly used, usually describes an algebraic term or number that is multiplied by other terms or numbers. The word "factor" is often used in other contexts that are properly considered as engineering slang.

•   Avoid using "differentiate" as a synonym for "distinguish". The word "differentiate" has a special meaning in mathematical analysis.

•   To insure means that money is paid by the insurance company if the event that is insured against occurs. To assure is to convince or promise someone. To ensure is to do something that will cause a particular result or outcome. Avoid confusing them. In U.S. industry, product inspectors are part of a "quality assurance" team. This is a misuse of assurance, the word ensurance should have been used. (But I have known some literate engineers who claimed that "quality assurance" was a sarcastic comment about the worth of the inspectors' efforts. <grin>) These distinctions only apply to American English.

•   Avoid constructions where "current flows". Many scientists insist, with considerable justification, that charge flows, not current. However, in speech everyone says "current flows", so this issue sounds pedantic and stilted. However, everyone will be happy with minor revisions that also omit needless words. For example, consider
the current that flowed in R1
could be changed to
the current in R1

•   Many engineers like to say that they have found the "optimum" value(s). The word "optimum" properly means "best possible" or "most favorable". Optimum values can be found by the calculus of variations and other sophisticated techniques. To use the word "optimum" to describe lesser efforts is an inflated use of the word, rather like calling a garbageman a "sanitation engineer".

•   Avoid an isolated use of the word this, as in:
This is discussed later.
Instead, add a word to specifically indicate what this refers to:
this feature
this circuit
this problem
this approach

•   Avoid colloquial expressions, such as the following list:
  1. Instead of hook up, used as a verb, say: connect.

  2. Instead of set up, used as a noun, say either: apparatus, system, circuit.

  3. Instead of burn out, say either: damage, destroy.

  4. Instead of write up, used as a noun, say either: report, essay, document.

•   In precise writing, one should distinguish between academic degrees and occupations. For example, it is common to use doctor as a synonym for physician. This carelessness probably arose because physicians are the only people with an earned doctoral degree who are personally known to most people. However, senior people in many occupations (e.g., science, medicine, psychology, history, engineering, religion, etc.) have earned a doctoral degree.

•   Journalists often write with hyperbole, in an attempt to make the news appear more significant than it really is, or in an attempt to make their writing more lively and more interesting. For example, one often reads that the value of stock in some company, or the value of some index of the stock market, "plummeted", when the actual decrease was only 2%. If one plotted the value on a scale from zero to its all-time high, a change of 2% would be barely discernable. If one routinely engages in hyperbole, then there is little force left in one's words for a truly remarkable or big event.

•   It is bad form to use "secondly" and "thirdly" to introduce multiple items in series. It is better to follow parallel structure and use "first", "second", and "third". Alternatively, one could use (1), (2), and (3), perhaps in an indented list, as in an outline.

a or an ?

The rule is simple:
Use a if the first letter of the following word begins with a consonant when pronounced.
Use an if the first letter of the following word begins with a vowel sound (a, e, i, o, u) when pronounced.

If the following word is either an abbreviation or a number, then consider whether it begins with a vowel sound when pronounced:
a UPS   (UPS = "you pea ess")
an MOV   (MOV = "em oh vee")
an RS-232   (RS = "are es")
a 1N4004   (1N = "won in")
an 1800 V   (1800= "ate teen hundred")
Sometimes the first letter in a word is silent, so one then considers the second letter in the word following the article:
an honor
an hour
an heir   (heir = "air")

the conjunction or

Most legal style manuals devote at least one page of diatribe to the meaning of and, or, and/or. I can only conclude that the intensity of these attorneys' argumentation must compensate for their ignorance!

The truth of the matter is that there are two distinctly different meanings of the word or in the English language:
  1. the inclusive or, which is satisfied if at least one of the items in the or'ed list is true. The inclusive or can be stated plainly as:
    "any one or more of the following: A, B, C, or D"
    Where there are only two alternatives, one can phrase the inclusive or as:
    "either A or B, or both",   "A and/or B"

  2. the exclusive or, which is satisfied if only one of the items in the or'ed list is true. The exclusive or can be stated plainly as
    "one, and only one, of the following: A, B, C, or D"
    Where there are only two alternatives, one can phrase the exclusive or as:
    "either A or B, but not both"

Regardless of the ambiguity in the English language, there is no ambiguity in the use of the word or as a mathematical term in symbolic logic, where or is always the inclusive or. Symbolic logic is the basis for digital computers and computer programming, which have revolutionized our way of working. For example, whenever an attorney does a search of a computer data base (e.g, search engine on the Internet, Dialog, Westlaw or Lexis), the OR command is always the inclusive or.

It is my impression that most physicists and mathematicians generally use or in the inclusive sense, and most attorneys in the USA generally use or in the exclusive sense.

The people who write legal style manuals don't seem to notice that their assertion about or not only contradicts the mathematical definition that is used in computer data bases, but also is a narrow-minded approach to the English language. For example, if one sees a sign on the library door that says:
drinking or eating prohibited
does this sign mean that it is acceptable to simultaneously eat and drink in the library? Of course not!

The situation is different in Latin, where "vel" is the inclusive or and "aut" is the exclusive or. The or operation in symbolic logic is often a "v", from the corresponding Latin word, although electrical engineers generally use a + sign for the or operation.

another logical issue

A U.S. Supreme Court opinion, Posadas De Peuerto Rico Associates v. Tourism Co. 478 U.S. 328 (1986), contains an example of faulty writing: "... is not misleading or fraudulent."
Does this mean
(not misleading) or fraudulent
not (misleading or fraudulent)   ?
As a matter of grammar, the Court's statement is, at best, ambiguous. Of course, only the second interpretation makes any sense. The Court should have said:
neither misleading nor fraudulent

An interesting additional problem in the use of conjunctions "and, or" in writing, which nearly all legal style manuals neglect, is the use of a negative with these conjunctions. If we adopt the nomenclature of v for inclusive or, & for and, and - for not, we can write two important theorems in symbolic logic as:
- (A v B) = (-A) & (-B)           (1)
- (A & B) = (-A) v (-B)           (2)
In words, Eq. 1 is
not (A or B) = neither A nor B
Nor (i.e., "not or") does not appear to have the ambiguity associated with inclusive or and exclusive or in English.

Consider the following practical example:
Suppose we want to write a sign for the library door to prohibit any combination of drinking or eating.
1.   no drinking or no eating
2.   drinking and eating prohibited
Sentences (1) and (2) are equivalent if the or is the inclusive or and if we understand "prohibited" in (2) to mean "no (drinking and eating)". These sentences are not a correct expression of the desired meaning, as they allow a person to either (drink and not eat) or (eat and not drink).

3.   no drinking and no eating
4.   drinking or eating prohibited
Sentences (3) and (4) are equivalent if the or is the inclusive or and if we understand "prohibited" in (4) to mean "no (drinking or eating)". Sentences (3) and (4) are logically correct. However, some attorneys will read sentence (4) and tell you that it is ok to both drink and eat in the library, following their erroneous interpretation that or is exclusive. Given the ambiguity of or in English, sentence (4) is vague.

5.   You may not eat or drink in this library.
Sentence (5) is logically correct if the not is distributed, to mean "not (eat or drink)". This sentence suffers from ambiguity about the scope of not, in addition to ambiguity about whether or is the exclusive or.

6.   Neither eating nor drinking is permitted.
Sentence (6) is my preferred way of expressing the thought. It is both logically correct and difficult to misunderstand.

use of pronouns

In informal English the pronoun "you" is often used to indicate an indefinite person, not necessarily the reader or listener. This can be mildly offensive, for example:
When you make a mistake....
It is preferable to use third person:
When one makes a mistake...
When an engineer makes a mistake....
Other synonyms for this usage of "you" include "a person", "people", etc.

The usage of "you" often results in childish constructions that indicate personal involvement of the reader when it is unnecessary or inappropriate:
When you apply a greater electric field....
should be
When the electric field is more intense....

Avoid sexist language. Because most scientists and engineers are male, one sometimes inadvertently uses a masculine pronoun to refer to a general technical person, for example:
An engineer should recognize that he represents his profession to the public.
A simple fix is to use the plural form, which is the same for masculine and feminine:
Engineers represent their profession to the public.
We represent our profession to the public.


There are several schools of thought about the use of hyphens. One can strictly follow elaborate rules for usage of hyphens, which tends to produce more hyphens than many good writers will accept. One can use hyphens only when they are necessary to avoid ambiguity, which tends to produce fewer hyphens than most copy editors will accept. My favorite reference for guidance on use of hyphens is The Chicago Manual of Style, although this manual recommends more hyphens than some technical writers will accept. The American Institute of Physics Style Manual says to "avoid the hyphen when it does not serve a useful purpose."

  1. most compound adjectives are hyphenated when they immediately precede the noun that they modify:
    common-mode voltage
    differential-mode current
    steady-state voltage or current
    well-known fact
    high-frequency behavior
    low-pass filter
    odd-integer terms

    Note that the same words used as a noun are not hyphenated:
    open-circuit voltage
    short-circuit current
    "The configuration was an open circuit."   (no hyphen here)

    first-aid kit,
    "You render first aid."   (no hyphen here)

    worst-case event,
    "Consider the worst case."   (no hyphen here)

    high-voltage supply
    low-voltage system
    But write
    There was high voltage on the wires that were supposed to carry low voltage.

  2. compound words containing numbers or "half-" are hyphenated:
    three-electrode tube
    four-terminal capacitor
    first-order approximation
    two-thirds   (Most technical writers would say "2/3", not "two-thirds".)
    8-bit digitizer

  3. Adverbs ending in "-ly" do not form hyphenated compounds.
    highly competent person
    fully ionized gas

  4. compound words formed by two words with equal functions, so that the compound is a single word with a different meaning:
    gain-bandwidth product
    analog-to-digital conversion
    one-to-one correspondence

  5. all compounds formed with "-free" are hyphenated:

  6. One generally does not hyphenate words formed by prefixes, except:
    1. all compounds formed with "self-, all-, ex-, quasi-" are hyphenated:

    2. to distinguish from homonyms:
      re-cover, re-solved (hyphenate when meaning "cover again, solve again")
      un-ionized (hyphenate when meaning "not ionized")

    3. when last letter of prefix is identical to first letter of following word:

    4. when there is otherwise a repeating sequence of letters that is confusing or ugly:
      For example, "nonoscillatory" looks like "no no scillatory" when read quickly.

    5. when the prefix or suffix is added to a name, symbol, or number:
      non-Newtonian physics
      Cd-free solder
      pre-1970 designs

      Note that this rule says that the following have no hyphen:
      infrared LED

citations and bibliography

In addition to the normal features of good writing (e.g., correct grammar, good style, punctuation, spelling, etc.), scholarly writing imposes an additional obligation on the author: citing relevant published literature. To educated people who engage in critical thinking, an unsupported assertion is worthless, no matter how authoritative the author or speaker is. References provide support for statements and add credibility to writing.

The rules for what needs a citation are an academic tradition, but are rarely stated explicitly:
  1. All direct quotations from another author must be cited. The writer has no discretion in this matter: the rule is absolute. Failure to cite quotations is known as plagiarism, a serious academic offense that is equivalent to fraudulent representation of someone else's property as the writer's. It is highly recommended that authors always include the indicia of a quotation [i.e., both (1) a citation to the original source and (2) enclose all quoted material within quotation marks or set it in a block of indented, single-spaced text] immediately after typing or inserting the quoted text into the author's document. It is not an acceptable defense to plagiarism to claim that the author forgot to include the indicia of a quotation.

  2. All substantial information taken from another source should be cited. There are four reasons for this:
    1. to give credit to the person who supplied the information or first made the discovery (i.e., to avoid misrepresenting an idea as the original work of the author),

    2. to relieve the writer from the responsibility for the accuracy or truth of the information,

    3. to lead the reader to a source of more detailed or complete information, or

    4. to give the reader a sense of the historical evolution of ideas in the field.

The meaning of "substantial information" in (2) is deduced on a case-by-case basis by considering the four reasons. If at least one of the reasons is appropriate or desirable, then a reference should be used.

However, one does not give a reference for well-known facts (e.g., Newton's Three Laws of Motion, Maxwell's Equations for Electromagnetism), except in history papers. The appropriate test is whether any person with an undergraduate education in the appropriate specialty would immediately recognize the fact: if yes, then no reference is needed.

Let us take a moment to expand on item 2b above. If the writer doubts the truth of the information, then the writer should indicate to the reader the basis for those doubts. This can be accomplished diplomatically by making a "on the one hand...on the other hand" argument by citing two different authorities who disagree. Doubts can also be raised and at least partly resolved by discussing alternative interpretations. The point is that the writer can not just serve the reader some cited material and then walk away from the mess: the writer has a responsibility to guide the reader to an informed decision in a coherent way. If a complete resolution of the facts is not possible, then the writer has an obligation to say so.

Some types of statements beg for a citation. For example,
It is commonly believed that....
It is widely known that....
The conventional wisdom is that....
These assertions need a citation of at least one (I prefer three) references that support the assertion. These references may be to textbooks, which are rarely cited in professional literature in other contexts. There are two reasons for requiring a reference to this type of statement: (1) the reader may disagree with the assertion and need to be convinced that it is a commonly held belief, or (2) the reader may not be an expert in the area and may need a source of more information in order to understand what is being discussed.

What kind of literature should be referenced?

It is preferred that all references be archival material: books, scholarly journals, and certain other publications. There is one test for "archival": Is it retained permanently by many major technical libraries? In general, any paper that is listed in standard databases (e.g., INSPEC's Physics and Electrical Engineering Abstracts) certainly qualifies as archival. In addition, patents and government reports qualify as archival materials, although they are often not considered scholarly materials.

Engineering standards, although they are important, are not archival: most university libraries have few standards, and libraries that do have standards typically only keep the current edition of the standard. It is almost impossible to locate a copy of an obsolete or withdrawn standard, unless one knows an old engineer who has a copy in the filing cabinet! For these reasons, standards are not archival documents. However, if one needs to cite to conventional good engineering practice or to cite to a performance specification, then one can cite engineering standards.

In general, one should avoid citing proprietary literature (manufacturer's application notes, specification sheets for products, etc.) or trade magazines. These materials are of an ephemeral nature and definitely not archival: most libraries have few of these publications and what they do have is nearly always discarded when a new edition becomes available. A trade magazine is a periodical that is characterized by an abundance of (1) advertisements and (2) articles written by a employee of manufacturer X about a specific product of manufacturer X. Such material is commonly not objective, and is often blatantly self-serving. Trade magazines are often distributed free to readers who are engaged in purchasing and specifying the products that are described in the magazine. Trade magazines are distinguished from a journal that is published by a professional society. Journals usually have two or three recognized experts review each prospective article for accuracy (a process called "peer review") before the article is published, while trade magazines publish anything that the editor wishes. Journals are supported by fees paid by the readers, while trade magazines are supported by advertising revenue. This suggests that journals are responsible to their professional readers, while magazines are responsible to their advertisers.

The Internet is not an archival source. One of the nice things about the Internet is that authors may make frequent revisions and additions to their works, which is one source of instability. Another source of instability is that people remove their web site from a local Internet Service Provider or university when they move. Webmasters sometimes rearrange directories or rename files, so only the domain name in the URL remains stable. In addition to these problems of instability, information on the Internet is only as trustworthy as its source: unlike books and scholarly journals, there is no peer review of material on the Internet.

format of citations and references

There are many different styles of references. The "proper" style is determined by each technical journal and is specified in its style manual. A writer should obtain a copy of the appropriate style manual before beginning writing. In the absence of specific advice in a style manual, here is the style that I prefer. To distinguish from conventional style, I call it preferred style.

Put a citation in the text that contains a citation to the author's name, date of publication, and perhaps the specific page number of work cited. For example:
A general survey of overvoltage protection techniques has been published (Standler, 1989).
Standler (1989) wrote a general survey of overvoltage protection techniques.
One should avoid using a neon indicator lamp as a surge-protective device (Standler, 1989, p. 130).
Then at the end of the document, one lists a bibliography in alphabetical order by the name of the author and — when more than one item by an author is cited — chronologically by time of publication for that one author:
Standler, R.B. Protection of Electronic Circuits from Overvoltages, Wiley-Interscience, 434 pp., 1989.

If an author publishes two or more items in one year (for example: 1989) that are cited, then, the earliest one is identified as 1989a, the second as 1989b, the third is 1989c, .... The same letters are used after the year of publication in the bibliography.

The reference list at the end of the scholarly article has the following format:
•   for a book written by John Doe:
Doe, J., Title of Book, publisher, total number of pages, year of publication.
If the book is a second — or later — edition, then the edition number (e.g., "2nd ed." or "3rd ed.") must also appear in the reference list.

•   for an article in a scholarly journal by John Doe:
Doe, J., "Title of Paper", Title of Journal, volume number: first page-last page, month and year of publication.

•   for a document by John Doe that is posted on the Internet:
Doe, J., Title of Document, URL, date of last revision.
The URL should include "http://", "ftp://", or "telnet://".
The MLA style requires the URL to be inside angular brackets (e.g., <>) which seems unnecessary to me, as the prefix "http://" clearly identifies the URL. Further, angular brackets are used to enclose commands in HTML, which could cause problems if the reference list is written in HTML and either posted on the Internet or sent via e-mail. If one wants to distinguish the URL from text, the URL could be set in a monospaced font.

The reference list has similar formats for other items. The author's name, title of item, and year of publication should always be given. For difficult-to-find items, other information that is useful in locating the item in a library, or in purchasing a copy, should also be included.

Titles are set in either italics or underlined. The modern form is to use italics. When I was a student in 1960s and 1970s, underlining was common, because one could underline (but not italicize) on a typewriter.

In the case of both books and journal articles, a specific page can be cited in the text, for example: (Doe, 1978, p. 132), but not in the reference list at the end of the scholarly article.

justification for this style

The preferred style of references suggested above is used in the following scholarly journals, among others:
Journal of Geophysical Research,
Quarterly Journal of the Royal Meteorological Society,
all journals that are published by the Institute of Physics in Great Britain (e.g., Journal of Physics),
American Scientist (journal of Sigma Xi),
Radio Science, and
The Journal of Comparative Neurology.
The preferred style is also recommended by the Modern Language Association (MLA). At my urging, the IEEE Editorial Office accepted this style of references in the revision of ANSI/IEEE Standard C62.41-1991. I also persuaded Wiley-Interscience Press to allow me to use this style in my book that was published in 1989.

This preferred style has several major advantages over the conventional use of reference numbers (e.g., a citation to [6]):
  1. The conventional style demands that the reader turn to the end of the paper to decode the reference numbers, so the reader needs to be looking at two pages simultaneously. Moreover, a reader who is familiar with the literature can recognize some of the classic papers by author and date of publication, whereas a citation to item 6 in the bibliography is meaningless until the reader looks at the bibliography.

  2. This style makes it easier for the author to prepare the paper because the citation is independent of its location in the paper. If paragraphs are rearranged or material is inserted in a draft, conventional reference numbers may need to be revised from the location of the change to the end of the paper, but references in this style do not need to be changed. I favor making it easy to rearrange, insert, or otherwise modify a draft without having to simultaneously revise the order of references.

  3. This style makes it easy for the author of the article, who is familiar with the references to the literature, to verify that the citations are accurate.

An editor may not care about making life easier for authors, but the real benefit of making life easier for authors is that there will be fewer errors in citations. Errors make citations worthless. Furthermore, it is commonly known that extensive revisions produce a better final product. We should encourage writers to revise their writing, and not punish them by making them also revise their reference numbers with each revision.

citing sources you have not read

When an author cites a source in a footnote or reference in the text, that citation is conventionally understood to indicate that the author has personally seen and read that source. The cliché "Never cite sight unseen" comes to mind.

Sometimes a book or scholarly paper mentions a fact or opinion, with a citation to some obscure source that is not available locally or is in an obscure foreign language. How does one cite this fact or opinion?

For example, suppose that you read a book by Richard Smith, A General Theory of Stuff, and on page 328, Smith mentions a fact and cites the source: Theodore Grobolink, "Stuff is Good," Transactions of the Mississippi Academy of Stuffology, Vol. 48, p. 547, published in the year 1878. For some reason, you can not obtain Grobolink's paper.

The citation for this fact would look something like the following:
Theodore Grobolink, "Stuff is Good," Transactions of the Mississippi Academy of Stuffology, Vol. 48, p. 547, (1878). Cited in Richard Smith, A General Theory of Stuff, at p. 328, Oxford University Press (1997).

However, if you think it is more significant that Smith, the preeminent modern scholar in the field of stuff, believes the discovery of Grobolink, then the citation would look something like the following:
Richard Smith, A General Theory of Stuff, at p. 328, Oxford University Press, 1997. Citing Theodore Grobolink, "Stuff is Good," Transactions of the Mississippi Academy of Stuffology, Vol. 48, p. 547, 1878.

In the above example, I say "would look something like" because there are oodles of different formats for footnotes and bibliographic entries. The word "Cited" could also be "Quoted" or another appropriate word. If you use the format that I suggest above you could write in the text:
Grobolink (1878) observed that ....
and write in the bibliography:
Grobolink, Theodore, "Stuff is Good," Transactions of the Mississippi Academy of Stuffology, Vol. 48, p. 547, 1878. Cited in Richard Smith, A General Theory of Stuff, at p. 328, Oxford University Press, 1997.
or if Smith's citation is more significant to you than Grobolink's work, then you might write in the text:
Smith's (1997, p. 328) authorative review of the subject cites Grobolink (1878) as the first to observe that ....
Smith (1997, p. 328) mentions that Grobolink (1878) observed ....

verb tense and voice

I prefer to use first person, active voice in writing, because such writing is both easier to write and to read. This is contrary to the advice found in some other sources, that would have one write "the voltage was measured by the author" rather than "I measured the voltage." If a reference is needed to convince someone else to let you write in first person, active voice, try the American Institute of Physics Style Manual, p. 14, third edition, 1978.

Many engineers believe that all formal technical writing should use the past tense. However, it is appropriate to use present tense for things that are true when the author writes about them and will still be true in the future when the text is read.

illogical rules in English

In the following items, I have defiantly adopted a logical practice that varies from conventional style in American English. You can decide for yourself (1) whether you want to join with me in being logical, and then receive scathing criticism from pinhead managers who exalt style above content and who refuse to recognize the possibility of acceptable alternatives in style, or (2) whether you want to take the easy route. <grin>

I write dates in the format of day-month-year, for example, 8 Mar 1999. In the USA, this is commonly known as the "military form" of dates. Actually, the military adopted this form to harmonize with our European allies in NATO. One of the leading organizations of scientists in the USA, the American Association for the Advancement of Science, uses the day-month-year format in its journal, Science. The journals of the American Institute of Physics also use the day-month-year format. Aside from harmonizing with international practice, there is a consistent order from small units (days) to large units (years), unlike the conventional American practice. Note that a date written 8-3-1999 is ambiguous to an international audience: is it the eighth day of the third month (German form) or the third day of the eighth month (American form)? Therefore, I recommend using a three-letter abbreviation for the month, instead of a number, to avoid this ambiguity.

The accepted rule in American English when using quotation marks is always to put a final comma or period inside the closing quotation mark. For example:
Arnold said "No," but I did not believe him.
Logically, this makes little sense. The comma is not part of Arnold's statement; the comma indicates a pause before the "but" in my sentence. Historically, this silly rule comes from antique typesetting practices, where small elements like a comma or period, were vulnerable to mechanical damage and were protected by including them inside the ending quotation mark. Large punctuation marks (e.g., colon and semicolon), are conventionally placed outside quotation marks. Question marks and exclamation points are conventionally placed logically, i.e., whether they are part of the quoted material or not. I like the statement in the 1979 edition of Mathematics Into Type, a style manual from the American Mathematical Society, which says "In general, mathematicians are probably hastening the process toward placing quotation marks logically." This example is just one of many that I could cite of a rule of style that was reasonable in the distant past, which rule is now continued, even though the reason has vanished. Rules should not be arbitrary, they should have a clear justification! The conventional rule for punctuation can mislead the reader, as in the following example:
The URL of my web site is ""
The final period indicates the end of the sentence and is not part of my URL, yet the conventional rule for punctuation invites the reader to type the final period as part of my URL. I would write this sentence as
The URL of my web site is "".
A better way to write this sentence is to avoid quotation marks and adopt the style used in software manuals. Set commands, URLs, and other material to be entered by the reader in a monospaced typeface and write the sentence to avoid punctuation at the end of the URL:
To go to my web site, type
in the URL window of your browser.

There are oodles of other examples of illogical or inconsistent usage in American English. For example:
  1. may not is properly written as two words, while conventional practice is to write cannot as one word, although these two examples are grammatically identical.

  2. The conventional English abbreviation for number is "No." There is no o in the word number, the abbreviation actually comes from the Latin word for number. I use the German abbreviation for number, "Nr.", when writing English text.

this document is at
first posted 24 Sep 1999,   last revised 4 Mar 2014

return to my homepage