How to Write and Illustrate: Scientific Papers, second edition

Document technical information

Format pdf
Size 6.7 MB
First found Jun 9, 2017

Document content analysis

not defined
no text concepts found


Steve Crocker
Steve Crocker

wikipedia, lookup

Port Rashid
Port Rashid

wikipedia, lookup




This page intentionally left blank
This Second Edition of How to Write and Illustrate a Scientific Paper will help both
first-time writers and more experienced authors, in all biological and medical
disciplines, to present their results effectively. Whilst retaining the easy-to-read
and well-structured approach of the previous edition, it has been broadened to
include comprehensive advice on writing compilation theses for doctoral degrees,
and a detailed description of preparing case reports. Illustrations, particularly
graphs, are discussed in detail, with poor examples redrawn for comparison.
The reader is offered advice on how to present the paper, where and how to
submit the manuscript, and finally, how to correct the proofs. Examples of both
good and bad writing, selected from actual journal articles, illustrate the author’s
advice – which has been developed through his extensive teaching experience – in
this accessible and informative guide.
B J Ö R N G U S T A V I I has been teaching courses in scientific writing for doctoral
(Ph.D.) students in medicine for 25 years. He brings his personal experience to this
book, both from writing more than 100 of his own research papers and from his
work as a journal editor.
How to
Write and
Scientific Papers
Björn Gustavii
Second Edition
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo
Cambridge University Press
The Edinburgh Building, Cambridge CB2 8RU, UK
Published in the United States of America by Cambridge University Press, New York
Information on this title:
© B. Gustavii 2003, 2008
This publication is in copyright. Subject to statutory exception and to the provision of
relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.
First published in print format 2008
ISBN-13 978-0-511-39316-7
eBook (EBL)
ISBN-13 978-0-521-87890-6
Cambridge University Press has no responsibility for the persistence or accuracy of urls
for external or third-party internet websites referred to in this publication, and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.
Preface v
1 Basic rules of writing
2 Comments on scientific language 3
3 Drafting the manuscript 15
4 Choosing a journal 18
5 Preparing a graph 20
6 Drawings
7 Figure legends 40
8 How to design tables 42
9 Title
10 Authors
11 Abstract
12 Introduction
13 Methods
14 Results 68
15 Discussion 74
16 Acknowledgments 79
17 References
18 Ph.D. and other doctoral theses 91
19 Letters and case reports 101
20 Numbers 105
21 Abbreviations
22 How to present statistical results 114
23 Typing
24 Dealing with editors and referees 132
25 Correcting proofs
26 Authors’ responsibilities
Literature needed on your desk 150
Further reading
Literature cited
Dear Novice Writer,
When I was in your shoes and preparing my first paper, I consulted
a book on how to write. I found there a sentence encouraging
the reader to stand in boiling water for an hour before doing the
After standing in boiling water for an hour, examine the
contents of the flask.
I had a pretty good idea of what was wrong with the sentence but,
at the time, I couldn’t figure out how to revise it, and the author
didn’t tell me. Now I can. If, an hour later, you are still alive:
Place the flask in boiling water for an hour, then examine its
So, in this book, every unfortunate example is followed by an
improved version. Good examples are provided with appropriate
bibliographic references. Bad ones, however, are presented with
references expunged.
Some examples were taken from manuscripts in preparation,
presented by participants in my courses on scientific writing. I have
been holding such courses for doctoral (Ph.D.) students in medicine
since 1980. Other specimens are from manuscripts submitted for
publication. They were collected when I served as an editor of Acta
Obstetricia et Gynecologica Scandinavica from 1986 to 1994. Yet
others are from published material.
From class discussions I have learned what candidates want to
know. Based on this information, some chapters are more detailed
than others, such as the one on how to prepare graphs.
The current edition contains a new and comprehensive chapter
on doctoral (Ph.D.) theses. Numerous other changes also appear in
this edition, for example, instructions for making drawings and a
description of preparing case reports.
Finally, don’t accept all my suggestions, because there is no
ultimate truth regarding how to write a paper – as I mistakenly
believed when I was a bit younger.
Good luck, my friend.
Björn Gustavii
I thank the following persons, who have read all or parts of the
manuscript of the second edition, for their advice and criticism.
Per Bergsjø, Norway
Carol Brimley-Norris, Finland
Joy Burrough-Boenisch, UK
Johan Ljungqvist, Sweden
Helen Sheppard, Sweden
Ray Williams, UK
Pål Wölner-Hanssen, Sweden
Special thanks to Tomas Söderblom, a layperson, who read the
manuscript for intelligibility; Richard Fisher, who corrected the
language; and Eva Dagnegård, who redrew the graphs and prepared
the electronic manuscript.
Basic rules of writing
Winston Churchill was sitting at his desk, working on his epic
about World War II, when his private secretary entered the room.
Churchill had reached the Blitz – the German air strikes against
London. His staff of researchers had earlier produced a 150-page
brief on the raids. The secretary had been asked to cut it down to
about two and a half pages and, after having “worked like stink,”
he could now proudly hand over the condensed version.
Churchill took out his red pen and started to edit. “All my
sloppy sentences were tightened up and all my useless adjectives
obliterated,” the secretary tells us in a documentary made about
50 years later (Bennet 1992). In the midst of it all, Churchill said
gently, “I hope you don’t mind me doing this?” The secretary
answered, “Thank you, Sir – you are giving me a free lesson in
writing plain English.” Churchill, Winston
We should emulate Churchill by excluding every nonessential
word. Professional writers do it that way. Brevity is an elementary
rule of all writing, not only to save valuable publication space, but
Basic rules of writing
also because verbose writing obscures meaning and wastes the
reader’s time and patience. And that is also the essence of the next
basic rule.
Logic and clarity
To convey information is above all a matter of logic and clarity.
What you want to say should be so arranged that the reader can
follow your argumentation step by step. Moreover, your sentences
should be so clear and easily understood “that the reader forgets
that he is reading and knows only that he is absorbing ideas” (Baker
Now to the importance of making the manuscript physically
attractive. Here is an illustrative example.
Clean typing
Paul Fogelberg, editor of a Finnish scientific journal, was one of the
teachers at a course on scientific writing. Late one evening, he told
us, he was perusing a manuscript in which only half of the letter
“a” was legible. Page after page, that half-letter pursued him until
eventually he began to feel vaguely that this must be something
directed at him personally.
I didn’t see Fogelberg again until 12 years later at a meeting
of editors. I mentioned the damaged typeface, without really
expecting that he would remember it. But he replied instantly, “It
wasn’t damaged. Much worse – it wasn’t cleaned.”
Does a dirty typeface of a mechanical typewriter, or an error
related to electronic word processing, really matter? Yes, because
editors know from experience that there is a close relationship
between a poorly prepared manuscript and poor science. So make
sure your manuscript looks carefully prepared; it may influence
editors and referees in your favor.
How to Write and Illustrate a Scientific Paper
Comments on
scientific language
A MEDLINE search showed that no fewer than 90 percent of
papers listed in Index Medicus in 1999 were written in English,
compared with 53 percent in 1966 (the year MEDLINE started).
The saying “Publish in English or perish” must therefore be taken
seriously. Regrettably, this means that many authors are obliged to
write in a language other than their native tongue – with all that
this can entail. Here I will share with you an episode from my own
experience as a non-native writer of English.
English as a foreign language
My first paper published in English was initially written in Swedish
and then translated into English by a professional translator.
“Brilliant,” I thought when I saw the translated version. But when
my supervisor read it, he shook his head and said, “Try to write
directly in English!” “Gosh,” I said to myself, thinking of my poor
grades in English at school, “I’ll never, ever be able to do that.”
But I decided to try and consulted the textbooks, which advised
me to read writers of fine English, such as Gibbon and his Decline
and Fall of the Roman Empire. I bought the book (running to
Comments on scientific language
3616 pages in three volumes!) but could find neither the time nor
the interest to read it.
Instead, I subscribed to the American weekly magazines
Newsweek and Time. As they often cover the same topics, the reader
is given the opportunity to learn twice, in different words, about
the same issues. I have found this very instructive. Time (weekly)
I have also found another method that has served me well.
When I have to tackle a new topic, I read leading English-language
publications, underline useful phrases and words, and then create
a list of the terms for each section (Introduction, Methods, etc.). I
noticed, however, that I seldom had to consult my list. During the
process of making the list, the brain seemed to have retained what
I had read and written.
I have hardly ever submitted a manuscript in English without
asking a linguist to look at it. Ideally, those correcting English
ought to be persons who: (1) not only are native speakers of English
but also live in your country and speak its language; (2) return to
their native country at least once a year to refresh their English;
and (3) have a knowledge of scientific writing. Correctors fulfilling
these criteria are a rare species. Many authors therefore have to
rely on English-speaking persons who, for instance, happen to be
working in their department or laboratory. That may not be so
bad, after all, because these persons are no doubt acquainted with
your field of research. But you must be aware that native-Englishspeaking researchers do not necessarily write good English – just as
not all Swedish researchers are good at Swedish.
I return to my early paper, translated from Swedish into English.
On rereading it 30 years later, I found to my embarrassment that it
didn’t express exactly what I meant to say, though I found the style
elegant. However, even clumsy writing would have been better
than this, had it conveyed the information accurately.
Why are papers in biomedicine often almost unintelligible?
Maybe an editorial in The Lancet (1995) had the answer when it
claimed that authors of scientific papers often write more to please
How to Write and Illustrate a Scientific Paper
the editor than to inform the reader. They dare not depart from the
traditional style for fear of having their work rejected.
Another mistake commonly committed by beginners is the
compulsion to be “complete.” Charlie Chaplin had something to
say about that.
Follow the “leitmotif”
The video film Unknown Chaplin (Brownlow and Gill 1983) shows
unused sequences from Chaplin’s productions. Some of them are
far funnier than those actually included in the final versions of his
films. Why were they excluded? Chaplin gives the answer in his
autobiography (Chaplin 1973). “If a gag interfered with the logic
of events, no matter how funny it was, I would not use it.” You
are thus recommended to do as Chaplin did and resist the urge to
include every item of evidence obtained. In other words, do not
include observations that depart from the main theme – no matter
how interesting these may seem to be (you will probably find space
for them elsewhere, or they could give rise to hypotheses to be
tested in future studies). However, if such information cries out
to be mentioned, you can insert it parenthetically – as I did in the
previous sentence.
Researchers are often short of time. I once heard of a scientist
who only had time to read papers while driving to work! That is
one reason for keeping a paper short; another is that superfluous
words obscure the meaning.
In the following paragraph, adapted from Kesling (1958), 36 of the
53 words can be omitted: .
Our research, designed to test the fatal effects of PGF2_ on dogs,
was carried out by intravenously introducing the drug. In the
Comments on scientific language
experiments, a relatively small quantity, 30 mg, was administered
to each animal. In each case, PGF2_ proved fatal; all 10 dogs
expiring before a lapse of five minutes after the injection.
Seventeen words are enough:
Intravenous injection of 30 mg prostaglandin PGF2_ to each
of ten dogs killed them within five minutes.
“Omit needless words!” is Rule 17 in Strunk and White’s The
Elements of Style (2000). In the introduction to the third edition of
the book, E. B. White, a pupil of Strunk, tells us that his teacher
omitted so many needless words in his course in English that he
would have been left with nothing more to say at the end of his
lesson if he had not used a simple trick: he uttered every sentence
three times, “Omit needless words! Omit needless words! Omit
needless words!” Strunk, William, Jr. White, E. B.
But do not go too far. The telegraphic style of the following
sentence taken from Contraception must be a riddle to a nonspecialist:
Young mature Sprague Dawley rats (200 g) (Charles River Italia)
were [used].
What do “young” and “mature” mean? What do “Sprague Dawley”
and “Charles River Italia” stand for? And did all the rats weigh
exactly 200 g? The average reader is probably better served by
The rats used in this experiment were obtained from Charles
River Breeding Laboratories and were derived from the Sprague
Dawley strain. The animals were sexually mature, 100 days old,
and weighed 190 to 215 g.
How to Write and Illustrate a Scientific Paper
Most writers no longer use male pronouns (he, his, him) to denote
both males and females. Does this mean that our language is less
sexist now? No. Instead, we have got constructions such as he/she or
s/he, which hardly solve the problem, but rather emphasize it. Here
is an example from a manuscript: gender
Each patient was interviewed at the out-patient unit that s/he
belonged to.
How to avoid constructions like this? The simplest way is often to
use the plural:
All patients were interviewed at the out-patient unit they belonged
On the odd occasion where the use of the plural seems impossible,
reword the sentence or try to remove the pronoun. For example, in
the following, the pronoun their could be removed.
I submitted the manuscript to the editor for their consideration.
Only when all else fails, use the less awkward form he or she. Finally,
I must relate an anecdote by Sheila McNab (1993).
In a serious road accident a father was killed and his son seriously
injured. When the boy was later brought into the hospital
operating theatre, the surgeon blanched and exclaimed, “I can’t
operate on this boy, he is my son!”
If you were unable to realize immediately that the surgeon was the
boy’s mother, you may have something to think about. When I tested
this anecdote on my graduate students, one male student could find
only one answer: the man who had died was the stepfather!
Comments on scientific language
Active or passive voice
Previously, scientists were obliged by tradition to use the passive
voice. The use of first-person pronouns (I or we) was seen as
pretentious, even impolite. Not so now. Scientists of today dare do
what Watson and Crick, back in 1953, had the courage to do in
the opening phrase of their classic on the structure of DNA – and
say we:
We wish to suggest . . . ,
which is more direct, easier to read, and shorter than the passive:
In this letter a suggestion is made . . .
Below is another example, drawn from New Scientist (1993). Its
former editor, Bernard Dixon, found the following sentence in a
manuscript submitted:
The mode of action of anti-lymphocytic serum has not yet
been determined by research workers in this country or
Dixon replaced it with:
We don’t know how anti-lymphocytic serum works.
“He was quick to telephone me,” Dixon recalls, “complaining about
editorial interference. [. . .] How could a magazine as prestigious
as New Scientist change an author’s meaning in such a cavalier
fashion? But, I replied, we had not altered his meaning. We had
simply made a sentence more readable and direct – and cut it to a
third of its original length.”
However, in methods and results sections the passive voice is
generally more effective. It emphasizes the action rather than the
person performing the action. Thus, the active form:
I stopped cell growth with colchicine
How to Write and Illustrate a Scientific Paper
has no real advantage over the passive:
Cell growth was stopped with colchicine
since nobody cares who performed the act. And further, when there
are several authors, the we in:
We stopped cell growth with colchicine
is probably not true – unless the authors each added a portion! Thus
active and passive voices both have their place in scientific writing.
Only two tenses are normally used in scientific writing: present
and past (Day 1995; Day and Gastel 2006). Present tense is used for
established knowledge (including your own published findings),
past tense is used for the results that you are currently reporting.
Most of the abstract section describes your own present work;
it is referred to in the past tense. Much of the introduction section
emphasizes previously established knowledge; given in the present
tense. Here is an example (Dembiec et al. 2004; emphases are
Tigers are often transported [but] the effect of transfer on them
has not yet been documented [2]. . . .
The methods and results sections describe what you did and found;
they appear in the past tense:
We simulated transport by relocating five tigers in a small
individual transfer cage. . . .
Average respiration rate of all tigers increased. . . .
Comments on scientific language
Finally, in the discussion section, where you compare established
knowledge with your own findings, you normally see-saw back and
forth between present and past tense – even in the same sentence.
Noun clusters and modifiers
In USA TODAY (October 13, 1992), I saw this:
Pig liver transplant woman dies
As a newspaper headline this phrase is acceptable. It is intelligible
and unambiguous; and the cramped space makes it necessary. But
in a scientific paper, such a sentence would have looked ridiculous.
Here, it has been written out in full:
The woman with a transplanted pig liver has died
The following phrase, quoted from Contraception, may be entirely
and immediately intelligible to an expert in the field:
Rabbit anti-mouse spleen cell serum . . .
But researchers not working in that field might wonder to which
animal the spleen had belonged. The writer could have saved some
readers a little trouble if he had written:
Anti-mouse serum of rabbits immunized with cells of mouse
spleen . . .
However, it is quite acceptable to couple a few nouns and modifiers
as long as it is crystal-clear what you mean and as long as the reader
can grasp the string of words at first reading, as in this example
from a methods section (Mehrotra et al. 1973):
Colony bred female albino rats . . .
and this used as a subheading (Gardiner et al. 1980):
Anaesthetized spontaneously breathing guinea pig
How to Write and Illustrate a Scientific Paper
Prevalence and incidence
The words prevalence and incidence are said to be among the most
misused terms in biomedical reports. Prevalence refers to the total
number of cases of a disease or condition existing at a specific time.
Incidence refers to the number of new cases that develop over a
specific time. In the following example from Newsweek (Begley
1996), the prevalence is 200 000 and the incidence 12 000.
Each year as many as 12 000 Americans join the more than
200 000 who already live with paralyzing spinal-cord injuries.
Avoid the use of “respectively”
Respectively obliges the reader to stop and reread the sentence, as in
the following, seen in a manuscript under preparation:
Phytate reduction in wheat, rye, barley with and without hulls
incubated with 40 g water/100 g cereal for 24 hours at 55 °C
was 45, 56, 48 and 77%, respectively.
The version below is direct and permits the reader to proceed
(revised text in boldfaced italics):
After incubation with 40 g water/100 g cereal for 24 hours at
55 °C, phytate reduction in wheat was 45%; in rye, 56%; in
barley with hulls, 48%; and in barley without hulls, 77%.
The “and/or” construction
The expression and/or disrupts the textual flow, as in this example:
The effect of intravenous streptokinase and/or oral aspirin . . .
which the reader would have found easier if it had read:
The effect of intravenous streptokinase, oral aspirin, or both . . .
A closer look at the text often reveals that and/or can be replaced by
Comments on scientific language
and (The ACS Style Guide 1977):
Our goal was to confirm the presence of the alkaloid in the
leaves and/or roots,
or by or (de Looze 2002):
Confidential information can only be given to the patients
and/or close relatives.
The construction and/or has no place in scientific writing.
Unnecessary hedging
Hedging is a way of saying “maybe” more than once. Two or more
hedges can drain all force from a sentence. The eminent writer in
the cartoon replaced seven hedges (“seems,” “not inconceivable,”
“suggest,” “may,” “indicate,” “possible,” “probably”) with just one:
“think.” One hedge is always enough.
Figure 2.1 The author seen in the figure thought twice before presenting his
message. (Redrawn, with permission, from Majewski 1994.)
How to Write and Illustrate a Scientific Paper
How old is young?
Consider this title of a paper:
Herniography in younger women with unclear groin pain
The abstract of this paper stated that the women were under
40 years of age. Well, viewed from my age these women were
young, but the average readers would probably have appreciated
a more precise definition of their age, as in this example (Sundby
and Schei 1996):
Infertility and subfertility in Norwegian women aged 40–42.
Prevalence and risk factors.
In other cases, an age range can be defined by a specific term, as in
this title (Gold et al. 1996):
Effects of cigarette smoking on lung function in adolescent
boys and girls
You are advised to use specific terminology, when available, to
report subjects’ ages. Here are the age groups recommended by
MEDLINE , as of January 2007:
All infants
All children
All adults
Preschool child
Middle aged
Middle aged + aged
80 and over
Comments on scientific language
birth–23 months
0–18 years
19+ years
birth–1 month
2–23 months
2–5 years
6–12 years
13–18 years
19–44 years
45–64 years
45+ years
65+ years
80+ years
Don’t ask me why the “adult” category excludes persons aged 18
years. Note that young is not listed – it is undefinable.
Persons below adult age may be referred to as boys and girls. For
adults, men and women are the correct terms.
Avoid synonyms to achieve elegant variation
In the list of abbreviations of a manuscript under preparation I found
this (boldface italics mine):
Cmax: maximum plasma concentration achieved.
Tmax: time at which the maximum plasma level was achieved.
Even if “plasma concentration” and “plasma level” here are true
synonyms, using both in the same paper may confuse your readers.
Choose one and stick to it. Scientific writing is not literary writing.
The remote verb
One of the most common errors in scientific writing is the use of
the “remote verb.” In the sentence below, 37 words and numerals
separate the subject (children) from the verb (were invited), quoted
from a dissertation:
All children (n = 99, 54 boys and 45 girls) born between 1990
and 1995, adopted during 1993–1997 from Poland, Romania,
Russia, Estonia, and Latvia through authorized adoption
agencies in Sweden and living in the region of Västra Götaland,
were invited to participate in the study.
The separation could have been avoided by beginning the sentence
with the first person, active voice, followed by the original last five
We invited to participate in the study all children . . .
How to Write and Illustrate a Scientific Paper
Drafting the
As no two authors write in the same way, no one can say which way
of writing will suit you best. You will have to find out for yourself.
The writing procedure described here is the one I personally have
found most useful – by trial and error. Hopefully you may find
some portion of it to adopt.
A central part of this writing scheme is to collect ideas while the
study is in progress.
Write down your thoughts as they arise
While the study is still in progress, jot down ideas as they occur
to you. The notes can be assembled, for example, in a loose-leaf
binder containing plastic sleeves, one for each section of the paper.
(Woody Allen, the moviemaker, works in a similar way; in a drawer
he gathers slips of paper with ideas for his forthcoming movie.)
Ideas can pop up anywhere – in bed, in the bath, in the street,
on the bus, on the train, in the car. So, place your notebooks
strategically so that you always have one at hand, wherever you are.
Use one sheet of paper per idea, even if the idea is only a single line
Drafting the manuscript
or phrase. Eventually, the reservoir may contain all the components
of the paper (or film script), waiting to be arranged.
Where and when to write?
As a beginner I made the cardinal error of taking two weeks off
and sitting down on a Monday morning in an attempt to write
the first draft continuously from beginning to end. It didn’t
work. Professional writers don’t do it that way. They know from
experience that they can work creatively for only a few hours per
day. They also know the importance of working uninterruptedly,
with no phone ringing and no visitors arriving. For example, when
Vilhelm Moberg, author of the great epic about Swedes leaving
their homeland for America, was at one time writing in California,
he could find only one place to work undisturbed – in the attic of
the house. No one could reach him there, because he pulled the
ladder up behind him.
Writing an epic is, of course, not the same as writing a scientific paper, whose well-defined sections can be used to divide
the text into separate stages. Short sections such as abstract and
introduction may be written in a single session each. Long sections
such as results and discussion might have to be split into smaller
parts, each to be written in one session.
How it can work in practice
Assume that during this particular sitting you intend to write the
introduction. You have three hours at your disposal. Before starting,
read and revise what you wrote during the previous sitting. Then
read the notes you have collected for the introduction. Let us say
that writing this section takes only about two of the three hours
you have available. Nevertheless, stop writing now – it will give you
a feeling of accomplishment. However, before you finish for the
day, read the notes you have collected for the next part and sketch
the main topics in brief, incomplete sentences.
How to Write and Illustrate a Scientific Paper
Even if you still have 20 minutes to spare, and are still full of
energy and creativity, do not start writing the next part. If you
do, you may have to leave the work uncompleted, with a feeling
of dissatisfaction. Ernest Hemingway once said about writing,
“Always stop on a high,” and that is exactly what you do if you
always stop when one part is finished.
Medical researchers with clinical duties rarely have as much
as three hours of uninterrupted time available. But this writing
program can be used for shorter (1–2 hour) spells if you adjust
the pieces of work accordingly. A great advantage of this writing
scheme is that you need not write every day.
The other way around
You do not have to begin with the abstract or the introduction.
You may begin by writing the easiest section, which could be the
methods or the results. This approach offers a psychological advance.
Starting with the information you know best (the methods or the
results) gets about a third of the paper done quickly, and you look
forward to writing more. Then, feel free to write the remaining
sections in whatever order you find easiest.
Handwriting or word processing?
Handwriting may be suitable for the first draft, but word processing
is without doubt the easiest method for revision. If you are going
to revise a section extensively, make a copy of the original version
and save it in a separate file – you may need it if you change your
Drafting the manuscript
Choosing a journal
You will most probably find the right journal for your paper among
those periodicals you most often read. That is where you have your
readership. Submission
If you think that more than one journal seems appropriate, you
may wish to rank them by quality. One way to do so is to look at
the “impact factor,” which tells how often the average article of a
journal is cited. Such information is provided by the Institute for
Scientific Information in its annual Journal Citation Reports.
The impact factor is especially useful for comparing journals
within a particular field of research. Let us take, for example,
Orthopaedics. The 41 journals listed for 2005 had an impact factor
in the range 0.1–4.2, with a median of 0.9. It is reasonable to
assume that journals with an impact factor of 4.2 attract the best
papers in the field, and that these journals have a greater impact
on science in that field than a median (0.9) impact-factor journal.
However, if you select a high-impact journal, the publication of
your paper may be delayed, as is hinted at in this question from a
course participant:
Should I send my paper to a journal with a high impact factor
and risk having it rejected, or should I send it to a journal with
a lower impact factor and get it published quicker?
How to Write and Illustrate a Scientific Paper
If you feel in your heart that yours is a first-class paper, then try
the high-impact journal – provided that it is a specialized journal
in your own field. However, if it is a journal outside your specialty
and your paper is accepted and published, this journal might turn
out to be a publication that researchers in your specialty do not
read. For example, a colleague of mine complained that his excellent
paper published in one of the highest-ranked medical journals, The
Lancet, was not cited. However, when you have been around for a
while, you may feel by intuition which journal is the right one.
The impact factor ranks journals; it does not evaluate individual
papers. Some articles may not be cited at all, while others become
classics. Although it may be outside the scope of this book, I will
tell you about one way to find the best papers in your field: visit the
website Faculty of 1000 Medicine ( or
Faculty of 1000 Biology ( These sites rate
individual papers according to their merit, irrespective of where
they are published. However, you must be aware that a top-ranked
article may not necessarily be well written.
Instructions to authors
When you have chosen a journal, the next step is to read the
current version of its Instructions to Authors. Several journals
print these instructions in every issue, others only in the first issue
of each volume. They also appear on the website of the journal.
If you work in a biomedical discipline you will find that many
journals use “Uniform requirements for manuscripts submitted to
biomedical journals” (Vancouver Document,, a
set of instructions intended to allow authors to use the same format
and style for papers submitted to different journals.
Choosing a journal
Preparing a graph
Assuming that your results show trends or movements over time,
such as nicotine concentration in plasma after smoking, a good
way to display your data would be to construct a line graph. But
do not rely on the computer to design it for you. Here are some
common errors.
The line graph
Figure 5.1 The effect of tyramine solution on pupillary size. (Adapted, with
permission, from a draft by Havelius 1994.)
How to Write and Illustrate a Scientific Paper
This seemingly excellent line graph nevertheless appears to have
two common defects: the curves are distinguished both by type of
line and by type of data-point symbol – either would suffice; and
the curves are identified by a separate key, obliging the reader to
scan back and forth to the key to see what they represent.
In the two redrawn graphs (5.2), the curves are labeled directly
and distinguished either by type of data-point symbol or by type
of line.
Open and filled circles, as in the left graph, are the data-point
symbols easiest to distinguish. They can also be used symbolically;
for instance, if an experiment has been performed with ( ) and
without ( ) treatment, the emptiness of the open circle suggests
that nothing has been administered.
Figure 5.2 Alternative displays of Figure 5.1
Other standard symbols for data points are open and filled squares
and triangles (
). If you need more symbols, you probably
have too many curves for one graph, and you should consider
dividing it into two or presenting your observations in a table.
In the right-hand graph you will probably not miss the data
points, as you can easily discern the change of line direction where
the points have been omitted. This graph may be the more attractive of
the two. Data points are probably overused in scientific papers.
Preparing a graph
Relationship between the lengths of the axes
In the following graph (5.3), the sharp decrease in the first part
of the line has been exaggerated in two ways: (1) the vertical axis
is longer than the horizontal axis and (2) the horizontal axis is
contracted because the distance between the first two tick marks
represents four hours whereas the same distance between the
following ticks represents only one hour.
PRL (ng/mL)
–12 –8 –7 –6 –5 –4 –3 –2 –1
Time (hrs)
Figure 5.3 The original line graph exaggerating the decrease. (Reproduced with
permission from Acta Obstetricia et Gynecologica Scandinavica 2001;80(1):34–8.)
The golden ratio, which is close to the format 3:2, is the most
aesthetically pleasing. But the 3:2 may invite misuse. Therefore,
the relationship between the axes should normally be 1:1, as in the
redrawn figure (5.4) on the next page.
Time point zero represents the time of delivery; this information
is given in the main text, but it could have been included in the
How to Write and Illustrate a Scientific Paper
Figure 5.4 Recommended version of Figure 5.3
Labeling axes
The reader must be in no doubt as to what the axes show. A few
drawing programs do not label the vertical axis parallel to the axis
but instead place a line of text above the figure, like a title. I have
even seen this practice used to describe the horizontal axis! So
always place the label parallel to the axis.
A graph is intended to show a trend, not exact figures. Therefore,
the number of tick marks should be limited. On the vertical axis,
the “1–2–5 -principle” is often used; that is, the axes are divided
into intervals of 1, 2, 3, …; 2, 4, 6, …; or 5, 10, 15 …; and not
7, 14, 21, etc. You may, if necessary, multiply by 10 (Figure 5.4)
or 100, but preferably not by 1000. As scientific units are often
expressed in multiples of thousands, three zeros are easily removed
by altering the unit, for example, from microgram to milligram.
Preparing a graph
The chart
A time series can also be displayed as a set of vertically arranged
bars, known as a column chart in most computer programs. Here
is an example.
Figure 5.5 Reported cases of diarrhoeal disease (cholera, dysentery, and
dehydration). (Reproduced, with permission, from Goma Epidemiology Group,
Public health impact of Rwandan refugee crisis: What happened in Goma, Zaire, in
July 1994? The Lancet 1995; 345(8946): 339–44, © The Lancet Ltd.)
This brings us to the question: Which type of presentation is preferable for a time series such as this – column chart or line chart?
As a basis for discussion I have redrawn this time series as a line
graph (5.6). To avoid ambiguity, I have converted x103 to thousands
on the vertical axis. Note that the curve is bolder than the axis
lines. Note also that the axes are separated, as zero is not common
to both.
Now let us compare the two designs, step by step. A column
chart may be preferable when there is no carry-over effect from
one time period to the next, that is, when each column represents
a new set of data with no addition from the preceding time period.
For example, the annual number of births could be shown as a
series of columns, while the total population could be plotted as a
continuous line (Chapman and Mahon 1986). As there is no carryover effect in the example of diarrheal disease, the lines joining the
data points are, strictly speaking, artifacts. Those who worry about
such things might prefer a column chart.
How to Write and Illustrate a Scientific Paper
Figure 5.6 Alternative display of Figure 5.5
Column charts are said in certain cases to exaggerate differences
between individual measurements. If this is so, it could be a reason
for not using column charts in such cases.
A figure should have no unnecessary lines. The column chart
(5.5) has 72 lines (three for each of the 24 columns), whereas the
curve graph has only one line. In a line graph the viewer can usually
see trends or movements more quickly than in a column chart.
Which graph would you have chosen in this case? I would probably
have preferred the line graph.
One type of illustration most readers dislike is those grouped
column charts that have more than two or three categories in each
group, as in the following example (5.7).
Figure 5.7 Beverage and gastric contents pH. (Reproduced with permission from
Medicine and Science in Sports and Exercise 1993; 251(1): 42–51.)
Preparing a graph
The key to this figure needs a key of its own. W, for example, appears
not to stand for watt, as it generally does, but water. C doesn’t
stand for carbon, but controls. Suppose you now look away from the
columns. What do you remember having seen, without looking
back? This type of data might well do better in a table.
Two or three categories in each group should be the maximum
in the grouped column chart. Figure 5.8 with two categories is easy
to grasp.
Figure 5.8 Probability of dying in a coronary care unit after admission with initial
working diagnosis of acute myocardial infarction. (Reproduced, with permission,
from Clarke et al. 1994.)
One way to remove the separate key in this graph could be to label
the first group directly, for example:
However, the main role for column charts is not to show time
series (such series are usually better shown by line graphs), but
to display categorical data. The following graph (5.9) presents
such data. What makes this graph especially good is that it shows
How to Write and Illustrate a Scientific Paper
the precise percentage for each of the columns. Such a design
combines the virtue of the table (giving exact values) with that of
the graph (quickly presenting the message). Note that the bars are
wider than the spaces between them. Note also that the bars have
a gray tone (somewhere in the middle of the gray scale), which is
more pleasing to the eye than a black or white tone or a striped
Figure 5.9 Intra-abdominal injuries associated with diaphragmatic rupture due
to blunt trauma. (Reproduced, with permission, from Sarna and Kivioja 1995.)
However, the cramped space of a journal column (about 8 cm)
allows a column chart to include only a few items. This chart is on
the verge of being overcrowded. Below the two columns on the far
right, the two separate texts nearly run into each other: Rupture of
Pancreatic . . . Such a problem can be overcome with the use of a
bar chart, the computer term for horizontally arranged bars.
The next figure (5.10) is a good example of a bar chart. The bars
are arranged in decreasing order of size. The chart also shows the
exact numerical value of each bar.
Preparing a graph
Figure 5.10 Annual alcohol consumption per inhabitant. (Reproduced, with
permission, from van Os and Neeleman 1994.)
We will turn to quite another item: the presentation of individual
data. I will show how much more information you can give the
reader by displaying raw data instead of summary data.
The following graph (Figure 5.11) shows individual data and
can be regarded as an opened-up column chart. At the bottom of
the right column, seven outliers (outside values) attract attention
(arrow mine).
How to Write and Illustrate a Scientific Paper
The seven outliers turned out to have an atypical variant of
Fabry’s disease, a serious metabolic abnormality. Had a summarydata graph been used instead (with columns having almost the
same mean and standard deviation), the outliers would have been
effectively camouflaged.
Figure 5.11 Plasma _-galactosidase activity in 89 normal male subjects and
230 male patients with left ventricular hypertrophy (LVH). (Reproduced, with
permission, from Nakao et al., An atypical variant of Fabry’s disease in men with
left ventricular hypertrophy, N. Engl. J. Med. 1995; 333:288–93. Copyright © 1995
Massachusetts Medical Society. All rights reserved.)
Preparing a graph
Now we will do the opposite. We will peek behind the columns
of a summary-data chart (5.12) at the raw data to find out what
they can tell us. The chart I have chosen shows the time needed to
induce abortion in the second trimester when using a conventional
method (prostaglandin alone) compared with a new approach
(prostaglandin + oxytocin).
Clinicians in the field seeing this graph may at first be impressed
by the seemingly dramatic shortening of the abortion time. However,
they would probably be less impressed when viewing the redesign
(Figure 5.13) consisting of individual data from the 53 subjects;
Figure 5.12 Interval (mean and SD) from induction to abortion for women treated
with intra-amniotic prostaglandin F2_ (PGF2_) alone or in combination with
intravenous oxytocin. (Reprinted from Prostaglandins, 2, M. Seppälä, P. Kajanoja,
O. Widholm, P. Vara, Prostaglandin-oxytocin abortion: A clinical trial on intraamniotic prostaglandin F2_ in combination with intravenous oxytocin, 311–9,
1972, with permission from Elsevier Science.)
How to Write and Illustrate a Scientific Paper
data obtained from two spike bars (also called needle bars; Harris
2006) in the same paper.
Figure 5.13 Alternative display of Figure 5.12
The redrawn graph shows that the improvement was not as clearcut as might be expected from the summary-data chart. I also
combined the graph with a table containing complementary data,
so that the reader does not need to search in the main text for this
information – a design recommended by Altman (1995, 40).
Altman (1995, 222) and Tufte (1983, 13–14) show with illustrative examples how the same summary data can rest on entirely
different sets of raw data. So before you start writing your paper I
recommend that you make, at least in sketched form, a graph with
your individual observations plotted, to see what you might find.
Preparing a graph
If you eventually decide to present your results as summary data,
avoid the following type of display.
Figure 5.14 Data (mean ± SE) compare rats adapted to low-salt with those
adapted to high-salt intake. A, kidney cortex. (Reproduced from Welch et al. 1997,
with permission from the American Physiological Society.)
To indicate a single mean value with a column is redundant. In the
following redesign the columns have been removed and the means
shown with data points.
Figure 5.15 Alternative display of Figure 5.14
How to Write and Illustrate a Scientific Paper
Here are some further comments on the original figure (5.14).
There is no need to distinguish the bars by making one of them
white and the other black; both bars in a gray tone would have been
preferable (see Figure 5.9). Note also that the label of the vertical
axis is too long, extending beyond the axis. An asset of the graph,
however, is that it presents the level of probability with a P value
rather than an asterisk. The use of asterisks (*P < 0.05, **P < 0.01,
***P < 0.001) should be avoided; instead, give exact P values (the
exception is P < 0.001; see Chapter 22, “P < 0.05 ≠ the truth”).
The box-and-whisker plot (box plot) has become a popular form
of presentation of data. As there are many variations (Harris 2004),
you will have to explain the details of the plot, as in the following
example from The Lancet (Chaparro et al. 2006):
Figure 5.16 Box-and-whisker plot of two-way interaction effect of treatment group
and maternal ferritin on infant body iron (mg/kg) at 6 months of age
Boxes represent the inter-quartile range (25th to 75th percentile), and whiskers indicate
the 5th and 95th percentiles for unadjusted data. The notch in each box represents CI
about the median, represented by horizontal line at the middle of the notch. Additional
horizontal line represents the mean of each subgroup. […] (Reproduced from The Lancet,
367, Chaparro et al., Effect of timing of umbilical cord clamping on iron status in Mexican
infants: a randomised controlled trial, 1981–9, © 2006, with permission from Elsevier.)
Preparing a graph
If you are uncertain whether to present your results as summary
data or as individual values, you can use both presentations, as in
Figure 5.17.
Figure 5.17 Mean (±SD) serum HCV [hepatitis C virus] RNA titers in the 33 mothers
with uninfected infants and the 7 mothers with HCV-infected infants. (Reproduced,
with permission, from H. Ohto, S. Terazawa, S. Sasaki, N. Sasaki, K. Hino, C. Ishiwata
et al. Transmission of hepatitis C virus from mothers to infants. N. Engl. J. Med.
1994; 330:744–50. Copyright © 1994 Massachusetts Medical Society. All rights
Another way to present both individual and summary data is to
combine an individual-data graph with a table, as in Figure 5.13.
One can wonder why these two informative ways of presentation
(Figures 5.13 and 5.17) are so seldom used.
Let us now discuss the three-dimensional graph. Today’s
computer technology renders the three-dimensional graph easily
displayed. As a result, this type of graph is seen more and more
How to Write and Illustrate a Scientific Paper
often in published reports. Unfortunately, the ease with which
it can be created often leads to its use even in cases where the
data have only two dimensions. A third dimension is thus falsely
introduced in such cases. Figure 5.18 is a typical example.
Figure 5.18 Mortality from coronary disease among Japanese-born men 55 to
64 years of age residing in Japan, Hawaii, or California, 1950. (Reproduced, with
permission, from Reed 1990.)
After conversion into a two-dimensional display, the graph is
easier to read (Figure 5.19).
Figure 5.19 A two-dimensional display of Figure 5.18.
Note that the bars have been shifted apart and given the same
gray tone. The bars are wider than the spaces between them. Note
also that the vertical axis does not extend beyond what the graph
demands, numbers on the scale are reduced, and tick marks point
Preparing a graph
A true third dimension is extremely uncommon in research results.
However, Tufte presented an excellent example of this rare species
in his classic The Visual Display of Quantitative Information (1983,
42), where he depicted air pollution over six counties in southern
California. Only thanks to the three-dimensional presentation
can one distinguish between the peak over San Bernardino, in the
background, and that over Los Angeles.
Pie charts may be of two kinds: three-dimensional, called pie
charts, or two-dimensional, called simple pie charts.
The three-dimensional pie chart is less suitable for presentation
of scientific data, as it is difficult to compare three-dimensional
segments (especially between charts of different sizes) and because
of the waste of space (few data per area of graphic). Tufte (1983,
178) says that “the only worse design than a pie chart is several of
The simple pie chart, however, is used increasingly in scientific
papers. As it is more illustrative than scientific, its usage may be
appropriate in a magazine article. Although I am not a supporter
of its employment in scientific writing, I will nevertheless describe
how to utilize it. A good simple pie chart has four characteristics
(Figure 5.20): (1) the largest segment begins at 12 o’clock; (2) it
continues with proportionally smaller portions in the clockwise
direction; (3) the number of segments does not exceed five; and
(4) labels are placed outside the circle. For emphasis, one sector
can be separated slightly; most software programs allow you to
do this. If, however, the space is cramped, the content of this
figure (5.20) could be given in the running text as follows:
Of the 20 patients studied, 12 had myoma uteri; 6, adenomyosis;
and 2, endometriosis.
How to Write and Illustrate a Scientific Paper
Figure 5.20 Clinical diagnosis of the patients studied. (Adapted, with permission,
from a draft by Chyi-Long Lee and Yung-Kuei Soong 1991.)
Before submission
Any graph must withstand reduction to a journal column width
(about 8 cm). After reduction, the text of the axis labels should
be similar in size to the running text. Reduce your graph on a
photocopier to see if that will be the case.
When manuscripts are submitted electronically, illustrations may
either be included with the text or uploaded separately. Consults
the Instructions to Authors of the journal you have chosen to find
what they prefer.
Preparing a graph
Although my first drawing could have been published with only
slight improvements, I let an artist redraw it, by hand. (We had no
computers in those days!) The picture was improved – but it cost
me a small fortune. Ever since, I have done my drawings myself.
This is perhaps not such a bad thing after all, because the pictures
will express exactly what I mean.
Such a drawing, done by one of the authors of a paper, is shown
Figure 6.1 How to prevent injuries to internal organs when inserting the trocar before
performing a laparoscopy. (My translation from the Swedish. Illustration adapted,
with permission, from Samuelsson and Sjövall 1973.)
How to Write and Illustrate a Scientific Paper
The report was initiated by a serious complication to a laparoscopy,
where a trocar (a sharp instrument) was stabbed into the large
intestine during the initial phase of the procedure. The paper
describes how this complication could be prevented – and the
drawing tells exactly what the authors wanted to say. The picture
cost them nothing!
Authors who wish to have their drawings done by a professional
artist are recommended to provide the artist with a detailed sketch.
The artist then scans the sketch into a computer where it is improved
– much cheaper than drawing it by hand. Skilful illustrators,
however, are a rare species and usually have a tight schedule. You
should therefore contact them in good time; preferably before you
begin to write the paper.
The following pair of figures shows the sketch (above) done by
the author and the improved drawing (below) by the artist. Every
detail is enhanced and given a professional look.
Figure 6.2 Drawing in sketch form (above) and final form (below). (The sketch
form printed with the permission of the author, Joanna Wallengren. The final form
reprinted from Journal of the American Academy of Dermatology, 39, Wallengren,
J. Brachioradial pruritus: A recurrent solar dermopathy. 803–6, Copyright (1998),
with permission from The American Academy of Dermatology, Inc. Illustration by
Ronny Lingstam.)
Figure legends
A figure legend or caption may include two items:
(1) the title, which states the topic of the figure, and
(2) the message, which explains the contents of the figure.
In some cases a message alone will suffice.
However, most legends do not convey a message. Figure 7.1,
with its typical legend, could be improved as shown in Figure 7.2.
Figure 7.1 Cumulative weeks to delivery of women in group A (n = 78) and group
B (n = 78). (Reproduced from a manuscript in preparation that after revision
was published in Acta Obstetricia et Gynecologica Scandinavica 1988; 67:81–4;
with permission from the author and Munksgaard International Publishers Ltd.,
Copenhagen, Denmark.)
How to Write and Illustrate a Scientific Paper
In the revised illustration, Group A has been changed to Treated
women and Group B to Controls so that the reader does not have to
search in the main text to interpret these terms. The tick marks now
face outwards, so that they do not encroach on the curve between
26 and 30 on the horizontal axis.
Figure 7.2 Outcome of twin pregnancy in women prescribed either prophylactic
leave of absence from work (treated) or not (controls). Gestational duration did not
differ between the groups.
An illustrative example of a legend including the message alone is
presented in Figure 8.1.
Figure legends
How to design tables
The content of a table may be either descriptive, as is most often the
case, or declarative (bearing a message). It helps the reader if this is
reflected in the table’s title.
The descriptive title
The descriptive title is used for tables that present detailed information, such as the one below (only part of the original table is shown).
Table 8.1 Maternal age, gestational age, indication, size and type (avascular
or vascular) of villi sampled, method used in processing the biopsy (direct
preparation, 24 h culture, long-term culture), and karyotype in 80 diagnostic cases
of first-trimester chorionic biopsy
Villi sampled
GestaCase Maternal tional
age (y) age (wk) Indication
Weight Avas- Vas(mg) cular cular
Previous child
Mb Down
Cytogenic method
24 h
Direct culture culture Karyotype
Source: Reproduced from Heim et al. 1985, with permission from Munksgaard International Publishers
Ltd., Copenhagen, Denmark (partial table).
How to Write and Illustrate a Scientific Paper
This table is from a paper of which I was a coauthor. Now, with the
benefit of hindsight, I see that this title contains the common error
of repeating every single heading, except, in this case, the first one.
No fewer than 31 of the 39 words in the title can be deleted. The
shortened version can be grasped at a glance:
Table 8.1 Details of 80 diagnostic cases of first-trimester chorionic
The declarative title
If a table shows a clear trend or relationship, a declarative title
could be preferable. The following table shows that the frequency
of fractured clavicles increased with increasing birth weight.
Table 8.2 Fractured clavicles and birth weight
Birth weight, g
Fractured clavicles
Source: Reproduced from a manuscript by Jójárt et al. 1992, with
But the title is neutral and one has to study the table for a while in
order to grasp the message. It would have helped the reader if the
table’s message had been stated in the title:
Table 8.2 Increase in fractured clavicles with birth weight
How to design tables
Rounding off
In the table’s far right-hand column, the percentages shown are overprecise. One decimal would suffice (see Chapter 20, “Percentages”). In
the middle column, the percentages add up to exactly 100.0 percent.
The authors have achieved this by adjusting the true percentage of
3.8 to 3.7 (second figure from bottom). Do not on any account adjust
figures to make them add up to the true total. Instead, give the total
after rounding (in this case 100.1 percent) and explain in a footnote,
“The sum of the percentages exceeds 100 percent due to rounding.”
Having said this, I also want to emphasize the assets of this table.
What is especially good about it is that it makes only one single
point. If, on the other hand, a table’s data lead to two conclusions,
the author had better try to divide the table into two smaller ones,
as two conclusions in the same table tend to obscure each other.
Another asset of this table is that numbers that are to be compared
follow down the columns, not across. They are easier to read that
way. (If you don’t believe me, try adding percentages from left to
right in a row!)
Table or graph?
As mentioned earlier, a table can be used either to report precise
numbers or to illustrate a trend. But often a trend is better illustrated
with a figure. Here is an example showing what I mean.
In its day, the following table (8.3) contained medical dynamite.
It presented the results of a study that showed, for the first time
in vivo, that certain anti-inflammatory substances, such as indomethacin or aspirin, inhibit the synthesis of prostaglandin. The
sensational outcome of this study is well hidden among the figures
within the table. Moreover, the title reveals nothing. And to open
a title with a long chemical name is deadening; Prostaglandin
metabolite would have sufficed; the full chemical name belongs to
the methods section.
How to Write and Illustrate a Scientific Paper
Table 8.3 Excretion of 7α-hydroxy-5, 11-diketotetranor-prostane-1, 16-diodic
acid in subjects receiving analgesics. Indomethacin (a, 4x50 mg/24 h [. . .] given as
indicated by asterisk.
Amount of metabolite (mg/24 h)
Day 1
Day 2
Day 3
Day 4
Day 5
Day 6
Day 7
2 .7
Source: Reproduced, with permission, from Hamberg 1972 (partial table).
Shown here (Table 8.3) is only the part of the table that includes the
results from the three subjects receiving indomethacin. I converted
this part into a graph (Figure 8.1) to show more clearly the dramatic
effect of indomethacin.
Figure 8.1 Urinary excretion of a prostaglandin metabolite decreased
following indomethacin administration in three humans
Note that the curves have the same type of line because they do
not need to be distinguished from each other; they are all intended
to show the same trend. The curves are bolder than the axes. Two
zeros are used to label the point where the axes meet. Tick marks
How to design tables
point outward. Axes are of equal length. The label of the vertical
axis is parallel to the axis and reads from bottom to top. Note also
that the legend gives the message of the figure.
On the next page, I will show how to design a table. To present
all these details in the running text would bore you. Instead, I
have arranged the instructions in note form around an imaginary
table. With the exception of the footnotes, the text of the table is
Once upon a time, typed repeatedly. So, you can concentrate on the
layout without being distracted by the content of the text.
How to Write and Illustrate a Scientific Paper
Typing the table
How to design tables
For every person who reads the whole of a scientific paper, about
500 read only the title (Kerkut 1983). One way to improve this
statistic could be to make the title declarative by including what
the paper says, not just what it covers.
Whenever possible, use a declarative rather
than a neutral title
This title is neutral:
Influence of aspirin on human megakaryocyte prostaglandin
John Vane, in his classic paper published in Nature in 1971, put it
more expressively:
Inhibition of prostaglandin synthesis as a mechanism of action
of aspirin-like drugs
(In 1988, Vane was awarded a Nobel Prize in acknowledgment of
his discovery. Vane told us how aspirin relieves pain.)
How to Write and Illustrate a Scientific Paper
The following declarative title is taken from the biological sciences
(Marvin 1964):
Birds on the rise
Goodman et al. (2001) recommended that the study design also be
included in the title, as follows (Lee et al. 1995):
Improved survival in homozygous sickle cell disease: Lessons
from a cohort study
However, the following formulation is not recommended (Quesada
et al. 1995):
Leaf damage decreases pollen production and hinders pollen
performance in Cucurbita texana
Do the authors really mean to imply that the issue is settled once
and for all? Your own present work should be referred to in the past
Leaf damage decreased . . . and hindered . . .
This way of using a verb in a title makes it into a sentence. It is
stronger than using a phrase, some say too strong. In a descriptive
study, however, you have to use a neutral title (Kitin et al. 2004):
Anatomy of the vessel network within and between the tree
rings of Fraxinus lanuginosa (Oleaceae)
Titles ending with a question mark
Scientists used to have the journal at hand when skimming
through the table of contents and could page straight to the article.
Today, most scientists skim lists of titles on a computer screen but
frequently are unable to get access to the article and sometimes
even the abstract. Thus, instead of a title in form of a question like
this (McWhorter and Martínez del Rio 2000):
Does gut function limit hummingbird food intake?
the reader would appreciate being told the answer from the beginning:
Limitation in hummingbird food intake by gut function
A review article can, however, have a title ending with a question
mark, because some papers reviewed say one thing; others, the
opposite. The title of a mini-review published in Nature (Pitnick
et al. 1995) presumably covering all or most facets of the debate is,
for example:
How long is a giant sperm?
Begin with the keywords
Not until the final words does the following title tell which disease was
being studied. Thus a reader who is in a hurry when scanning the table of
contents may not have time to find out what it is about.
The effect of calcium antagonist felodipine on blood pressure,
heart rate, working capacity, plasma renin activity, plasma
angiotensin II, urinary catecholamines and aldosterone in
patients with essential hypertension
When the keywords are placed at the beginning of the title, it is
immediately clear what disease was studied:
Essential hypertension: The effect of . . .
Use verbs instead of abstract nouns
The following is a common way of formulating a title:
Treatment of polycystic ovary syndrome
How to Write and Illustrate a Scientific Paper
but if you turn the noun into a verb, you will make the sentence
more dynamic:
How to treat . . .
Avoid abbreviations in the title
In The Lancet (1993), which asks us to avoid abbreviations in titles,
I saw the following heading for an editorial (the editor has denied
that it was a joke):
OCs o-t-c?
Written out in full it means “Oral contraceptives over-the-counter?”
or, in other words, should pharmacies be allowed to sell oral
contraceptives without a doctor’s prescription?
Admittedly, I read this editorial because its unintelligible title
caught my attention. So this instance could be regarded as an exception from the rule of thumb, not to use abbreviations in the title.
Of course, you are allowed to use abbreviations and symbols
that are more familiar than the words they stand for, such as DNA
and pH. In case of doubt, however, use both the term in full and
its abbreviation, as in this title (Stockdale 2000):
Contaminated material caused Creutzfeldt-Jacob disease (CJD)
in some undersized children who were treated with growth
hormone (GH)
A title for your thesis
In JAMA I saw the following title, which is composed of a main
heading for the broad aspect and a subheading for the details
(Hodgen 1981).
Antenatal diagnosis and treatment of fetal skeletal
with emphasis on in utero surgery for neural tube defects and
limb bud regeneration
This type of title could be useful for your doctoral dissertation:
the main heading for the nonspecialists, including your relatives,
neighbors, and friends; and the specific subheading for experts in
the field.
Running title
As an aid to readers, most journals print a running title at the top
(running head) or bottom (running foot) of every page or alternate
pages. Thus, if a journal is opened in the middle of an article, the
reader will know what it is about.
As your main title will probably be longer than the stated limit
for the running title, provide (on the title page of the manuscript)
an abbreviated version. When doing so, focus on the keywords,
as did the authors of this title of a paper published in Obstetrics &
Gynecology (López-Jaramillo et al. 1997):
Calcium supplementation and the risk of preeclampsia in
Ecuadorian pregnant teenagers
which was condensed to:
Calcium supplementation reduces preeclampsia
However, as observed earlier, the presentation of results in the
current paper should be in the past tense, thus “reduced”; present
tense, “reduces,” is used for established knowledge.
A course participant, listening to my advice regarding these
principles, asked, “Do you really mean that Tjio and Levan [who
discovered that we have 46 chromosomes, not 48] should have
entitled their paper The chromosome number of man was 46?” Of
course not, that would have misled the reader completely. This is
an illustrative example of Rule No. 1 of writing a scientific paper:
the author’s common sense should always take precedence over the
principles of authorities.
How to Write and Illustrate a Scientific Paper
In the running title discussed here, however, the past tense “reduced”
is acceptable.
Calcium reduced the risk of preeclampsia
Note that I have reinstated “the risk of,” because a condition cannot
be reduced – only its frequency, degree of difficulty, or both can.
Who should be an author? Let us discuss this delicate issue.
Guidelines on authorship
The criteria for authorship as defined by the International Committee
of Medical Journal Editors (1985; 2002) are scarcely known to
many scientists. And even if known, they are often ignored, as they
are considered too restrictive. According to the current criteria, as
of January 27, 2002 (, all persons designated as
authors of a paper should have:
conceived and planned the work that led to the paper or
interpreted the evidence it presents, or both;
(2) written the paper, or reviewed successive versions and taken
part in the revision process; and
approved the final version.
As all three conditions must be met, this implies that a skilled laboratory technician who does the necessary spadework or a clinician
contributing patient material is unworthy of acknowledgment as a
How to Write and Illustrate a Scientific Paper
coauthor. So, then, which guidelines for authorship are applied in
the real world?
In the real world of research, most investigations nowadays are
necessarily made by a team. Each member contributes different
talents and skills. Irrespective of the nature of their contributions
– intellectual (creative) or practical (doing the experiments) – all
members of the team are usually acknowledged in the author
The names also require some sort of ranking. The most prominent
position, heading the list or in last place, is usually occupied by
the team leader (often the author). The coauthors are then listed
in descending order, reflecting how much each person contributed
to the work. My experience is that the team leader is the person
best suited to decide this ranking. He or she will be anxious to
retain the coworkers within the team for forthcoming trials and
has therefore a self-interest in making the ranking fair.
Are the coworkers then happy with the ranking? Usually not,
because many tend to overestimate their own particular contribution. When authors of a multiauthor paper were asked to assess
their own contribution, the total contribution came to 300 percent
(Broad 1981)!
However, the introduction of a credit line stating exactly what
each person did in the study – rather like film credits – will probably
make the ranking of authors less important. Such an approach was
suggested at a conference on authorship in Nottingham, England,
in 1996 (Godlee 1996). Since then the proposal has been adopted
by leading journals such as BMJ, JAMA, and The Lancet. These add
the list at the end of the paper, under the heading “Contributors.”
The contributors list
Here is an example from The Lancet that highlights the contribution
of a graduate student (Tønnes-Pedersen et al. 1997). The title of the
article, with its author byline, reads:
Hormone replacement therapy and risk of non-fatal stroke
Anette Tønnes-Pedersen, Øjvind Lidegaard, Svend Kreiner,
Bent Ottesen
and here is part of the contributors list:
Anette Tønnes Pedersen carried out the study, as part of her
PhD project, and was responsible for all parts of the research
project, including the writing of the paper. Øjvind Lidegaard
was responsible for the initial study design . . . Svend Kreiner
did the statistical analysis . . . Bent Ottesen was responsible for
overall supervision . . .
Thus, Tønnes-Pedersen did all that could be expected of a graduate
student and, in the contributors section, she had the opportunity
to say so.
Readers will also be served by such a list. They will find it easier
to identify which person to contact to discuss some point in the
paper or to request a reagent.
Even if your target journal has not yet adopted the “contributor”
concept, you are encouraged to include a contributors list in your
manuscript. The editor would love it – and it would also give
impetus to the idea.
But don’t be too detailed. In The Lancet (1997, 350:620–3)
I saw the following statement in the contributors list of a report
concerning a survey using questionnaires (family name is fictitious):
. . . Isabel Moe . . . mailed the letters and questionnaries. . . .
A question of coauthorship
Here is a question from a course participant concerning coauthorship:
A multidisciplinary study included, among many other things,
a psychiatric evaluation of groups of patients, according to
a standardized method. The psychiatrist was coauthor of a
number of articles. Should he be listed as a coauthor also in all
future papers based on further analyses of the material – even if
he in no way takes part in the production?
How to Write and Illustrate a Scientific Paper
I do not have the answer. But let us consider the problem from
both sides, the psychiatrist’s and your own.
Without the psychiatrist’s professional evaluation of the patients,
which apparently was not routine work, you would have had no
material to analyze and no results to report. Maybe you should list
him as a coauthor and also ask him to check the manuscripts. He
might be able to contribute valuable ideas on parts of the papers
dealing with his specialty.
However, you honestly feel that the psychiatrist has already
received his due share of the proceeds. You also believe that you
no longer require his expertise. If so, in future papers, you might
thank him in the acknowledgments. This will also mean that the
psychiatrist, too, can write a paper based on this material as seen
from a psychiatrist’s point of view – and in return mention you in
his acknowledgments.
A third way might be that you talk the situation over with the
psychiatrist. But apparently this approach is blocked; otherwise
you would have used it already.
This discussion shows how tricky it can be to decide on
By writing a working abstract at an early stage, you will provide
yourself with a framework for the rest of the article. It will guide
you in deciding what to include and what to omit, thus sparing you
tedious rewriting. Then, when you have completed the paper, you
can return to the working abstract and recast it in its final form.
Let us consider here what to include. In both types of abstract,
the conventional (unstructured) and the formally structured, the
same four basic sections are required: Background (including the
purpose of the study), Methods, Results, and Conclusions.
The conventional abstract
In the conventional abstract, the four basic components are not
identified with explicit headings, which may explain why some
authors miss out vital parts. In the worst instance, the abstract
contains neither background nor conclusion. Many abstracts
written in the conventional manner are uninformative and cause
editors much unnecessary editorial trouble. The following abstract,
quoted from an article in Contraception, commences directly with
the method:
How to Write and Illustrate a Scientific Paper
80 female Wistar rats were employed in this work. They were
homogeneously divided into four groups. . . .
Most journals reporting laboratory studies use the conventional
abstract, and it is usually applied to Case Reports and Brief
Communications. Clinical journals, however, often require a
structured abstract for full-length papers.
The structured abstract
The structured abstract differs from the conventional abstract
by having a heading for each section. Here is such an abstract
of the classic The chromosome number of man by Tjio and Levan
(1956). The paper was published before abstract sections had come
into common use. So, based on the content of that paper, I have
composed a structured abstract in order to show what could be
Background. It is generally accepted that the chromosome
number in humans is 48. But to count chromosomes has been
difficult, as they clump and partially cover each other. In this
study, cultured cells were treated with solutions that spread the
chromosomes and made them easier to count.
Methods. Cultured cells from human embryonic lung were
treated with both colchicine and hypotonic solution.
Results. Among 265 mitoses counted, all but 4 had a
chromosome number of 46.
Conclusion. The results suggest that the chromosome number
in humans is 46, not 48.
In several clinical journals the structured abstract is more detailed,
with the methods section split into, for example, Design, Setting,
Patients, Intervention, and Measurement. The Instructions for
Authors will tell you what headings to use, if your target journal
requires a structured abstract.
The structured abstract has been criticized. It is longer than the
abstract in standard format. Its imposed style may be a strait-jacket
constraining the author, inhibiting creativity. Its rigid uniformity
may bore the reader.
Admittedly, these objections are serious, but the advantages of
the structured abstract outweigh its greater length and the extra
effort needed for its preparation. In fact, the structured abstract
conveys information so accurately and efficiently that readers could
be dissuaded from going on to read the rest of the article. And that
is not the intention.
Finally, if you cannot avoid an abbreviation in the abstract, you
must explain it, the reason being that the abstract will stand alone in
abstracting publications. For the same reason, if you are convinced
that your abstract must include a reference to a significant work,
give a full reference.
How to Write and Illustrate a Scientific Paper
Michael Crichton, the author of Jurassic Park and other bestsellers,
has a background in medicine. He once wrote the following
introduction for a paper published in the New England Journal of
Medicine (1975):
Most medical communications are difficult to read. To determine
why, contributions to three issues of the New England Journal of
Medicine were studied and the prose analyzed.
Crichton’s first sentence awakens interest. It is succinctly written in
only seven words. The entire section is short – a mere three lines
long, and not a word needs to be added.
Here is another fine introduction, to a paper published in the
BMJ (McGarry 1994):
Nose bleeds in adults are the commonest reason for emergency
admission to an otolaryngology ward, but the cause of
the condition remains unknown.1 Case reports suggest an
association between nose bleeds and regular, high alcohol
We conducted a prospective case-control study to compare the
alcohol habits of adults with nose bleeds with those of controls
being treated for other otorhinolaryngological conditions.
These introductions, like many other well-written introductions,
contain a brief description of two items:
the problem;
(2) the proposed solution.
However, the first few sentences sometimes contain general, even
vapid, statements, as in this example from a manuscript under
preparation submitted by a course participant of mine:
Respiratory diseases are important health problems throughout
the world and often lead to morbidity and death.
These platitudes could be omitted ruthlessly, as could also the
empty words of the next sentence. Only in the third sentence does
the author come to the point:
An important risk factor for developing Chronic Obstructive
Pulmonary Disease (COPD) is chronic cigarette smoking (1).
I advised the author to open the introduction with the third
sentence. The reference (1) should be to a carefully chosen review
article describing the problem.
You probably need more than two or three sentences for your
introduction, but it should preferably not exceed one page in length
(typed double-spaced). More space may be required for certain
topics, such as occupational science, medical ethics, and nursing
and health care. Check the current version of the Instruction for
If you have previously published part of the work other than in
the form of a congress abstract, you should say so in a few words at
the end of the introduction.
How to Write and Illustrate a Scientific Paper
Methods are usually best described in the order in which they
were used. So, in the following sentence, taken from a manuscript
in preparation, it would be helpful if you reversed the order of
presentation, as from:
Cell growth was stopped with colchicine after incubation for
65–70 hours at 37 °C.
After the incubation of cells for 65–70 hours at 37 °C, their
growth was stopped with colchicine.
Unless a previously published method is generally known, the reader
will appreciate being told its essential features. Thus, a reference
figure may well be considered inadequate, as in this example from
a manuscript in preparation:
Kidney volume was measured as previously described.3
Little need be added to give the reader the broad outline of the
method (revised text in boldfaced italics):
The kidney volume was measured with an ultrasound apparatus
containing a built-in volume program.3
A new procedure, however, should be described in sufficient detail
to allow a trained scientist to repeat the investigation. For the
novice writer it can be hard to find a middle course between too
much and too little information; often the novice errs on the side
of too much. An experienced colleague can help to remove excess
The subjects
I once saw a note taped to a door asking for volunteers – completely
healthy, nonsmoking women aged 55 to 60 – to participate in a
study of the effect of estrogen on blood circulation in the legs. The
door led from a huge bicycle garage (reserved for workers at the
hospital) into the elevator hall. In a subsequent report of this study
each of these details has to be related, because bicycling women
of this age might not be representative of the reader’s patients
– especially not concerning the main topic of the study, blood
circulation in the legs.
In fact, most doctors reading your paper will ask themselves,
“Does this apply to my patients?” To assess this, they need to
know precisely the source of the participants and the details of the
entry and exclusion criteria. If you are uncertain how to present
it, I recommend as a model the section “Subject Recruitment and
Enrollment Criteria” in a paper on treatment of the common cold
in children, published in JAMA (Macknin et al. 1998).
Informed consent
Before ethics committees became common, lecturers teaching
research methods, including the matter of obtaining informed
consent, could get the following comment from course participants,
“But if we ask the subjects they might say no” (Holmes 1997).
That’s right. When patients have a free choice, few studies have
a consent rate of 100 percent. So, if you are writing the often seen
How to Write and Illustrate a Scientific Paper
construction “All patients gave informed consent,” give it a second
look. Probably you are referring only to those patients who had
already been enrolled in the study. If that is the case, say so.
If, on the other hand, you are referring to all those patients who
fulfilled the criteria for being enrolled (called eligible patients), then
the reader would be interested to know whether the patients really
were informed of possible side effects of the trial, and if they were
given a copy of the written informed consent form.
Normally, the following could suffice, although it is a minimum
(Jha et al. 1998):
Patients were informed of the purpose of the trial and had to
give their signed informed consent before being enrolled.
Then state in the results section how many patients declined to
participate. If you know their reasons for doing so, these should be
reported too. But remember, patients are free to say “no” without
giving a reason.
Let us now take a closer look at one particular type of study, the
randomized controlled trial. When properly conducted, it is the
most reliable way of comparing treatments. However, reports of
such trials frequently omit important features of the study design.
We shall consider some of them.
Omissions in reporting of randomized
controlled trials
Randomization means that subjects are allocated to the treatment
and control groups by chance (at random). But it is inadequate
just to say that a study was randomized, without telling how. The
reason is that all methods of randomization are susceptible to
conscious or unconscious manipulation by investigators, the degree
of which varies from one method to the other. The two methods
considered most reliable are the following: a telephone call to an
independent center for a computerized “flip of a coin”; and the
use of carefully prepared envelopes (sealed, opaque, identical, and
serially numbered). Procedures easier to manipulate are alternate
cases, odd and even birth dates, or file numbers. They should
be avoided. So, to evaluate the trial properly, the reader needs to
know how the assignment was made, as in this example (European
Carotid Surgery Trialists’ Collaboration Group 1998):
We randomised . . . by telephone to the Clinical Service Unit
in Oxford. A computer program generated the randomisation
schedule . . . making it impossible for the local investigators to
know whether the next allocation was going to be to surgery or
Specify the blinding method used. For example, when three parties
are involved – the patient, the treating physician, and the evaluator
– some designate the study triple-blind, others double-blind. And
what is implied in the term double-blind in the paper entitled
“Double-blind study . . . on . . . rats,” published in Acta Chirurgica
Belgica in 1978?
Moreover, how should we denote blindness in a study in which
only the evaluators are masked, as was the case in the famous
trial of streptomycin in the treatment of pulmonary tuberculosis
(Medical Research Council 1948). The term blind was not even
mentioned in the paper, which merely described what was done:
The [X-ray] films have been viewed by two radiologists and a clinician, each reading the films independently and not knowing if
the films were of C [control] or S [streptomycin] cases. There was
a fair agreement among the three; at a final session they met to
review and discuss films on which there had been difference of
interpretation, and agreement was reached without difficulty on
all films.
This is a careful reporting of how blindness was accomplished.
How to Write and Illustrate a Scientific Paper
Number of subjects
Before you start the investigation, calculate the sample size needed
to demonstrate a difference, if it exists. This calculation should be
reported. Here is an example from a study on the healing of leg
ulcers, published in the BMJ (Morell et al. 1998):
Sample size
To have an 80% chance of detecting as significant (at the 5%
level) an increase in healing from 50% to 70%, 206 patients
were required.
However, the number of patients needed in a study refers to the
number of patients who complete the trial, rather than the number
who start it. To add a calculated drop-out rate, say 10%, would
therefore be advisable:
To allow for a 10% drop-out rate, 230 patients were enrolled.
Sample-size calculation is so important a part of the study design
that it deserves a separate subheading.
Let us turn briefly to animal studies. An example of a fine study
design for a trial on dogs with perianal fistulas was reported by
Mathews and Sukhiani (1997) in Journal of the American Veterinary
Medical Association. The number of dogs needed was calculated,
and the random allocation of the dogs to the treatment and control
groups was described. In most studies on animals, such information
is lacking.
A checklist
In 1994, two expert groups independently published detailed
checklists for the reporting of randomized controlled trials. Two
years later these groups produced a unified list, the CONSORT
(Consolidated Standards of Reporting Trials) statement. It states in
detail which items must be included in a report (Begg et al. 1996;
Moher et al. 2001; CONSORT Website 2001).
As the reporting of the randomized controlled trial covers most
aspects of the principles of scientific writing, I have used it as a
model for this chapter. This approach spares us tedious repetition.
The chapter therefore has two parts: the flow of participants and
their follow-up; and the outcome of the study.
Participant flow and follow-up
Figure 14.1 Trial profile. (Reproduced, with permission, from K. Sliwa, D. Skudicky,
G. Candy, T. Wisenbaugh, P. Sareli, Randomised investigation of effects of
pentoxifylline on left-ventricular performance in idiopathic dilated cardiomyopathy, The Lancet 1988; 351(9109):1091–3, © The Lancet Ltd.)
How to Write and Illustrate a Scientific Paper
This seemingly fine flow chart (14.1) would have been even more
informative if it had included two additional boxes above the first
one, giving the numbers of patients screened and excluded. Here is
part of a flow chart with such information.
Figure 14.2 Part of a flow chart showing that no fewer than 93% of the patients
screened were excluded. (Reproduced, with permission, from The Publications
Committee for the Trial of ORG 10172, in Acute Stroke Treatment [TOAST]
Investigators. JAMA, April 22/29, 279:1265–72. Copyright 1998, American Medical
Only 1904 of the 25 624 potential subjects were selected. This
information is useful to have in case you intend to make a confirmatory study. In the main text of this paper, the reasons for exclusion
were given in detail. Such data can help readers to assess potential
bias in patient selection.
A high rate of withdrawal (say 15 percent or more) can invalidate
the conclusions of the study (Lang and Secic 1977, 24). The number
of dropouts and their reasons for withdrawal should be reported for
each group separately.
Moreover, dropouts should be included in the analysis of the
study on an “intention-to-treat” basis. This is often neglected and
can result in incorrect reporting. If you simply ignore the drop-outs
and if, for example, the reason for their exclusion is chiefly the side
effects of the new treatment tested, then the ensuing comparison
would be biased in favor of this treatment. Hence, all randomized
participants should be analyzed in the groups they were originally
allocated to, including even those patients who did not receive
the intended treatment and even those who, for some reason,
subsequently received the treatment of the alternative group. The
use of intention-to-treat analysis should be indicated in the flow
chart, for example in either of the following ways.
Figure 14.3 Alternative displays of the right arm of Figure 14.1 to indicate analysis
on intention-to-treat basis
Loss of participants to follow-up
Participants who completed the treatment but who were lost to
follow-up are likely to be atypical in critical ways. They could
be patients who did not return because they had recovered, or
because they had died, or who were still sick and did not wish to
burden the doctor with an unsatisfactory outcome. Describe their
characteristics as at the last examination.
Details of randomized participants
In a manuscript under preparation, I saw a detailed description of
the study group, but the controls were presented as follows:
A control group of sixteen healthy volunteers underwent investigation in the same manner.
How to Write and Illustrate a Scientific Paper
Such scanty presentation of the controls has been the reason for
many a rejected paper. The controls should be described just as
thoroughly as the subjects of the treated or exposed group, as in
the following table from a study on the duration of pregnancy after
laser conization of the cervix.
Table 14.1 Clinical characteristics
(n = 64)
(n = 64)
Age at delivery (y)
31.3 ± 4.19
31.4 ± 4.15
35 (54.7%)
33 (51.6%)
Previous preterm delivery
2 (3.1%)
3 (4.7%)
Previous miscarriages
7 (10.9%)
8 (12.5%)
Previous voluntary abortions
4 (6.3%)
2 (3.1%)
2 (3.1%)
1 (1.6%)
13 (20.3%)
12 (18.8%)
9 (14.1%)
16 (25.0%)
Operative vaginal delivery
Cesarean delivery
NS = not significant
Data presented as mean ± standard deviation or n (%).
Source: Reprinted from Raio et al., with permission from the American
College of Obstetricians and Gynecologists (Obstetrics & Gynecology
1997; 90:978–82).
Be cautious, however, when comparing baseline characteristics
with the use of statistical tests. A minor imbalance in a key prognostic factor can have a profound effect on a treatment comparison,
even when the imbalance shows “no significance.” Significance
testing can thus obscure an important imbalance. Therefore, in
addition to the statistical testing (some experts say “instead of
statistical testing”), compare the baseline characteristics with the
use of clinical experience – and common sense (Altman 1985;
Bailar 1986). Then, state in the main text what you have found.
The P values could thus have been omitted from this table and
replaced in the main text with, for example:
There were no clinically meaningful differences between the
groups in their baseline characteristics.
Study outcome
Try not to repeat tediously in the text what is already clear from
a perusal of the tables. The text should emphasize the important
observations and present them in order of decreasing interest,
beginning with the main finding. Tables are used for reporting the
details of the outcome. The following text repeats the content of
the table:
As Table 1 shows, the mean ± SD of nocturnal plasma-melatonin
concentrations was 19.0 pg/mL ± 11.9 in the 6 patients in the
suicidal group and 45.5 pg/mL ± 27.1 in the 22 controls (P <
There is no need for such repetition; merely state the main point:
Patients with a history of attempted suicide had significantly
lower nocturnal plasma-melatonin concentrations than did
controls (P < 0.05) (Table 1).
An exact P value would have made the presentation even more
informative. Below is part of an excellent table that presents both
confidence intervals and exact P values, thus enabling readers to
judge for themselves the clinical importance of the results.
Table 14.2 Outcome of treatment
Number (%) of patients
with outcome
n = 60
Difference in
percentage (95% CI)
n = 60
Clinical success
48 (80%)
39 (65%)
15.0 ( – 0.7 to 30.8)
60 (100%)
54 (90%)
10.0 (2.4 to 17.6)
Source: Reproduced, with permission, from M.A. Salam, U. Dhar, W.A. Khan, M.L. Bennish,
Randomised comparison of ciprofloxacin suspension and pivmecillinam for childhood
shigellosis. The Lancet 1998; 352(9127):522–7, © The Lancet Ltd.
How to Write and Illustrate a Scientific Paper
Be careful also in the reporting of side effects. It could be the most
important part of the study. (Does the drug cause so much nausea
that patients will not take it?) Therefore, describe adverse effects as
thoroughly as beneficial ones. If no adverse reactions were found,
say so.
Toward the end of World War II, General Patton advanced with
his tanks through the enemy lines. From motorcycle dispatch riders
he received intelligence reports from other parts of the front. When
a rider began reading out the dispatch from the very beginning,
Patton most often asked him to go directly to the bottom line.
In the same way, hurried readers of scientific papers thumb
through the pages to find the final paragraph of the discussion.
Why? Apparently, because that is where the reader can usually get
a comprehensive conclusion of the results.
But there is no generally accepted form on how to arrange the
various parts leading up to the conclusion. So to help you I have
chosen, as a model, an especially well-designed discussion in a
paper by Logan et al. (1993). Based on this I have composed a
structure for the discussion section that can be used as a guide
when writing this section. It has four parts: Main message, Critical
assessment, Comparison with other studies, and Conclusions. Let us
discuss them step by step.
(1) Main message, which answers the question posed in the
introduction section and includes the main supporting
How to Write and Illustrate a Scientific Paper
Most often, however, the opening paragraph of a discussion
unnecessarily recapitulates in detail what the readers have already
been told twice, once in the abstract and again in the results section,
as in the following example. (Imagine General Patton listening to
Results of the first phase of this study show that men assigned
to Hospital Corpsman and Mess Management Specialist
occupations have the highest overall hospitalization rates
across the three decades of a 30-year navy career. Rates also are
elevated for the groups of Construction/Manufacturing, Deck,
Ordnance, and Engineering/Hull while the lowest rates across
the three decades are observed for the group of Miscellaneous/
Technical, Electronics, and Administrative/Clerical.
Now, compare this with the next example in which Logan et al.
open the discussion with an answer to the question posed in the
Our data support the hypothesis that taking aspirin or other
non-steroidal anti-inflammatory drugs protects against the
development of colorectal cancer and suggest that it does so by
reducing the prevalence of colorectal adenomas.
Having seen this, readers may well wonder how valid the paper’s
arguments are. So, here is the right place to present the strengths
and weaknesses of the study.
(2) Critical assessment, that is, opinions on any shortcomings
in study design, limitations in methods, flaws in analysis, or
validity of assumptions.
Logan et al. gave primacy to this important part of the paper by
giving it a separate heading:
Bias and confounding
Could bias or confounding account for these findings? . . .
Whether convinced by the arguments or still skeptical, readers
will then want to know how the findings agree (or contrast) with
previously published work.
(3) Comparison with other studies, where inconsistencies are
If you intend to discuss several observations, start with the most
significant, continue with the next most important, and so on.
Thus, beginning with your most significant finding, you start the
comparison with studies whose results are largely consistent with
your own. Then consider studies less compatible with yours, and so
on. Conclude with any results that contradict your findings.
All the time, discuss similarities and differences. If you cannot
explain conflicting evidence, you could suggest how the discrepancy
might be resolved by conducting a new trial.
Having completed such a comparison, Logan et al. rounded off
the discussion in a paragraph headed “Conclusions.”
(4) Conclusions, that is, comments on possible biological or
clinical implications and suggestions for further research.
. . . Studies are now needed to confirm these findings, to determine how non-steroidal anti-inflammatory drugs might act, and
particularly to see if [these] drugs can prevent the recurrence of
adenoma or even cause sporadic adenomas to regress.
In the following I will set out further suggestions for writing the
discussion section.
How to Write and Illustrate a Scientific Paper
Evaluate the results – not the authors
Consider this:
A simple but very keen-sighted observation was made in the
Gothenburg study, namely, that there was a relationship between
the waist to hip circumference ratio and myocardial infarction.
[My translation from the Swedish.]
In my view, the only thing these investigators did was to use a
measuring-tape. So, in what way was this keen-sighted? Reading
the measuring-tape? And why “simple but” instead of “simple and,”
as if “simple” were something negative? I would have deleted the
first ten words: “The Gothenburg study showed that . . . ”
Avoid claiming priority
In a discussion section, I once saw this:
Our study appears to be the first one in which an open-end
catheter method was applied to the study of tubal motility in
the primate.
Some time later, in Letters to the Editor, a reader told us that
several similar works on humans had been published and called
the claim of priority “intellectual piracy.” As the study was made
on apes (primates include both humans and apes), the author of the
paper replied:
The quotation about being the first study in the primate is
obviously wrong. We . . . should of course have said nonhuman
Following this line of argument, most studies could be designated
“the first,” because most of them have a design of their own. Try
instead to establish the novelty of your work by telling the reader
in the introduction how the design of your study differs from that
of previous works, as in this fictitious example:
Most studies have been made on humans; ours was made on
Then tell the reader why this approach ought to be superior.
The reference-13 trick
At this point, let us assume that you have completed your study
after three years of hard work. When you now scan the latest issue
of your target journal, you come across a paper on a subject very
similar to yours; same questions, same answers.
You are horrified. Your first intuition is to ignore the article,
“After all, editors and referees are too busy to have time to read
journals.” At that very moment, Reference-13 Himself approaches
you and whispers in your ear: “Bury it in the discussion!” –
meaning that if you refer to the paper in a short subordinate
clause in the depths of the discussion section, it will hopefully pass
unnoticed (BMJ editorial 1985). But the editor and referee will find
it, for sure, because they know the trick.
Your citation should instead be inserted in the introduction
section, as Reference 1 or 2. That doesn’t mean that your cause is
lost; your investigation could be an important confirmatory study.
Here is an example of a generous way of handling this
problem. Two researchers, Karman and Potts (1972), had developed a surgical method. When they subsequently searched the
literature, they found that a similar technique had been described
35 years earlier in a Ukrainian journal in the Russian language
with no English summary. They honestly gave full credit to the
initial inventor:
Since the development of this apparatus it has come to our
attention that Bykov developed an analogous procedure in
How to Write and Illustrate a Scientific Paper
You should not forget to thank the people who have really helped
you, but whose contributions do not justify authorship. But be
specific. When I saw this acknowledgment, I wondered what
exactly these persons had done (family names are fictitious):
We thank C. Roe, D. Doe, and S. Poe.
In the following example it is evident what each person did:
We thank Betsy Roe and Gerri Doe for their assistance in
preparing the data; William Poe for the medical photography;
Marian Loe and David Coe for their critique of the findings
of this study; and Fred Noe for reviewing the 200-+m and
400-μm specimens.
But take care not to give others credit for your own work. Otherwise
the reader will wonder what your contribution was.
The often used “wish to thank” can be shortened to “thank.”
Avoid using professional or courtesy titles in the acknowledgment.
The example above was written accordingly. It also tells us the given
name of those acknowledged.
Persons you want to thank should be asked if they are willing
to be acknowledged and if they approve the wording you have
used to mention them. This is because colleagues who have read
and corrected your manuscript may disagree with some of its
central points. To acknowledge them could imply their approval of
the content of the paper.
Always give credit for financial support
When you thank sources of financial assistance, be careful how
you present their names. If, as in this case, the name of the funding
agency is not in the language of the journal:
The study was supported by “Kronprinsessan Margaretas Arbetsnämnd för synskadade.”
you should use either a translation alone:
The study was supported by Crown Princess Margareta’s
Working Group for the Visually Handicapped.
or, as a courtesy to the granting authority, both the original name
and, in brackets, the translated name:
The study was supported by “Kronprinsessan Margaretas Arbetsnämnd för synskadade” [Crown Princess Margareta’s Working
Group for the Visually Handicapped].
But do not thank all grant-giving agencies who have supported
your research work over several years past. Thank only those which
supported the study you are now reporting because, just as research
teams compete with one another, so too do granting authorities.
And if one authority alone has chosen and supported a promising
work, you should give that authority all due credit.
Note that some journals ask that funding organizations be
named on the title page instead of being included in the acknowledgments.
How to Write and Illustrate a Scientific Paper
At one time, there were over 250 different styles of reference in
the scientific literature (Garfield 1986). The editors of some
major biomedical journals therefore had good reason to convene
in Vancouver, Canada, in January 1978, to work out a uniform
reference style. One of their suggestions was that authors should
number references in the order in which they appear in the text
(International Committee of Medical Journal Editors 1997).
Vancouver versus Harvard style?
Although many of the major journals in the biomedical field have
adopted the Vancouver style, some still prefer the Harvard system
(first used in 1881 by a zoologist at Harvard University [Chernin
1988]) in which the author’s name and the year of publication are
cited in the text. In the fictive sentence below, I have mixed the two
styles to illustrate their differences:
A reference figure (17) in the Vancouver style says less than
a name-and-year reference (Einstein 1941) according to the
Harvard system.
Most readers prefer the Harvard system because they like to know
just what author is being cited as they read the text. Still, the
name-and-year system does have disadvantages: difficulty for readers
who see an interesting item in the reference list in locating that
reference in the main text; and, more important, the disruption of
the text when a large number of references need to be cited within
a paragraph, as in this example (Bengtsson 1968):
This method was introduced by Aburel in 1938, but he was
followed by only a few workers in the succeeding 20 years
(Bommelaer 1948; Cioc 1948; Kosowski 1949; de Watteville
and d’Enst 1950). During the 1960’s however, hypertonic saline
has been increasingly employed (Bengtsson and Csapo 1962;
Jaffin et al. 1962; Wagner et al. 1962; Larsson-Cohn 1964;
Møller et al. 1964; Sciarra et al. 1964; Wiqvist and Eriksson
1964; Bora 1965; Short et al. 1965; Turnbull and Andersson
1965; Wagatsuma 1965; Cameron and Dayan 1966; Gochberg
and Reid 1966; Klopper et al. 1966; Christie et al. 1966; Ruttner
1966; Olsen et al. 1967).
Using the Vancouver system, the text above can be condensed to
about one-third of its original length:
This method was introduced by Aburel in 1938,1 but he was
followed by only a few workers in the succeeding 20 years.2–5
During the 1960s, however, hypertonic saline has been
increasingly employed.6–22
As shown, an important name and year can be featured within
the text even when using the numbering system. Nevertheless, you
have to follow the style of the journal to which the paper is to
be submitted. So read the current version of its Instructions for
Now at last you are ready to submit your paper. At this moment
you notice that you have used an inappropriate citation system.
Today, this is not a catastrophe. By using a suitable computer
program you can, at the touch of a few keys, produce reference lists
in the format of your choice and, in the text, substitute names for
numbers (or vice versa).
How to Write and Illustrate a Scientific Paper
In two pages, I would like to show in detail how references are written
in the main text and in the reference list according to the Vancouver
and the Harvard systems. I also take the opportunity to show how
to refer to unpublished results and personal communications.
The Vancouver system – also called the numbering system
The Harvard system – also called the name-and-year system
How to Write and Illustrate a Scientific Paper
Accuracy of references and quotations
When I was a novice editor, I endeavored to evaluate personally
the content of each submitted paper (in addition to the assessments
made subsequently by referees). If I didn’t know the subject well
enough, I marked the key citations in the reference list when
reading the text and then tried to retrieve them.
I was surprised to find how often I was unable to track down an
article. In some submitted papers as many as every second reference
was untraceable by direct search. Through the journal’s volume
index, however, I found some of them, but under the wrong year,
wrong volume, or wrong page. Other errors found in retrieved
publications included incorrect title, wrong author(s), misspelling
of author’s name or, more seriously, misquoting of other authors’
What can an author do to make citations more accurate? Most
misquotations can be avoided by rereading the publications cited.
Thus, never rely on memory alone! The number of errors can be
reduced by checking and rechecking not only new references but
also those lifted from your own computer bank of citations. Even
references downloaded from MEDLINE can conceal errors! This
means that you should always have a copy of the publications to be
cited – at hand. (Some journals require that authors send, with the
submitted paper, a photocopy of the first page of every reference
Quoting from another article is allowed only if the original
publication is unavailable. However, citing a publication you have
not seen can be hazardous, as the following illustrates. For nearly
50 years, several authors of English-language papers referred to a
Dr. O. Uplavici (Dobbel 1938). He was assumed to be the author
of a Czech-language article, reporting the first experiments in
which amebic dysentery was transmitted from man to cat. Actually,
“O úplavici” was the title of the paper and means “On dysentery.”
The first author who cited the paper, Kartulis (1887), made the
mistake. Hence, if you refer to a paper you have not read, say so:
Amebic dysentery was successfully transmitted from man to cat
(Uplavici 1887, cited by Kartulis 1887).
Kartulis is thus held responsible for what he says Uplavici said. (The
original article of “O úplavici” was written by Jaroslav Hlava 1887.)
Where you have a choice, avoid quoting sources that are not
widely accessible; many books and most conference abstracts of
meetings belong to this group. The citation of abstracts is particularly to be discouraged, as only about one-third of them lead
to full-length, peer-reviewed articles (Liu 1996). Some journals no
longer allow the citation of abstracts.
Your own author name
Decide early in your career the form in which you wish your name
to appear. If you have a common name, it may be wise to use your
middle initials in order to be distinguished from other scientists. If
you are uncertain, make a search on MEDLINE to see whether you
should use your initials.
If you change your surname and start using the new name as
your author name, your previous scientific work will become hard
to trace. So, stick to the name under which you wrote or coauthored
your first papers (van Loon 1997).
Problems of Western names are nothing compared with those in
some Asian countries. Here I present only one example, the names
of Chinese. (More details on this subject are given in Scientific Style
and Format 1994, 136–7.)
Chinese names
It is especially troublesome to distinguish the names of Chinese,
because their family names are shared by so many, the 11 most
common accounting for about 40 percent of the entire population
How to Write and Illustrate a Scientific Paper
(Xu ZhaoRan and Nicolson 1992). This means that, on average,
each of these names is shared by 42 million people! So, to distinguish
their names, you have to spell out in full the given names in your
reference list. (Given names of Chinese citizens are always placed
last, thus after the family name.) So:
Deng X.
Jiang Z.
would have been easier to identify as:
Deng Xiaoping
Jiang Zemin
To maximize clarity, two writers (SUN Xiao-Ling and ZHOU
Jing 2002) suggested that Chinese authors write their surnames in
all capital letters and hyphenate two-syllable given names. Many
Chinese in Hong Kong and in the USA adopt hybrid names, using
a Western given name, such as Tony K.H. Chung and Felix W. S.
Wong. Such names are referred to in the Western manner: Chung,
T. K. H., Wong, F. W. S.
You may have to refer to an unknown author in at least two kinds
of situation: one involves articles published during the cultural
revolution in China and some years afterwards as no author was
allowed to appear in an author byline; the other concerns unsigned
editorials. Although anonymous is often the term used in such
cases, consider how much more you can convey if you use the
method described below.
An example from China is a much-cited pioneering article
reporting the first clinical use of placental biopsy. As Chinese
articles in medicine without an author give the name of the author’s
hospital, this could serve as the reference in the text, for example
(Harvard style):
(Tietung Hospital 1975)
In Vancouver style:
In 1975, Tietung Hospital (7) reported…
In your reference list it would appear as (Harvard style):
Tietung Hospital. 1975. Fetal sex prediction by sex chromatin
of chorionic villi cells during early pregnancy. Chin Med J
An unsigned editorial can be referred to in the text as in the
following example:
(Nature editorial 2006)
and in the list of references:
Nature editorial. 2006. Save the lungfish. An Australian dam
project threatens a living fossil. 442:224.
Editorials or other articles by unknown authors can also be referred
to by citing in the text the first few words of the title:
(Save the lungfish 2006)
and in the list of references:
Save the lungfish. An Australian dam project threatens a living
fossil [editorial]. 2006. Nature 442:224.
In general, citing an author as Anonymous is to be avoided in
scientific writing.
Record titles in the language of publication
I once saw a manuscript in which the first seven references listed
were written in five different languages: Bulgarian, Hungarian,
Italian, German, and English. Readers are best served if only one
language is used, that is, the language of the journal in question,
How to Write and Illustrate a Scientific Paper
which its readers can be assumed to understand. Let us look at one
of the references mentioned, that in Bulgarian:
Tanchev S, Asparuhov A, Tanchev P, Gramcheva O. Vurkhu
vuzstanoviavaneto sled rodova fractura na kliuchitsata. Akush
Ginecol (Sofia) 1987; 26:49–XX.
As the language of the journal is English, the Bulgarian title needs
to be translated into English. The translated title is then placed
in brackets and additional information in parentheses. (Like
the original reference, the following one is written in Vancouver
Tančev S, Asparuhov A, Tančev P, Gramčeva O. [Healing of
fractured clavicle in newborns] (In Bulgarian with English
abstract) Akush Ginecol (Sofia) 1987;26:49–XX.
Note that I have placed a diacritical mark over the letter c (replacing
ch) when rewriting the names of the authors, which is the way they
themselves have written their names in the English abstract. To
do so is a courtesy. If your computer cannot cope with this mark,
insert it by hand, indicate the change in the margin, and inform
the editor about it in the covering letter.
You can also provide both a non-English-language title and the
Svedin G. Transkutan nervstimulering som smärtlindring vid
förlossning. [Transcutaneous electrical nerve stimulation for
analgesia in childbirth.] (In Swedish with English abstract.)
Läkartidningen 1979;76:1946–8.
Names of journals
Abbreviate journals according to the listing in Index Medicus
( If a journal is not listed (about three in four
biomedical journals are not), spell out the journal’s title in full.
How to refer to World Wide Web
Because a website may be updated after you have seen it, even
disappear, you should give the date you accessed the site and
also keep a printed copy of it. No firm rules exist yet for citing
electronic materials. Consult the Instructions to Authors of your
target journal. In this book I have used the following format in
most cases:
Animal Info. 2002. Information on rare, threatened and
endangered mammals. Severna Park (MD): Animal info. (accessed 2 February 2002).
How to Write and Illustrate a Scientific Paper
Ph.D. and
other doctoral theses
The Ph.D. (Philosophiae Doctor) is the highest university degree.
It is acquired after writing a doctoral thesis (or dissertation) and
defending it at an oral examination.
There are almost no generally accepted rules for thesis preparation. The structure varies from country to country, “from institution
to institution and even from professor to professor in the same
department of the same institution” (Day and Gastel 2006).
However, theses are of two kinds: monographs and compilations
of articles. Monographs are the most common form, especially in
the humanities, theology, and law. But compilations are increasingly
used in medicine, technology, and the natural sciences.
Compilations: the theses of the future
Compilations are based on articles that have “been scrutinized by
international peer review, probably more prestigious than local
committees” (Carling 2006). They are recognized in Argentina,
Australia, India, Japan, the United States, and in such European countries as the Czech Republic, Finland, Germany, the
Netherlands, Norway, Spain, and Sweden; they are permitted
Ph.D. and other doctoral theses
in the United Kingdom, but are not common there (BurroughBoenisch 2006).
Compilations are of two types. In one, the reprinted articles are
sandwiched between introductory and concluding chapters. In the
other, the reprinted articles are appended to a summary of their
contents. In both types, the articles are published or publishable in
refereed journals and often have several authors, with the doctoral
student as first author of most of them. The subject of this chapter
is the type of compilation that includes a summary; for lack of an
official name, I propose to call that summary a thesis overview.
Many of the details described in other parts of this book apply
to the writing of a thesis in general. So, to avoid tedious repetition,
I will discuss only new features here. The advice is applicable also
to other doctoral and postgraduate degrees.
Most research is nowadays performed in groups. You must do your
utmost to indicate clearly what parts of the work were yours. The
examiner will be curious to know how much you contributed to the
study design, data collection, data analysis, and, especially, writing
of the manuscript. Table 18.1 is an exemplary presentation of a
list of contributors to the papers of a thesis, shown as a facsimile
(Theander 2005). See the next page.
How to Write and Illustrate a Scientific Paper
Table 18.1 Detailed list of contributors to the papers of a doctoral thesis (facsimile
from Elke Theander 2005, with permission)
Ph.D. and other doctoral theses
Thesis at a glance
Thesis at a glance could be described as abstracts of the abstracts and
is extremely helpful to the reader. Here is an example of one part
of it (Theander 2005):
Paper II – Does treatment with gammalinolenic acid (GLA)
alleviate fatigue and glandular dysfunction in primary
Sjögren’s syndrome?
Patients: 90 patients (+30 in pilot trial).
Methods: Double-blind placebo-controlled randomized trial.
Conclusions: GLA [had] no effect on fatigue or glandular signs
and symptoms.
General introduction
You may well have come to know more than anyone else on the
planet about your narrow subject, so your general introduction
could and should be a highly readable piece of work. You may, if
relevant, go back millions of years in the opening phrases, as in
this introduction to a thesis on the use of ultrasound in medical
diagnostics (Andolf 1989):
For millions of years, bats and dolphins have used ultrasound
as a method for localization. It was not until 1912, when the
Titanic catastrophe occurred, that scientists proposed that
man as well should use ultrasound . . . In Lund in 1953, the
cardiologist Inge Edler [and] the physicist Helmut Herz [made
observations that led to] the application of ultrasound in the
medical field.
The thesis overview should be intelligible even to a non-specialist.
Baranto (2005), for example, devoted 18 pages to introducing
the reader to the anatomy of the spine, its function and possible
degenerative changes. This was illustrated by drawings borrowed
from a textbook. Such an approach is highly recommended.
How to Write and Illustrate a Scientific Paper
One short sentence for each aim is often sufficient. Do not use
abbreviations here; if you do, explain them. A non-specialist reader
would probably find this aim (quoted from a thesis) unclear:
• to elucidate the [presence of] urogenital carriage of GBS
in mothers of GBS infected infants and/or in GBS colonized
women giving birth to neonatally healthy infants, with respect
to . . .
. . . but not this (Christensen 1980):
• to elucidate the presence of group B streptococci (GBS) in the
urogenital tract among adults;
Methods and results
I had just completed the papers for my thesis, in May 1973. Now,
it was time for the overview. A senior colleague of mine suggested
that I should take the papers, a pair of scissors, and a roll of tape in
one hand and a bottle of brandy in the other, find a quiet place and
cut-and-paste the overview. (We had no word processors in those
days!) His advice had a point; the methods and the results sections
could have been presented in that way. But if you take parts from
your papers you have to paraphrase them, that is, present them
in a new way. If you do not, you need quotation marks around
the borrowed lines. Paraphrasing is difficult, especially concerning
methodology, so I suggest the following instead.
As you have already described in detail the methods and the
results in your papers, the specialist readers will, in the printed
thesis, have access to them there. Instead, in the overview, present
these sections on a non-specialist level. Choose words similar to
those you would use in discussing these sections over lunch with a
colleague from a speciality other than your own.
Ph.D. and other doctoral theses
If you have one common group of subjects or samples, use a flow
chart to show how they were allocated to the different studies. If
you have two common bases, create two flow charts. Circles can
be used in a similar manner; the size of the circles corresponds to
the number of subjects or samples. Overlapping circles can be used
to indicate that some studies had a part in common. Use a table
to present studies based on different subjects or samples. Design
simple drawings to explain your methods. For the results, it may
be useful to construct tables synthesizing data from more than one
Color photographs or diagrams are said to enhance an article,
but they must add something, not only be decorations. If you have
color photos or figures in your journal papers, the less expensive
black-and-white photos will suffice in the thesis overview – if you
find it necessary to reproduce them there at all.
General discussion
The general discussion in a thesis requires a slightly different
approach from the discussion section of papers, in that you must
consider both the whole picture and the individual pieces. Use
a new subheading for each individual piece. Open the general
discussion by explaining how you achieved your aims. In one thesis
(Bergström 1994), one of the aims was:
• To develop a technique for subretinal transplantation of
retinal cells . . .
and the general discussion began:
The transplantation procedure
The subretinal transplantation technique that we have developed
has turned out to be easy to use and [gave] good and consistent
How to Write and Illustrate a Scientific Paper
What makes the writing of this sentence especially good is that the
author used almost exactly the same wording as in the sentence
describing the aim.
If you suffer from writer’s block at the end of your writing, these
acknowledgments could be a way out (family names fictitious;
Azem 2005):
Thanks . . .
My supervisors Ann-Mari Soe and Samuel Loe for supporting
and inspiring me, and sharing their great knowledge in mucosal
immunology. Present and former colleagues, collaborators,
students, professors, and administrators & technical staff for
[contributions to] this thesis.
This study was supported by grants from . . .
This is better than skipping the acknowledgments; only once have
I encountered a thesis without an author’s thanks.
Most acknowledgments, however, are written in the traditional
manner where you thank people individually. But, instead of
merely thanking a coworker “for helping me with a lot of work,”
tell the reader precisely what she did, “Ewa . . . for showing me
how to perfuse rat livers.” Though most authors create their own
illustrations these days using computer software, a few turn to
a professional artist. If so, do not forget to thank him or her. I
remember an artist showing me a textbook in botany that she
had illustrated. She put her soul into it, she said; but she was not
mentioned in the acknowledgments. A statistician would also
appreciate a word, “Jan Doe, for guiding me through the jungle of
statistics,” as would the language corrector.
You are your own editor of your thesis. You may acknowledge
people without their permission, contrary to the custom in papers.
Ph.D. and other doctoral theses
But you must be absolutely certain that the person acknowledged
would have permitted a particular wording, such as:
. . . Yvette Soe, for preparing good fillets of beef; . . .
Cover illustration
If you have a cover illustration, explain it and credit the photographer
or the artist, on one of the first pages. Here is an example (Naylor
Cover picture: A [. . .] rat that has been under too much stress
(left). In comparison, an [. . .] exercising rat that is alert and
ready for the next challenge.
Illustration by Joen Wetterholm.
. . . and here is the illustration. The two rats illustrate the essence
of the thesis.
Figure 18.1 Cover illustration. (Reproduced from Andrew S. Naylor 2005, with
permission from the author; illustration by Joen Wetterholm, JoenArt.)
How to Write and Illustrate a Scientific Paper
A fresh look
I took a fresh look at 100 theses in biomedicine published in Sweden
in 2005 and found the following trends. The average thesis was
based on four papers or articles. The doctoral student was usually
listed first among the authors; the main supervisor, last. Few papers
had more than six authors. Of the four papers or articles, two
were published and two were in press, submitted or in manuscript.
Roughly, the thesis overview was 55 pages long (reference list
excluded) and cited on average 170 bibliographic sources.
The structure of the thesis overview
Most types of scientific writing are highly structured. Thesis
writing is not. So you are free to begin with the parts most likely
to be read first: Abstract, Summary in your native language, and
Acknowledgments. Give each of them a whole sheet of paper and, if
you can restrain yourself, use only one side of the sheet, leaving the
back blank. It will give your dissertation an inviting opening. This
approach I found in a technology thesis (Synnergren 2005). In the
following, you will find a list of headings for the thesis overview
and how to order them. After the presentation of the list, I will give
some further comments on it.
Summary in your native language
Glossary (or Definitions)
List of papers
Contributors to the papers
Thesis at a glance
Ph.D. and other doctoral theses
General introduction
Methods (or Materials and methods)
Results (or Results and comments)
General discussion
Clinical implications [if relevant]
Implications for further research
Papers I–
The heading Glossary denotes here a list of words that are not
explained in the thesis, because they are so well known to specialists
such as one for example, Glial cells. A reader outside the speciality
would appreciate being told that it means, Cells in the supporting
structure of nervous tissue. This example was taken from a two-page
Glossary of a thesis (Andersson Grönlund 2005).
In Definitions you can explain, for example, Preterm delivery, the
meaning of which is evident to any obstetrician but not to every
non-specialist; thus denoting, Delivery before 36 completed weeks.
I have already described and highly recommended Contributors
to the papers and Thesis at a glance.
In the heading Materials and methods, Materials should be
replaced by Patients or Subjects when appropriate as human beings
are not referred to as material.
How to Write and Illustrate a Scientific Paper
Letters and
case reports
The thalidomide letter
The X-ray image of a woman in late pregnancy showed a fetus
without arms. “Once in a lifetime, we are supposed to see something
like this, but I have seen it twice in a couple of months,” said one
radiologist. Rumors of an increase in similar defects were frequent
– the cause unknown.
About two months later, in December 196l, a letter appeared
in The Lancet stating that women receiving thalidomide in early
pregnancy for morning sickness frequently had babies with missing
or deformed limbs (McBride 1961). The letter, comprising only 15
lines, ended with the following question:
Have any of your readers seen similar abnormalities in
babies delivered of women who have taken this drug during
The response was overwhelming. Eventually more than 10 000
babies in almost 50 countries were born with such defects
(Thalidomide UK 2006). After the publication of the letter, the
drug was immediately withdrawn worldwide. This letter counts as
Letters and case reports
the first milestone for The Lancet (, accessed 9
December 2006) since its announcement of the value of penicillin
in 1940.
Format and size of a letter
Just as in the thalidomide letter, you should be brief and to the
point. The format of a letter is almost always the same: a title; a
salutation, for example, “To the Editor”; the letter itself without
subheadings; and a list of references. You may include one figure
or one table, though such additions will be published only if they
add substantially to the letter. The length of a letter, however,
varies widely from one journal to another. Read the Instructions to
Authors for further information.
Letters are of two kinds: those discussing recent articles and
those describing preliminary research. A letter discussing a recent
article is usually sent by the editor to the authors of the article
discussed, and both the letter and any response will be published
together. Research letters are brief reports of novel findings that
might stimulate further research.
Transforming a paper into a letter
It was a clear case of hubris. I had submitted a report in the form
of a full paper on a single patient to The Lancet. Of course, the
paper was rejected. But the editor offered me and my co-workers
the opportunity to have it cut to a letter.
The original paper had 12 pages of running text, two tables,
and 14 references. We decided to accept the letter format only on
condition that the journal managed to include every important
piece of information. The Lancet did (Hoyer et al. 1979). The huge
reduction of text from an original paper to a letter tells us that the
format of original papers may sometimes be an uneconomical form
of presentation.
How to Write and Illustrate a Scientific Paper
Case reports
Ideally, a case report should suggest a hypothesis that can be tested
by others. For example, the associations between estrogens and
endometrial carcinoma (Fremont-Smith et al. 1946) and between
birth control pills and high blood pressure (Woods 1967) were first
suggested by isolated case reports that led to controlled research
and confirmation of the initial hypotheses.
Unfortunately, most case reports submitted to journals are just
another observation of an unusual condition already well known.
The case should instead have been presented at a departmental
The common subtitle “A review of the literature” is inappropriate,
because a case report is too brief (usually only two pages of
running text) to permit even a mini-review. Couldn’t the young
author (it often is a junior) write a review article separately? No,
an acknowledged expert should write the review, often after an
invitation. He or she is supposed to have the experience to evaluate
the articles, emphasize the good ones, merely mention others,
and, above all, have the courage to exclude works that are below
standard. Reviews may include hundreds of references and are
often used when you have to limit your bibliography – and you
must be able to rely on them.
So which case reports do reach the stage of peer review? Most
likely those that formulate a testable hypothesis, or those that have
something to add – for example, a new diagnostic tool or a new
treatment. And, of course, a case report is appropriate if a new
phenomenon has been observed.
Format and size of case reports
If you have such a case to report, do it. The structure and size vary
widely from one journal to the other. Consult the Instructions
to Authors of the journal you choose. Usually, after a short
introduction, the case (headed: “Case report”) is presented,
Letters and case reports
followed by a short discussion and a list of references. Roughly,
the length is limited to two pages (double spaced) of running text,
five references, and one figure or one table. To save space, present
only relevant findings and do not give whole strings of normal
serum electrolytes and white-cell counts. To limit the extent of
the bibliography, cite references to comprehensive reviews of the
A case report was rejected on formal grounds as it had six figures
whereas the journal allowed only one. The author was quick to
resubmit the manuscript with the same figures, although he had
changed Figures 1–6 to Figure 1, a–f! This was not a joke; the
author probably thought he could cheat an overworked editor with
this trick.
Another case report dealt with an observation made during
a routine abdominal operation. The report shed new light on
such cases as would merit publication. However, the report had
seven authors – too many to find room at the operating table.
The manuscript was returned to the corresponding author who
was asked to declare the contribution of each author. Instead, he
resubmitted the manuscript with five of the seven authors omitted.
The manuscript was accepted. A case report seldom needs more
than two authors, one who made the observation and, if necessary,
one who monitored the writing.
How to Write and Illustrate a Scientific Paper
It has long been the custom to spell out numbers below 10, as
shown in this example from Newsweek (Ridley 2003):
. . . a rat has seven neck and 13 thoracic vertebrae, a chicken 14
and seven . . .
But authorities on scientific style now agree that all numbers
should be expressen in numerals, rather than in words, in most
. . . a rat has 7 neck and 13 thoracic vertebrae, a chicken 14
and 7 . . .
However, we should still spell out numbers that begin a sentence.
The following example is from the abstract of a published paper:
Three thousand eight hundred and seventy-six mothers were
examined by ultrasound at 7–12 weeks of gestation. One hundred
and sixty-six (4.3%) were found to have a dead fetus.
But many readers find it difficult to grasp large numbers written
in words, as in the example shown. Note how much easier to
comprehend the passage becomes when it is recast so that the
numbers fall somewhere in the middle:
Ultrasound examination of 3876 women at 7–12 weeks of gestation showed that 166 (4.3%) had a dead fetus.
Two numbers side by side
Placing unrelated numbers next to each other confuses the reader,
as in this example taken from Mosteller (1992):
This group of patients with leukemia had an average white-cell
count of 257, 112 lymphocytes and 145 other types.
Separate the numbers:
This group of patients with leukemia had an average white-cell
count of 257, of which 112 were lymphocytes and 145 other
Here is another confusing construction:
2 500-mg tablets
Spell out the number easier to express in words and leave the other
in numerical form:
two 500-mg tablets
Decimal point
The decimal sign in English is a point, not a comma:
0.3 (not 0,3)
Use a zero before the decimal point:
0.3 (not .3)
How to Write and Illustrate a Scientific Paper
American and British practice has been to indicate thousands
with commas. In many non-English speaking countries, however,
the comma serves as a decimal marker. To avoid confusion for
an international readership, the Council of Biology Editors’ Style
Manual Committee (1994, 196) recommends the use of a space to
mark off thousands in English writing:
12 345 (not 12,345)
Not all journals have adopted the practice of spacing. In deference
to the editors of such a journals, follow the house style.
Numbers with several zeros
Modern standard units of measure go up and down in steps of 1000.
An appropriate unit to remove surplus zeros is therefore easy to
3 μL (not 0.003 mL)
In other cases, multipliers (exponents) can be used:
1.6 x 109 bacteria per mL
although multipliers should be avoided if they can be easily
12 million inhabitants (not 12 x 106 inhabitants)
but never use “billion” in a scientific paper; it means 109 in the
USA, but 1012 in most European countries.
Quotients of units
You are allowed to use one slash (/) to express quotients of units:
but not two or more. Thus, “milligram per kilogram per hour” is
preferably presented by means of negative exponents:
mg · kg–1 · h–1 (not mg/kg/h)
The use of negative exponents may be unfamiliar to some readers.
For example, a veterinarian seeing that cows were fed 10 kg · day–1
suggested that they were probably fed at night (Lindsay 1989).
However, the use of negative exponents has come to stay and we
have to get accustomed to it.
In a manuscript in preparation I once read that five percent of the
patients with claudicatio intermittens had home-help service once
a week. How, I wondered, could the community afford this luxury
for people with a relatively moderate handicap. Then I found that
the total number of patients studied had been no more than 20. So,
in fact, only one single patient had received the service!
That mode of presentation contained two errors: the original
data were missing, and the number of patients was too small to
warrant expression as a percentage.
Here are some conventions of scientific writing concerning the
use of percentages:
(1) If the total number is less than, say, 25, percentages should
not be used at all.
(2) If the total number is between 25 and 100, percentages
should be expressed without decimals (7 percent, not 7.2
(3) If the total number is between 100 and 100 000, one decimal
may be added – and only one (7.2 percent, not 7.23 percent).
How to Write and Illustrate a Scientific Paper
(4) Only if the total number exceeds 100 000 may two decimals
be added (7.23 percent).
(5) The original data should always be included:
Death occurred in 209 (7.2%) of the 2901 patients.
Note that the percentage is reported in parentheses, to give
primacy to the original data. Thus, not 7.2 percent (209).
(6) The original data should never be presented with a slash
construction. Thus, not 209/2901 (7.2 percent).
Rounding to two significant digits
Ehrenberg (1977) maintains that numbers are easier to compare
after rounding to two significant digits. (Final zeros do not
matter, as the eye can readily filter them out.) Compare these two
(1) Between 1970 and 1975 the number of legally performed
abortions in Sweden increased, from 17 134 to 33 926.
(2) Between 1970 and 1975 the number of legally performed
abortions in Sweden increased, from about 17 000 to about
34 000.
In the second statement, the numbers have been rounded to two
digits, and the two-to-one relationship between them is much
clearer than in the first statement. However, when exact values
for numerical data matter, such a drastic rounding off is not
recommended, but can be used, say, in the discussion. (See also
Chapter 8, “Rounding off.”)
Numerals within parentheses are used to enumerate items, as in
this example (Animal Info 2002):
Animal Info (2002) lists three mammals as endangered in
Afghanistan: (1) snow leopard, (2) markhor, a member of the
goat family (not recently confirmed), and (3) tiger (may be
extinct here).
However, if references in the text are numbered, italic letters must
be used instead, to avoid confusion:
Animal Info (1) lists three mammals as endangered in
Afghanistan: (a) snow leopard, (b) markhor, a member of the
goat family (not recently confirmed), and (c) tiger (may be
extinct here).
What do we mean by “often”?
When presenting numerical data in text, readers feel more at ease
with prose description than with actual numbers. But be careful!
When 51 researchers fluent in English were asked to quantify the
term often, they suggested a rate somewhere between 28 and 92
percent (average 59 percent; Toogood 1980)! So, nonnumerical
expressions alone should best be avoided. Thus:
Most of the patients (82%) . . .
How to Write and Illustrate a Scientific Paper
Abbreviations should be kept to a minimum. So a formulation such
as the following is not to be recommended (quoted from Spiers
. . . a patient with ASHD and PHMI, SPCABG, who PTA for
ERCP had an episode of BRBPR.
It requires some years in the profession to grasp immediately that
this patient with atherosclerotic heart disease and a history of myocardial infarction, status post-coronary-artery-bypass graft, had
an episode of bright red blood per rectum prior to admission for
endoscopic retrograde choledochopancreatography!
The abbreviations used in this sentence are probably all accepted
in the specialty. But just because an abbreviation is permitted does
not mean that you are obliged to use it.
So when should you consider using an abbreviation? Let us
take an example. The term nonsteroidal anti-inflammatory drug
(accepted abbreviation NSAID) may not warrant abbreviation
unless it occurs, say, a dozen times in a paper of standard length.
Some abbreviations are more readily understood than the full
forms: DNA, AIDS, laser. Often such abbreviations are accepted
in the major bibliographic databases. If so, you are free to use them
without definition – even in the title and the abstract section.
Refrain as far as possible from inventing your own abbreviations.
Try instead to find substitute expressions. Assume that you have
made a study of young mature Sprague Dawley rats. You are now
writing the paper and need to refer frequently to this group. You
therefore consider devising a more convenient construction, such as
the YMSD rats. Forget it! The editor will never accept it. So, what
to do? If there are no other rats mentioned in the paper, just simply
use “the rats”; otherwise, for example, “experimental” or “treated”
Units of measure
Units are abbreviated when they follow a numeral. Otherwise, they
are spelled out:
2 mg (but two milligrams)
Singular and plural have the same abbreviation:
1 wk
6 wk (not 6 wks)
An abbreviated unit takes no period unless it ends a sentence:
mo (not mo.)
Abbreviated units need no explanation.
The title of the film 48 HRS. (1982) – the movie with Nick Nolte
and Eddie Murphy in the leading roles – is fine. But in scientific
writing you should use 48 h. Note that hours is abbreviated with a
lower case letter. The capital H is the symbol for hydrogen.
By the way, Halliwell’s Film Guide 2006 gives a numeral followed
by m after each film. That must be for the freaks, I thought, but
soon I realized that m meant minutes, not meter. In scientific
writing, minutes is abbreviated min; and meter, m.
How to Write and Illustrate a Scientific Paper
General principles
You should introduce your abbreviations one by one as they first
occur in the text, in this way:
nonsteroidal anti-inflammatory drug (NSAID).
But readers will miss this information if they turn directly from the
abstract or introduction to the discussion section, as most readers
probably do. Therefore it would be helpful if you also listed the
abbreviations (headed: “Abbreviations used”). You could place this
list either at the foot of the abstract page (see Chapter 23, “Typing”)
or on a following separate page. In the printed version, the list will
usually appear as a footnote on the article’s title page.
Finally, do not mix abbreviations and spelled-out terms;
use either “nonsteroidal anti-inflammatory drug” or “NSAID”
throughout the paper.
How to present
statistical results
Too often, statistics are used “as a drunken man uses a lamp post,
more for support than illumination” (Sumner 1992). Experts in the
field can tell whether your study really needs statistics; if it does, they
can help you to plan the statistical part of your study, for example,
to estimate the sample size needed to demonstrate a difference (if it
exists) and to choose appropriate statistical methods.
Then, when your study is completed, you will encounter another
serious matter: how to present the statistical results. About half of
such presentations contain statistical errors (Murray 1991). Here
are the most common ones.
Using mean when median is meant
In a descriptive study on back pain in pregnancy, the women were
asked to bend over with their arms hanging down. The distance
between fingertips and floor was then measured. The result (mean
and standard deviation ) was reported as
12 ± 14 cm,
thus ranging between –2 and 26 cm, suggesting that some of the
women must have poked their fingertips a couple of centimeters
How to Write and Illustrate a Scientific Paper
through the floorboards. This surprising conclusion is the result of
reporting asymmetrically distributed (skewed) data by using mean
(the average) and standard deviation instead of median (the value
midway between the lowest and the highest value) and a percentile
range, such as the interquartile range (25th to 75th percentile).
One rule of thumb says that if the standard deviation is greater
than half the mean, the data are unlikely to be normally distributed (bell-shaped). In fact, most results in biomedical science are
asymmetrically distributed (Lang and Secic 1997, 47).
If you present, in the same table, both normally and nonnormally distributed data, this should be indicated in a footnote
(see Chapter 8, “Typing the table”).
Using standard error
instead of standard deviation
Standard error of the mean (SEM; often incorrectly abbreviated as
the unspecified SE) is invariably smaller than standard deviation
(SD). It is therefore tempting to describe a set of observations with
mean and SEM so as to suggest less variation in the observations.
But to do so is inappropriate, as SEM, rather than being a descriptive
term, reports the precision of an estimate of the mean in relation to
its unknown value. SD, on the other hand, measures the spread of
individual results around an observed mean.
Failure to distinguish between
statistical significance and biological importance
In the following example, borrowed from Lang and Secic (1997, 58),
a drug was found to lower the diastolic blood pressure by a mean
of 8 mm Hg, from 100 to 92 mm Hg – which was statistically
significant (P < 0.05).
How to present statistical results
However, as Lang and Secic say, a more informative way to estimate
an effect is to construct a confidence interval. (Simply put, a
95 percent confidence interval [CI] is the range within which one
can be 95 percent certain of including the true value.) In this case,
a 95 percent CI was 2 to 14 mm Hg. This tells us that the reduction
in blood pressure could be as much as 14 mm Hg, which would
be clinically important, whereas a reduction of 2 mm Hg would
not. Thus, a result can be statistically significant yet clinically
inconclusive. In the running text the result could be presented in
this way:
Diastolic blood pressure was lowered by a mean of 8 mm Hg,
from 100 to 92 mm Hg (95% CI = 2 to 14 mm Hg; P = 0.02).
Thus, P values estimate the statistical significance while confidence intervals also estimate the clinical significance. So, when
the confidence interval is used, readers do not have to rely on the
author’s interpretation; they can judge for themselves.
Selected presentation of multiple statistical
Multiple testing can generate significant differences where none
exist. With the conventional threshold of P = 0.05 to define a
significant result, there is a 1 in 20 risk of finding a significant
difference even when comparing two groups that are actually
alike. To present only significant results of multiple testing,
as if they were the only analyses performed, is inappropriate
– to say the least. “If the fishing expedition catches a boot,
the fishermen should throw it back, not claim that they were
fishing for boots” (Mills 1993).
One way to accommodate the multiple testing problem is to
adjust the P value by the Bonferroni method, that is, to divide
the P value by the number of tests made. However, for large
numbers of comparisons, the adjusted P value may be almost
How to Write and Illustrate a Scientific Paper
A better approach could be to decide, even in the planning stage,
which test is of major interest and focus your attention on this
variable when analyzing the data and writing the paper. Other data
should be analyzed too, and interesting findings used for further
research. (In fact, many fundamental breakthroughs stem from
such unexpected findings.)
The next section deals with subgroup analyses, which pose problems similar to those of multiple testing.
Overinterpretation of subgroup effects
In a trial conducted on 16 027 patients with suspected acute
myocardial infarction, Collins et al. (1987) made the incidental
observation that the benefit of treatment was fourfold greater for
patients born under the astrological sign of Scorpio than for patients
born under all other signs put together. Computer searching
through numerous subgroups makes it almost inevitable that some
spurious “significant” results, like this one, will appear. However,
it is reasonable to carry out a small number of subgroup analyses
– provided that these are specified in advance (Altman 1995, 466).
This kind of subgroup analysis could provoke ideas to be confirmed
(or refuted) in future studies.
Another aspect of subgroup analysis deals with heterogeneity
analysis in trials with statistically significant results, in order to
assess whether the results are applicable to all patients (Rothwell
1995). For example, one study showed a statistically significantly
higher perinatal mortality among newborns of immigrant women
than among those of women of Swedish origin. Subgroup analyses,
however, revealed that the results were applicable only to newborns
of women from Sub-Saharan Africa, who had a particularly high
perinatal mortality.
How to present statistical results
Using relative instead of absolute figures
In a large Swedish trial of mammography screening, a 24 percent
reduction in mortality from breast cancer was reported. This
impressive figure led to public clamor for screening programs,
which would probably not have arisen if an absolute figure had
been reported instead of a misleading relative figure. As breast
cancer mortality declined from 0.51 to 0.39 percent, the reduction
in absolute terms was actually only 0.12 percent.
A third way to present the result would be by giving the Number
Needed to Treat in order to protect one of them from the disorder
(Chatellier 1996). In this case, 833 women would have had to
be screened regularly for 12 years to prevent a single death from
breast cancer. This way of presenting results is easily understood
by both doctors and patients. The relative risk reduction should
therefore not be cited without simultaneously indicating the
absolute risk reduction or Number Needed to Treat (Laupacis et
al. 1992).
Finally, the results could be expressed also in terms of events per
100 000 of person-years.
Some further comments
The ± sign
The notation of an observed mean as 12.3 ± 0.4 gives no indication
as to whether the second figure is a standard deviation or something
else. A clearer presentation would be:
the mean was 12.3 (SD 0.4)
the mean (SD) was 12.3 (0.4).
With this construction you also avoid the ± sign. Some journals do
not allow its use.
How to Write and Illustrate a Scientific Paper
P < 0.05 ≠ the truth
A firmly ingrained idea is that P < 0.05 = the truth, while P > 0.05
= unpublishable. But P values of 0.04 and 0.06, which differ very
little, ought to lead to similar interpretations rather than radically
different ones. To emphasize this point, the editor of one of the
world’s major journals, The Lancet, was prepared to strip! At a
workshop for editors, he removed his jacket, tie, and shirt to display
a T-shirt bearing a crossed-out sign stating P < 0.05 (Crossan and
Smith 1996). Some journals now ask for exact P values when values
fall above 0.001. The sign < (less than) attached to the P would
thus be used only at the extreme P < 0.001.
How to present statistical results
On the following pages I will show you a manuscript with a layout
that can almost invariably be used as a model when typing your
own. It follows the Vancouver recommendations but, as a model,
can be adapted to comply with most other instructions.
Figure 23.1 The editor at work. (Cartoon by Louis Hellman, first published in
H.E. Emson. BMJ 1994; 309:1738; reproduced with permission.)
How to Write and Illustrate a Scientific Paper
If you follow this layout, you will never again need to ask yourself
in front of your word processor: “Well, how shall I do it this time?”
Instead, you can concentrate on what you have to say. Again I have
used Once upon a time as the running text.
Text in 12-point Times and double-spaced
How to Write and Illustrate a Scientific Paper
How to Write and Illustrate a Scientific Paper
How to Write and Illustrate a Scientific Paper
“Twain spacing”
Samuel L. Clemens (Mark Twain) claims in his autobiography that
Tom Sawyer (published in 1876) was the first typewritten book
manuscript (White 1988). It was double-spaced.
Although most authors now double-space the main text of their
manuscripts, many appear to regard single spacing as adequate for
the reference list. Yet this is the section in which most editorial
changes have to be made. Single-spaced text is impossible to edit
The sole exception from double spacing is text within a table that
forms groups of words, where each group may be single-spaced,
with a double space between groups (see also Chapter 8, “Typing
the table”):
Once upon a time once upon a
time once upon a time
Once upon a time once upon a
time once upon a time
12-point Times
The Times font is a standard choice for newspapers and other
periodicals. (It was cut originally for the daily newspaper The Times
of London.) Text set in Times is easy to read and its compact design
saves space. So, Times might be the right choice for you when
typing the manuscript. A font size of 12 points is recommended.
There is one exception, however. Helvetica, a sans serif typeface,
is considered better for text that is not intended for continuous
reading, such as that of graphs.
How to Write and Illustrate a Scientific Paper
Avoid using block capitals
whereas lowercase letters (small letters) have ascenders, such as b,
and descenders, such as p, which distinguish the letters from each
other and make reading easier.
However, in first-level headings, such as RESULTS, capitals are
recommended, in order to distinguish these headings from subheadings.
Up-and-down-style versus down style
Most journals in the USA use the “up-and-down-style” in titles
and headings. This style means that the first letter of each word is
capitalized. Exceptions are articles (a, the), prepositions (on, in),
and coordinate conjunctions (and, or). Here is an example from
N. Engl. J. Med. (Grüters et al. 1995):
Persistence of Differences in Iodine Status in Newborns after
the Reunification of Berlin.
Most other countries use the “down style,” also called sentence style,
meaning that only the first letter of the first word is capitalized:
Persistence of differences in iodine status in newborns after the
reunification of Berlin.
In this book I have used the down style, except for the title of the
book (How to Write and Illustrate a Scientific Paper). Well, I have
never said that this style is without charm, only that down style is
more reader-friendly – at least for an international readership.
Do not mimic the journal’s style
Many of the 600-odd journal publishers that have adopted the
Vancouver style do not themselves adhere to its requirements in
every detail. They do recommend, however, that authors submitting
manuscripts “should not try to prepare them in accordance with the
[journal’s style] but should follow the [Vancouver requirements]”
(; accessed 20 January 2002). The purpose of this
recommendation is to relieve authors of time-consuming and
unproductive effort to make their manuscripts comply with any
one journal’s particular requirements. Any necessary changes will
be made by the journal’s copy editor.
The length of the manuscript
Editors are biased in favor of short articles. Even hard facts can be
explained in a few pages. A classic example is the report by Watson
and Crick (1953) on the structure of DNA, which occupies just over
one page in Nature, and is understandable even to non-chemists.
Most biomedical journals will not accept manuscripts exceeding
3000 to 4000 words (plus references, figures, and tables). As one
double-spaced page takes about 300 words, that means a maximum
of about 10 to 14 pages of running text. However, presentation of
results from certain subjects, such as occupational science, medical
ethics, and nursing and health care, may need more space. Check
the Instructions for Authors of your target journal for the number
of words allowed. And never exceed that number! Never!!
Let us postulate that yours is a 12-page paper. To achieve fair
proportions between the different parts of your paper, you should
devote about one page to the introduction, and about 3–4 pages
each to the methods, results, and discussion sections.
How to Write and Illustrate a Scientific Paper
The importance of punctuation
Pay attention to punctuation. For example, a comma placed incorrectly can make a difference, as in Lynne Truss’s Eats, Shoots &
Leaves (2003). As Truss explains on the jacket, a panda walks into
a café, eats a sandwich and shoots a gun into the air. On his way
out, he tosses a badly punctuated wildlife manual at the confused
waiter and tells him to turn to the section about his species. The
waiter turns to the page and reads:
Panda. Large black-and-white bear-like mammal, native to
China. Eats, shoots and leaves.
I saw the following passage in the weekly magazine Time. In the
absence of appropriate punctuation, it will be misunderstood:
Woman without her man has no reason for living,
which should read:
Woman: without her, man has no reason for living.
Some learn the typographical conventions early, like the five-yearold girl who was asked to explain why we have “dots” at the end of
sentences (Henshaw, quoted by Hartley 1994). She answered:
It’s to finish a sentence. If you don’t put a dot and you write
a letter people might think you’ve forgotten to post the other
Dealing with editors
and referees
Here is a question from a course participant:
Am I entirely left to the tender mercies of the editors and the
referees? Or do I dare to argue for my own view when I feel that
the referee might have misunderstood a certain point? Am I
impolite if I do so?
No, you aren’t – if you do it politely. Thus not exactly in the way
quoted below from a covering letter to the editor of Cardiovascular
Research (Hearse and the Editorial Team 1992):
Many of the “problems” the referee had with our manuscript
appear to stem from his limited understanding of electrophysiology or from our failure to explain observations at a more
basic level.
In this case the referee happened to be a most eminent researcher in
electrophysiology. Try instead to write as though the referee were
God the Father Himself. But don’t hesitate to make your point:
Thank you for the constructive criticism of my paper. Here are
my comments on the referee’s suggestions.
How to Write and Illustrate a Scientific Paper
Page 3, lines 2–5. What I wanted to say here was . . .
I have rewritten this passage to make my point of view more
Page 4, lines 3–5. . . .
Don’t forget that the referee might have sacrificed hours of unpaid
effort on your manuscript.
Do referees delay?
Here is another question from a course participant:
How big is the risk that the paper goes to a competitor who
delays the whole thing?
That referees delay publication while they incorporate the ideas
of the refereed work into their own publication is extremely rare.
With few exceptions, referees are honorable men and women.
If these words do not allay your fears, what can you do to
protect your ideas from being stolen? One way could be to present your results at a conference before submitting the paper for
publication – your results will then be safeguarded in a conference
abstract. But can you be absolutely sure that the person reading
your submitted abstract will not leak your ideas before the abstract
is published? No, there can be no such certainty. So, if you cannot
live with that, you might consider not publishing – and science will
be deprived of your interesting findings.
Unpublished work
The referee (also known as reviewer) must have access to all
papers you refer to in your manuscript. So when you submit your
manuscript, enclose copies of any works “in press,” “in manuscript,”
or “in preparation” that you mention in your paper. If you haven’t
Dealing with editors and referees
done so, the assessment of your work may be incomplete. One
reviewer wrote:
Much of the key cited methodological material is “in press” and
cannot be judged by this reviewer.
But don’t forget that referees have access to their own works. One
referee told me in her answer:
A lot of what they say is virtually a direct quotation of my own
Shortening the manuscript
An author had been asked to shorten the text of his manuscript. In
his (serious?) covering letter, he said (Baumeister 1992):
You suggested that we shorten the manuscript by 5 pages, and we
were able to accomplish this very effectively by altering the margins
and printing the paper in a different font with a smaller type face.
We agree with you that the paper is much better this way.
A more ambitious author did it in this way:
As suggested in your letter, we have reduced the text by close to
30%. The word count in the revised version, compared with the
previous version:
Total count
Is now
Percent cut
Perhaps you can find a middle-of-the-road approach.
How to Write and Illustrate a Scientific Paper
Figure 24.1 The editor at his desk.
Accepted or rejected
If your paper is accepted, you may receive a preprinted card with a
short statement, as I did:
Dear Doctor:
Your manuscript has been accepted for publication. It is now
being sent to the Publisher and in due time you will receive a
In contrast, a letter of rejection may seem almost movingly
considerate, as this one I received from The Lancet:
Dear Dr Gustavii,
I hope we shan’t dismay you by failing to accept this paper. I
should like to have enabled our readers to see your further interesting findings but at present we are in such trouble from pressure
on space. . . . I am sure you will readily make an alternative
Dealing with editors and referees
I was dismayed, but found consolation in the fact that I was not
alone. In 1937, Nature rejected a submission from Hans Krebs in
which he described the citric acid cycle – one of the central features
of cellular metabolism, now known as the Krebs cycle. The paper
was accepted instead by the editor of Enzymologia. In 1953, Krebs
was awarded a Nobel Prize in recognition of his work.
How to Write and Illustrate a Scientific Paper
Correcting proofs
Popeye, the beloved cartoon character, would probably never have
been created had it not been for a misplaced decimal point. As you
know, Popeye gets his strength by eating spinach, assumed to be
rich in iron. This misconception derives from a report indicating,
due to a misplaced decimal point, that spinach has an iron content
tenfold higher than its true value. An overlooked error seldom has
such amusing consequences, however.
How to read proof
When you receive your masterpiece, nicely typeset in the form of a
proof, you may be tempted to read it straight through at that very
moment. My advice is to follow your intuition. You will be on the
alert and will easily notice if the reading makes sense, thus catching
errors of omission, such as a dropped line or a lost paragraph. In
order not to overlook printer’s errors, however, you will have to
reread the proof at least once more.
For the second reading, persuade someone to slowly read the
manuscript aloud while you check the text in the proof. If you can’t
find a reader, place a finger under the first line of the manuscript
and a finger under the first line of the proof, just under the first
character. Look from manuscript to proof and back again, checking
Correcting proofs
word by word, numeral by numeral, and punctuation mark by
punctuation mark. Be especially careful in checking the tables and
the reference list.
Another option that I have heard of but not used is to read the
manuscript into a tape recorder and then listen to the recording
while looking at the proof.
If you check the proof too hastily, you may live to regret it. I
once overlooked a mistyped numeral and afterward had to correct
it by hand in 300 reprints.
What to correct
The main reason for sending you the proofs is so you may correct the
typing errors. At this stage, you are not allowed to polish the prose.
You may, however, correct a serious mistake, such as inconsistency
between data in the abstract and in the main text of the paper.
Moreover, if you have changed your mind about part of the
content, or if you have acquired relevant new information, you can
write an addendum (also called “Notes added in proof”) placed
at the end of the main text, before the list of references. The main
reason for using an addendum is the ethical aspect of adding to the
body of the text some new matter or a revision containing material
not seen by the referee. Here is an example of presentation of new
information (Federle et al. 1982):
Since this manuscript was submitted, 25 additional patients
have been studied. DR [Digital Radiography] pelvimetry
has completely replaced the conventional method at this
If you have referred to a paper as in “in press” and this paper has
already been published, update it by providing the volume, year,
and pagination.
How to Write and Illustrate a Scientific Paper
Correction marks
There are several systems of proof correction: the continental
European system, the British system, the American system, and
various systems used in other countries. But typesetters worth their
salt can cope with them all. So there is no need for you to learn
more than one system. And the one I recommend is the American
system, as, in my opinion, it is the easiest to use. On the following
pages you are shown the most common American proofreader’s
marks and how to use them.
Unlike corrections in the manuscript, corrections in the proof
must be marked twice, once at the point where the error occurs and
once in the margin. Typesetters scan the margins and won’t notice
corrections indicated in the text if you have failed to indicate the
change in the margin.
You need not know more than the marks given in this chapter;
they cover most cases of correction in the average proof. But, in
those few cases where you cannot find a suitable mark here, just
write an instruction to the typesetter in the margin and circle it.
Write the correction beside the instruction and make an appropriate
mark in the text to show where the new, corrected material is to be
Electronic proof
Technology now exists whereby those correcting electronic proofs
can electronically show correction.
Correcting proofs
Table 25.1 Commonly used American correction marks
How to Write and Illustrate a Scientific Paper
Here is a sample of text with errors marked, corrected, and commented upon.
Correcting proofs
Subjects’ right to privacy
How to protect a subject’s identity? In this figure, executed with a
dash of humor, the anonymity of both subjects has been protected
by the traditional black band across the eyes.
Figure 26.1 Simian stance, which resembles the posture of an anthropoid
ape, can be a valuable clue to diagnosis of spinal stenosis. (Reproduced, with
permission, from P.A. Simkin. Simian stance: a sign of spinal stenosis, The Lancet
1982; ii (8299):652–3, © The Lancet Ltd.)
How to Write and Illustrate a Scientific Paper
Nevertheless, a few weeks later, in Letters to the Editor, a reader told
us that despite the bar across the animal’s eyes, he had immediately
recognized it as one he had seen in Melbourne Zoo and suggested
that authors should take greater care to preserve anonymity when
presenting ape data (Millar 1982).
This observation tells us that a black band across the eyes may be
insufficient to disguise the subject. So, in cases like this, informed
consent should be obtained, as recommended by the International
Committee of Medical Journal Editors (1995). The woman in the
figure had given such consent. Simkin’s figure also tells us that
authors need not always be deadly serious in their reporting. Rather,
humor can help to convey the message.
Duplicate submission
One of my course participants asked:
If I submit a paper to a journal, can I at the same time, in a revised form, send a paper (same subject, same material) to
another journal? Or would my young research career finish
there and then?
Yes, that would be the end! A unanimous jury would give you
nothing less than a life sentence for self-plagiarism, as happened
to an author who had submitted the same manuscript to two
journals in the USA. All American journals publishing in that field
agreed never again to consider any manuscript from that scientist’s
laboratory (Abelson 1982).
Borrowing published material
Here is another question from a course participant:
If I redraw a published picture and make some small alterations,
can I then call it “my own”?
Authors’ responsibilities
No, usually not. If the published picture has “original features,”
and if these are retained in the redrawn figure, then you should
seek permission to reproduce the picture. Now, what constitutes
the originality of an illustration? As even judges in lawsuit cases
on this subject can disagree, we can do no better than use our
intuition to find an answer when this question is raised. Here is my
own personal view on two cases.
I saw the original Viking cartoon in a daily newspaper, (top
left in Figure 26.2). The Viking encourages the readers to say NEJ
(NO) to the European Union, a subject far from teaching how to
write a scientific paper. But I found the figure useful for the purpose
by replacing the text on the stone and, after whiting out the eye
and mouth regions, remaking the facial expression. However, these
alterations do not make the redrawn figure “mine,” as its original
features are retained almost unchanged. I therefore had to obtain
permission to publish it (bottom in Figure 26.2; see also Figure
Figure 26.2 Permission is needed for publication of the redrawn cartoon (bottom),
as the originality of the picture (top left) is retained. (Redrawn, with permission,
from Majewski 1994.)
How to Write and Illustrate a Scientific Paper
On the other hand, when a figure’s characteristics are not copied,
permission is not needed. For example, I used the textbook figure
(the left one in Figure 26.3) without permission as a model when
drawing the figure to the right.
Figure 26.3 Permission not needed for the redrawn figure (right), as the
characteristics of the original figure (left) are not copied.
However, to reproduce these two figures in the present book, I did
have to obtain permission. Thus, the left figure is reproduced,
with permission, from Arey 1954; that on the right is reproduced
from Gustavii 1975, with the publisher’s permission. Note that I
had to obtain permission to reproduce even a figure I myself had
drawn for a paper I had authored myself. As is often the case, I was
required before the publication of my paper to sign a statement
transferring my copyright to the publisher, including both text and
When you seek permission to reproduce a figure (original or
redrawn), write not only to the publisher but also to the author and,
if appropriate, to the illustrator as well. For example, the copyright
owner of Figure 23.1 in this book appeared to be the artist. (The
publisher told me so and gave me her address.) Thus, although the
copyright often belongs to the publisher, this is not always the case.
Authors’ responsibilities
Most journals require permission from both copyright owner and
author. Permission is needed also to reproduce a table (or part of it)
or to quote text at length (say, 100 words or more).
Many publishers give permission freely, but some charge a fee.
For the copyrighted material (44 graphs, seven tables) in the first
edition of this book, fees were charged for three graphs and one
table; the graphs cost me (in 1998) £30, £30, and US$20; the table,
US$25. Most publishers responded to my request for permission
within one to three months. One publisher, however, replied only
after ten months and two reminders.
Nowadays, however, many publishers of scientific journals
provide online forms for requesting permission. Because these
forms can be exhaustingly detailed, it may be easier for you to send
a request by e-mail. Use a template (Figure 26.4) that you can save
in your computer. Permission is often granted within hours, even
minutes – if the request is received during working-hours.
The old template in Figure 26.5 is still useful. For example, in
2006, when I sought permission from a number of authors, I was
unable to find e-mail addresses in two cases, and I had to send the
template by regular mail.
The wording of the credit line I suggest, for example, “Reproduced, with permission, from Simkin 1982,” is not always accepted.
Some copyright holders wish to specify in detail the wording of the
acknowledgment, which must be followed strictly, even if it means
that the credit line becomes much longer than the figure legend
(see Figure 14.1).
How to Write and Illustrate a Scientific Paper
Figure 26.4 A template to be used when requesting, by e-mail, permission to use
copy-righted material.
Authors’ responsibilities
Figure 26.5 A request sent by regular post.
How to Write and Illustrate a Scientific Paper
Saving your original data
When you have completed your paper you may wonder what to do
with all the raw material you have collected. Should you discard
it? No, definitely not. Your target journal may require access to
it. For example, if you submit a paper to JAMA, you have to sign
a statement saying, “I certify that if requested, I will provide the
data or will cooperate fully in obtaining and providing the data
on which the manuscript is based for examination by the editors
or their assignees” (JAMA 2002). Even after publication, you are
obliged to provide those who request it, with the original data of
your paper – a responsibility you will bear for at least five years
after publication (some say 10 years).
Authors’ responsibilities
Literature needed
on your desk
On phraseology
Hornby, A. S.; Ashby, M. 2005. Oxford Advanced Learner’s
Dictionary of Current English. 7th edn. New York: Oxford
University Press.
On synonyms
The Merriam-Webster Dictionary of Synonyms and Antonyms. 1998.
Springfield, MA: Merriam-Webster.
On manuscript preparation
International Committee of Medical Journal Editors.
Uniform Requirements for Manuscripts Submitted to Biomedical Journals (the Vancouver Document,
How to Write and Illustrate a Scientific Paper
On how to abbreviate a journal’s title
List of journals in the Index Medicus. Bethesda, MD: National
Library of Medicine; published annually as a list in the
January issue of Index Medicus. For sale on the Net by, as a separate publication. Also available at:
Literature needed on your desk
Further reading
Guides to writing
Albert, T. 2000. The A–Z of Medical Writing. London: BMJ Books.
Alley, M. 1997. The Craft of Scientific Writing. 3rd edn. New
York: Springer-Verlag.
Booth, V. 1993. Communicating in Science. Writing a Scientific
Paper and Speaking at Scientific Meetings. 2nd edn.
Cambridge: Cambridge University Press.
Browner, W. S. 1999. Publishing and Presenting Clinical Research.
Baltimore: Williams & Wilkins.
Byrne, D. W. 1998. Publishing Your Medical Research Paper. What
They Don’t Teach You in Medical School. Baltimore: Williams
& Wilkins.
Davis, M. 2004. Scientific Papers and Presentations. 2nd edn. San
Diego: Academic Press.
Day, R. A.; Gastel, B. 2006. How to Write and Publish a Scientific
Paper. 6th edn. Westport, CT: Greenwood Press.
Goodman, N. W.; Edwards, M. B.; Black, A. 2006. Medical
Writing: A Prescription for Clarity. 3rd edn. Cambridge:
Cambridge University Press.
Hall, G. M., editor. 2003. How to Write a Paper. 3rd edn. London:
BMJ Books.
How to Write and Illustrate a Scientific Paper
Huth, E. J. 1999. Writing and Publishing in Medicine. 3rd edn.
Baltimore: Williams & Wilkins.
Iles, R. L. 2003. Guidebook to Better Medical Writing. Washington,
D.C.: Island Press.
King, S. 2002. On Writing. A Memoir of the Craft, pp. 139–288.
New York: Pocket Books.
Malmfors, B.; Garnsworthy, P.; Grossman, M. 2005. Writing
and Presenting Scientific Papers. 2nd edn. Nottingham:
Nottingham University Press.
Matthews, J. R.; Bowen, J. M.; Matthes, R. W. 2001. Successful
Scientific Writing. A Step-by-Step Guide for the Biological and
Medical Sciences. 2nd edn. Cambridge: Cambridge University
Montgomery, S. L. 2002. The Chicago Guide to Communicating
Science (Chicago Guides to Writing, Editing, and Publishing).
Chicago: University of Chicago Press.
O’Connor, M. 1991. Reprint 1999. Writing Successfully in Science.
London: E & FN Spon.
Schoenfeld, R. 1989. The Chemist’s English. 3rd edn. New York:
Woodford, F. P. 1999. How to Teach Scientific Communication.
Bethesda: Council of Biology Editors.
Taylor, R. B. 2005. The Clinician’s Guide to Medical Writing. New
York: Springer.
Zeiger, M. 1999. Essentials of Writing Biomedical Research Papers.
2nd edn. New York: McGraw-Hill Professional Publishing.
Carey, G. V. 1976. Mind the Stop. A Brief Guide to Punctuation.
London: Penguin.
Truss, L. 2003. Eats, Shoots & Leaves. The Zero Tolerance Approach
to Punctuation. London: Profile Books.
Further reading
The English language
Day, R. A. 1995. Scientific English. A Guide for Scientists and
Other Professionals. 2nd edn. Phoenix: Oryx Press.
Strunk, W. Jr.; White, E.B.; Angell. R. 2000. The Elements of
Style. 4th edn. Boston: Allyn & Bacon.
Style manuals
The ACS Style Guide: Effective Communication of Scientific
Information (An American Chemical Society Publication).
2006. 3rd edn. Edited by Anne M. Coghill and Lorrin R.
Garson. Washington, DC: American Chemical Society.
American Medical Association Manual of Style. 2007. 10th edn.
New York: Oxford University Press.
Publication Manual of the American Psychological Association.
2001. 5th edn. Washington, DC: American Psychological
Science & Technical Writing. A Manual of Style. 2001. 2nd edn.
Edited by Philip Rubens. New York: Routledge.
Scientific Style and Format: The CBE Manual for Authors, Editors,
and Publishers. 1994. 6th edn. Edited by Edward J. Huth.
New York: Cambridge University Press.
Briscoe, M. H. 1996. Preparing Scientific Illustration. A Guide to
Better Posters, Presentations, and Publications. 2nd edn. New
York: Springer-Verlag.
Harris, R. L. 1999. Information Graphics. A Comprehensive
Illustrated Reference. New York: Management Graphics.
Tufte, E. R. 2001. The Visual Display of Quantitative Information.
Cheshire, CT: Graphics Press.
How to Write and Illustrate a Scientific Paper
Altman, D. G. 2006. Practical Statistics for Medical Research. 2nd
edn. London: Chapman & Hall.
Lang, T. A.; Secic, M. 2006. How to Report Statistics in Medicine.
Annotated Guidelines for Authors, Editors, and Reviewers. 2nd
edn. Philadelphia: American College of Physicians.
Further reading
Literature cited
Abelson, P. H. 1982. Excessive zeal to publish. Science 218:953.
Altman, D. G. 1995. Practical Statistics for Medical Research. London:
Chapman & Hall.
Altman, D. G. 1985. Comparability of randomised groups. Statistician
Andersson Grönlund, M. 2005. Ophthalmologic characteristics and
neuropediatric findings – with special emphasis on children adopted
from Eastern Europe. (Dissertation.) Göteborg, Sweden: The
Sahlgrenska Academy at Göteborg University.
Andolf E. 1989. Sonography of the female pelvis with emphasis on ovarian
tumours. (Dissertation.) Lund, Sweden: University of Lund.
Animal Info. 2002. Information on rare, threatened and endangered
mammals. Severna Park (MD): Animal info.
(accessed 2 February 2002).
Arey, L. B. 1954. Developmental Anatomy. A Textbook and Laboratory Manual
of Embryology, p. 130. 6th edn. Philadelphia: Saunders.
Azem, J. 2005. Approaches to analyses of cytotoxic cells, and studies of
their role in H. pylori infection. (Dissertation.) Göteborg, Sweden: The
Sahlgrenska Academy at Göteborg University.
Bailar, J. C. 1986. Science, statistics, and deception. Ann. Intern. Med.
Baker, J. R. 1955. English style in scientific papers. Nature 176:851–2.
Baranto, A. 2005. Traumatic high-load injuries in the adolescent spine.
Clinical, radiological and experimental studies. (Dissertation.)
Göteborg, Sweden: The Sahlgrenska Academy at Göteborg University.
How to Write and Illustrate a Scientific Paper
Baumeister, R. F. Dear Journal Editor, it’s me again. Dialogue [cited by
Hearse, 1992].
Begg, C.; Cho, M.; Eastwood, S.; Horton, R.; Moher, D.; Olkin, I.; et al.
Improving the quality of reporting of randomized controlled trials. The
CONSORT statement. JAMA 1996; 276:637–9.
Begley, S. 1996. To stand and raise a glass. Newsweek, July 1, 42–5.
Bengtsson, L. P. 1968. Therapeutic abortion by means of intra-uterine
injections. Techniques, effects, risk and mechanisms of effect. Medical
Gynecology & Sociology 3:6–14.
Bennet, S. 1992. Churchill [documentary]. London: British Broadcasting
Bergström, A. 1994. Experimental retinal cell transplants. (Dissertation.)
Lund, Sweden: University of Lund.
BMJ editorial 1985. Reference 13. 291:1746.
Broad, W. J. 1981.The publishing game: Getting more for less. Meet the least
publishable unit, one way of squeezing more papers out of a research
project. Science 211:1137–9.
Brownlow, K.; Gill, D. 1983. Unknown Chaplin [video]. Thames Video
Burrough-Boenisch, J. 2006. EATAW (European Association for the
Teaching of Academic Writing) email forum.
(accessed 8 February 2006).
Carling, P. 2006. EATAW (European Association for the Teaching of
Academic Writing) email forum. (accessed
9 December 2006).
CBE. 1994. See Council of Biology Editors’ Style Manual Committee.
Chaparro, C. M.; Neufeld, L. M.; Alavez, G. T.; Cedillo, R. E-L.; Dewey,
K.G. 2006. Effect of timing of umbilical cord clamping on iron status in
Mexican infants: A randomised controlled trial. The Lancet 367:1981–9.
Chaplin, C. 1973. My Autobiography, p. 208. Harmondsworth: Penguin.
Chapman, M.; Mahon, B. 1986. Plain Figures, p. 74. London: HMSO.
Chatellier, G.; Zapletal, E.; Lemaitre, D.; Menard, J.; Degoulet, P. 1996.
The number needed to treat: A clinically useful nomogram in its proper
context. BMJ 312:426–9.
Chernin, E. 1988. The “Harvard system”: A mystery dispelled. BMJ
Christensen, K. K. 1980. Group B Streptococci. Aspects on urogenital
epidemiology and obstetrical significance. (Dissertation.) Lund, Sweden:
University of Lund.
Clarke, K. W.; Gray, D.; Keating, N. A.; Hampton, J. R. 1994. Do women
with acute myocardial infarction receive the same treatmen as men?
BMJ 309:563–6.
Literature cited
Collins, R.; Gray, R.; Godwin, J.; Peto, R. 1987. Avoidance of large biases and
large random errors in the assessment of moderate treatment effects: The
need for systematic overviews. Stat. Med. 6:245–50.
CONSORT Website. 2001. (accessed 20 January
Council of Biology Editors’ Style Manual Committee. 1994. Scientific Style and
Format. The CBE Manual for Authors, Editors, and Publishers. 6th edn.
New York: Cambridge University Press.
Crichton, M. 1975. Medical obfuscation: Structure and function. N. Engl. J.
Med. 293:1257–9.
Crossan, L.; Smith, R. 1996. The BMJ/EASE workshop for editors. CBE Views
Danel, C.; Moh, R.; Minge, A.; et al., for the Trivacan ANRS 1269 trial group.
2006. CD4-guided structured antiretroviral treatment interruption
strategy in HIV-infected adults in West Africa (Trivacan ANRS 1269
trial): a randomised trial. The Lancet 367:1981–9.
Day, R. A. 1995. Scientific English. A Guide for Scientists and Other Professionals.
2nd edn. Phoenix: Oryx Press.
Day, R. A.; Gastel, B. 2006. How to Write and Publish a Scientific Paper. 6th
edn. Westport, CT: Greenwood Press.
De Looze, S. 2002. Rules on virgules. European Science Editing 28:108–10.
Dembiec, D. P.; Snider, R. J.; Zanella. A. J. 2004. The effects of transport
stress on tiger physiology and behavior. Zoo Biology 23:335–46.
Dixon, B. Plain words please. 1993. New Scientist 137:39–40.
Dobbel, C. 1938. Dr O. Uplavici (1887–1938). Parasitology 30:239–41.
Ehrenberg, A. S. C. 1977. Rudiments of numeracy. J. R. Stat. Soc. A.
140(pt 3): 277–97.
Emson, H. E. 1994. Christmas wishes. BMJ 309:1738.
European Carotid Surgery Trialists’ Collaborative Group. 1998. Randomised
trial of endarterectomy for recently symptomatic carotid stenosis: Final
results of the MRC European Carotid Surgery Trial (ECST). The Lancet
Federle, M. P.; Cohen, H. A.; Rosenwein, M. F.; Brant-Zawadzki, M. N.;
Cann, C. E. 1982. Pelvimetry by digital radiography: A low-dose
examination. Radiology 143:733–5.
48 Hours. 1982. Director Walter Hill. Paramount/Lawrence Gordon.
Fremont-Smith, M.; Meigs, J. V.; Graham, R. M.; Gilbert, H.H. 1946. Cancer
of endometrium and prolonged estrogen therapy. JAMA 131:805–8.
Gardiner, P. J.; Copas, J. L.; Schneider, C.; Collier, H. O. J. 1980. 2-decarboxy2-hydroxymethyl prostaglandin E1 (TR4161), a prostaglandin
bronchodilator of low tracheobronchial irritancy. Prostaglandins 19:
How to Write and Illustrate a Scientific Paper
Garfield, E. 1986. The integrated Sci–Mate Software System. Part 2: The
Editor slashes the Gordian knot of conflicting reference styles. Current
Contents, March 17(11):81–8.
Godlee, F. 1996. Definition of “authorship” may be changed. BMJ 312:
Gold, D. R.; Wang, Xiaobin; Wypij, D.; Speizer, F. E.; Ware, J. H.; Dockery,
D. W. 1996. Effects of cigarette smoking on lung function in adolescent
boys and girls. N. Engl. J. Med. 335:931–7.
Goodman, R. A.; Thacker, S. B.; Siegel, P. Z. 2001. What’s in a title? A
descriptive study of article titles in peer-reviewed medical journals.
Science Editor 24:75–8.
Grüters, A.; Liesenkötter, K. P.; Willgerodt, H. 1995. Persistence of
differences in iodine status in newborns after the reunification of Berlin.
N. Engl. J. Med. 333:1429.
Gustavii, B. 1975. The distribution within the placenta, myometrium, and
decidua of 24Na-labelled hypertonic saline solution following intraamniotic or extra-amniotic injection. Br. J. Obstet. Gynaecol. 82:734–9.
Halliwell’s Film, Video & DVD Guide 2006. 21st edn. Walken, J.; Halliwell, L.
New York: HarperCollins.
Hamberg, M. 1972. Inhibition of prostaglandin synthesis in man. Biochem.
Biophys. Res. Commun. 49:720–6.
Harris, R. L. 1999. Information Graphics. A Comprehensive Illustrated
Reference. New York: Management Graphics.
Hartley, J. 1994. Designing Instructional Text, p. 31. 3rd edn. London: Kogan
Hearse, D., and the Editorial Team. 1992. Of humour, music, anger, speed,
and excuses: Reflections of an editorial team after one year in office
[editorial]. Cardiovasc. Res. 26:1161–3.
Heim, S.; Kristoffersson, U.; Mandahl, N.; Mineur, A.; Mitelman, F.; Edvall,
H.; Gustavii, B. 1985. Chromosome analysis in 100 cases of first
trimester trophoblast sampling. Clin. Genet. 27:451–7.
Helenius, G. 2005. Tissue engineering of blood vessels. (Dissertation.) Lund,
Sweden: University of Lund.
Hlava, J. 1887. O úplavici [On dysentery] [Journal of Czech Physicians] Jan.
Hodgen, G. D. 1981. Antenatal diagnosis and treatment of fetal skeletal
malformations: With emphasis on in utero surgery for neural tube
defects and limb bud regeneration. JAMA 246:1079–83.
Holmes, W. 1997. Minimum ethical standards should not vary among
countries. The Lancet 314:1479.
Literature cited
Hoyer, L.W.; Lindsten, J.; Blombäck, M.; Hagenfeldt, L.; Cordesius, E.;
Strömberg; P.; Gustavii, B. 1979. Prenatal evaluation of fetus at risk for
severe von Willebrand’s disease. The Lancet 2:191–2.
International Committee of Medical Journal Editors. 2002. Uniform
Requirements for Manuscripts Submitted to Biomedical Journals. www. (accessed 20 January 2002).
International Committee of Medical Journal Editors. 1995. Protection of
patients’ right to privacy. BMJ 311:1272.
International Committee of Medical Journal Editors. 1985. Guidelines on
authorship. BMJ 291:722.
JAMA. 2002. Instructions for authors. Published in the first issue of each
January and July and available at
Jha, T. K.; Olliaro, P.; Thakur, C. P. N.; Kanyok, T. P.; Singhania, B. L.;
Singh, I. J.; et al. 1998. Randomised controlled trial of aminosidine
(paromomycin) v sodium stibogluconate for treating visceral
leishmaniasis in North Bihar, India. BMJ 316:1200–5.
Karman, H.; Potts, M. 1972. Very early abortion using syringe as vacuum
source. The Lancet i:1051–2.
Kartulis, S. 1887. O. Uplavici, Ueber die Dysenterie (review). Centralblatt für
Bacteriologie und Parasitenkunde 1(18):537–9.
Kerkut, G. A. 1983. Choosing a title for a paper. Comp. Biochem. Physiol.
Kesling, R. V. 1958. Crimes in scientific writing. Turtox News 36:274–6.
Kitin, P. B.; Fujii, T.; Abe, H.; Funanda, R. 2004. Anatomy of the vessel
network within and between the tree rings of Fraxinus lanuginosa
(Oleaceae). Am. J. Botany. 91:779–88.
Kurki, T. 1992. Preterm birth. A clinical, biochemical and bacteriological
study. (Dissertation.) Helsinki, Finland: University of Helsinki.
Lancet, The. 1993. OCs o-t-c? [editorial] 342:565–6.
Lancet, The. 1995. English as she is wrote [editorial]. 346:1045.
Lang, T. A.; Secic M. 1997. How to Report Statistics in Medicine. Annotated
Guidelines for Authors, Editors, and Reviewers. Philadelphia: ACP.
Laupacis, A.; Naylor, C. D.; Sackett, D. L. 1992. How should the results of
clinical trials be presented to clinicians? [editorial]. ACP Journal Club,
Lee, A.; Thomas, P.; Cupidore, L.; Serjeant, B.; Serjeant, G. 1995. Improved
survival in homozygous sickle cell disease: Lessons from a cohort study.
BMJ 311:1600–2.
Lindsay, D. 1989. A Guide to Scientific Writing. Manual for Students and
Research Workers, p. 36. Melbourne: Longman Cheshire.
Liu, L. 1996. Fate of conference abstracts. Nature 383:20.
How to Write and Illustrate a Scientific Paper
Logan, R. F. A.; Little, J.; Hawtin, P. G.; Hardcastle, J. D. 1993. Effect
of aspirin and non-steroidal anti-inflammatory drugs on colorectal
adenomas: Case–control study of subjects participating in the
Nottingham faecal occult blood screening programme. BMJ 307:285–9.
López-Jaramillo, P.; Delgado. F.; Jácome, P.; Terán, E.; Ruano, C.; Rivera,
J. 1997. Calcium supplementation and the risk of preeclampsia in
Ecuadorian pregnant teenagers. Obstet. Gynecol. 90:162–7.
Macknin, M. L.; Piedmonte, M.; Calendine, C.; Janosky, J.; Wald, E. 1998.
Zinc gluconate lozenges for treating the common cold in children. A
randomized controlled trial. JAMA 279:1962–7.
Majewski, J. 1994. Sydsvenska Dagbladet [The South Swedish Daily News],
Nov. 6, Sect. A:2 (cols. 2–3).
Marvin, P. H. 1964. Birds on the rise. Bull. Entomol. Soc. Amer. 10:194–6.
Mathews, K. A.; Sukhiani, H. F. 1997. Randomized controlled trial of
cyclosporine for treatment of perianal fistulas in dogs. J. Am. Vet. Med.
Assoc. 211:1249–53.
McBride, W. G. 1961. Thalidomide and congenital abnormalities. The Lancet
McGarry, G. W.; Gatehouse, S.; Hinnie, J. 1994. Relation between alcohol
and nose bleeds. BMJ 309:640.
McNab S. M. 1993. Non-sexist language. TWIOscoop 11(5):148–53.
McWhorter, T.J.; Martínez del Rio, C. 2000. Does gut function limit
hummingbird food intake? Physiological and Biochemical Zoology
Medical Research Council. 1948. Streptomycin treatment of pulmonary
tuberculosis. Br. Med. J. 2:769–82.
Mehrotra, P. K.; Karkun, J. N.; Kar, A. B. 1973. Estrogenicity of some
nonsteroidal compounds. Contraception 7:115–24.
Millar, J. A. 1982. Anonymity of anthropoid apes featured in medical
journals. The Lancet ii:940.
Mills, J. L. 1993. Data torturing. N. Engl. J. Med. 329:1196–9.
Moher, D.; Schulz, K. F.; Altman, D., for the CONSORT Group. 2001.
The CONSORT statement: Revised recommendation for improving
the quality of reports of parallel-group randomized trials. JAMA 285:
Morell, C. J.; Walters, S. J.; Dixon, S.; Collins, K. A.; Brereton, L. L. M.;
Peters, J.; et al. 1998. Cost effectiveness of community leg ulcer clinics:
Randomised controlled trial. BMJ 316:1487–91.
Mosteller, F. 1992. Writing about numbers, p. 378. In Bailar J. C. and
Mosteller, F., editors. Medical uses of statistics. 2nd ed. Boston: NEJM Books.
Murray, G. D. 1991. Statistical aspects of research methodology. Br. J. Surg.
Literature cited
Naylor, A. S. 2005. Differential effects of voluntary running on hippocampal
plasticity in the adult rat brain. (Dissertation.) Göteborg, Sweden: The
Sahlgrenska Academy at Göteborg University.
Os, J. van; Neeleman, J. 1994. Caring for mentally ill people. BMJ 309: 1218–21.
Pitnick, S.; Spicer, G.S.; Markow, T.A. 1995. How long is a giant sperm?
Nature 375:109.
Publications Committee for the Trial of ORG 10172 in Acute Stroke
Treatment (TOAST) Investigators. 1998. Low molecular weight
heparinoid ORG 10172 (danaparoid) and outcome after acute ischemic
stroke. JAMA 279:1265–72.
Quesada, M.; Bollman, K.; Stephenson, A. G. 1995. Leaf damage decreases
pollen production and hinders pollen performance in Cucurbita texana.
Ecology 76:437–43.
Raio, L.; Ghezzi, F.; Di Naro, E.; Gomez, R.; Lüscher, K. P. 1997. Duration
of pregnancy after carbon dioxide laser conization of the cervix:
Influence of cone height. Obstet. Gynecol. 90:978–82.
Reed, D. M. 1990. The paradox of high risk of stroke in populations with low
risk of coronary heart disease. Am. J. Epidemiol. 131:579–88.
Ridley, M. 2003. What makes you who you are. Newsweek, June 2, 50–7.
Rothwell, P. M. 1995. Can overall results of clinical trials be applied to all
patients? The Lancet 345:1616–19.
Samuelsson, S.; Sjövall, A. 1973. Komplikationer och komplikationsprofylax
vid gynekologisk laparoskopi. [Complications and prophylaxis in
gynaecological laparoscopy.] (In Swedish with English abstract.)
Läkartidningen 70:2570–4.
Sarna, S.; Kivioja, A. 1995. Blunt rupture of the diaphragm. Ann. Chir.
Gynecol. 84:261–5.
Scientific Style and Format. 1994. See Council of Biology Editors’ Style
Manual Committee.
Spiers, A. S. D. 1984.Transatlantic medical English. The Lancet ii:1451–3.
Stockdale, T. 2000. Contaminated material caused Creutzfeldt-Jacob disease
(CJD) in some undersized children who were treated with growth
hormone (GH). Nutr. Health. 14:141–2.
Strunk, W., Jr.; White, E. B. 2000. The Elements of Style. 4th edn. Boston:
Allyn & Bacon.
Sumner, D. 1992. Lies, damned lies – or statistics? J. Hypertens 10:3–8.
SUN, Xiao-Lin; ZHOU, Jing. 2002. English versions of Chinese authors’
names in biomedical journals. Observations and recommendations.
Science Editor 25:3–4.
Sundby, J.; Schei, B. 1996. Infertility and subfertility in Norwegian women
aged 40–42. Prevalence and risk factors. Acta Obstet. Gynecol. Scand.
How to Write and Illustrate a Scientific Paper
Synnergren, O. 2005. Time-resolved X-ray diffraction studies of phonons and
phase transitions. (Dissertation.) Malmö, Sweden: Universities of Lund.
Thalidomide UK. 2006. (accessed 27 July 2006).
The ACS Style Manual. 2002. A Manual for Authors and Editors. 2nd edn.
Edited by Janet S. Dodd. Washington, DC: American Chemical Society.
Theander, E. 2005. Living and dying with primary Sjögren’s syndrome.
Studies on aetiology, treatment, lymphoma, survival and predictors.
(Dissertation.) Malmö, Sweden: University of Lund.
Tjio, Joe Hin; Levan, A. 1956. The chromosome number of man. Hereditas
Toogood, J. H. 1980. What do we mean by “usually”? The Lancet i:1094.
Tønnes-Pedersen, A.; Lidegaard, Ø.; Kreiner, S.; Ottesen, B. 1997. Hormone
replacement therapy and risk of non-fatal stroke. The Lancet 350:1277–
Truss, L. 2003. Eats, Shoots & Leaves. The Zero Tolerance Approach to
Punctuation. London: Profile Books.
Tufte, E. R. 1983. The Visual Display of Quantitative Information. Cheshire,
CT: Graphics Press.
Van Loon, A. J. 1997. Making it easier to trace articles in scientific and
technical periodicals: The importance of the first page. European Science
Editing 23:9–12.
Vancouver Document. 2002. Uniform Requirements for Manuscripts Submitted
to Biomedical Journals. (accessed 20 January 2002).
Vane, J. R. 1971. Inhibition of prostaglandin synthesis as a mechanism of
action of aspirinlike drugs. Nature 231:232–5.
Waldron, H. A. 1995. English as she is wrote. The Lancet 346:1567–8.
Wallengren, J. 1998. Brachioradial pruritus. A recurrent solar dermopathy.
J. Am. Acad. Dermatol. 39:803–6.
Watson, J. D.; Crick, F. H. C. 1953. Molecular structure of nucleic acids.
A structure for deoxyribose nucleic acid. Nature 171:737–8.
Welch, W. J.; Peng, B.; Takeuchi, K.; Abe, K.; Wilcox, C. S. 1997. Salt
loading enhances rat renal TxA 2/PGH2 receptor expression and TGF
response to U-46,619. Am. J. Physiol. 273 (Renal Physiol. 42): F976–
White, J. V. 1988. Graphic Design for the Electronic Age. The Manual for
Traditional and Desktop Publishing, p. 201. New York: Watson–Guptill.
Woods, J. 1967. Oral contraceptives and hypertension. The Lancet 2:653–4.
Xu, ZhaoRan; Nicolson, D. H. 1992. Don’t abbreviate Chinese names.
Taxon 41:499–504.
Literature cited
commonly accepted, 111
in abstracts, 60, 111
in article title, 51
in tables, 47
of journal names, 89
abbreviation, 60
for conference, 121
structured, 59
working, 58
Acceptance letter, 135
examples of, 79
financial support, 80
Acta Chirurgica Belgica, 66
Active voice, 8
Addendum, 138
Age groups in MEDLINE, 13
Allen, Woody, 15
Altman, Douglas G., 31, 71, 117
and/or, 12
Anonymous, 87
in figures, 33
in tables, 47
Asymmetrically distributed, 115
contributers list, 55
criteria for authorship, 54
mailing address, 121
order of listing, 55
Axes in graphs, 21, 22, 45
Baseline characteristics, 71
Bias, 69, 71
Blinding, 66
Block capitals, 129
BMJ, 55, 61, 67, 78, 120
Bonferroni method, 116
Box-and-whisker plot, 33
Box plot, 33
Capitalization in title, 126
Capitals, 123, 129
Cardiovascular Research, 132
Case report, 103
Categorical data, 26
Chaplin, Charles (Charlie), 5
Charts, 24
Chinese names, 86
Churchill, Winston, 1
How to Write and Illustrate a Scientific Paper
Circular graph. See Pie chart
Claiming priority, 77
Clemens, Samuel L. – a.k.a. – Twain,
Mark, 128
Collins, Rory, 117
Column chart, 24
Conclusion, in discussion section, 69
Conference abstract, 62, 121
Confidence interval, 72, 116
Consent for publication of
photographs, 142
Contraception, 6, 10, 58
Contributors’ list, 55, 56
Controlled trial. See Clinical trial
Copyright, 145
Correction marks, 139
Council of Biology Editors’ Style
Manual Committee, 107
Credit line, 146
Crichton, Michael, 61
Crick, Francis H. C., 8, 130
Curves in graphs, 20, 45
Data-point symbol, 21
Decimal point, 106
Diacritical mark, 89
Disclosure. See Conflict of interest
Dissertation, 91
Dixon, Bernard, 8
Doctoral thesis, 91
Double-blind, 66
Double spacing, 47, 125, 128
Drawings, 38
Dropouts, 69
Duplicate (dual) publication, 143
Ehrenberg, A. S. S., 109
Electronic style, 90
Elements of Style, The, 6
English as a foreign language, 3
Enumeration, 110
Enzymologia, 136
et al., 84
Ethics, 56, 85, 142
Exclusion criteria, 64
Faculty of 1000 Biology, 19
Faculty of 1000 Medicine, 19
Figures. See Graphs
First person, usage, 8
Flow charts, 69
Follow-up, 70
Footnotes, 47
Gender, 7
Gibbon, Edward, 3
Golden ratio, 22
Grants, acknowledgement, 80
axis labels, 23, 24, 46
box-and-whisker plot, 33
lines, 20
pie charts, 36
reduction of, 37
Grouped column charts, 25
Halliwell’s Film Guide, 112
Hand-drawn figures, 39
Harvard system, 81, 82, 84
He/she, 7
Headings, running title, 123
Health care, writing on, 130
Helvetica (typeface), 128
Hemingway, Ernest, 17
Heterogeneity, 117
Hyphenation at line ends, 124
Illustration. See Graphs
Impact factor, 18
Incidence, 11
Inclusion criteria, 64
Index Medicus, 3, 89, 122, 126
Individual data, 28
Informed consent, 64
Institute for Scientific Information,
Instructions for Authors, 19, 130
Intention-to-treat analysis, 70
International Committee of Medical
Journal Editors, 54, 143
Interquartile range, 47, 115
JAMA, 51, 55, 64, 149
Journal Citation Reports, 18
Journal of the American Veterinary
Medical Association, 67
Jurassic Park, 61
Kesling, Robert V., 5
Key words, 122
Krebs, Hans, 136
Lancet, The, 4, 19, 24, 51, 55, 56, 68,
72, 119, 135, 142, 148
Lang, Thomas A., 115
Language, 3
active voice, 8
English as a foreign language, 3
he/she, 7
incidence, 11
passive voice, 8
prevalence, 11
sexist, 7
tense, 49
verbosity, 5
Legend. See Figure legend
Length of the manuscript, 130
Levan, Albert, 59
Line graphs, 20–25, 44–45
Loss to follow-up, 70
Manuscript preparation
authorship, 54
constructing outlines, 58
graphs, 20, 44
headings, 123
length of the manuscript, 130
proof reading, 137
working abstract, 58
Material and Methods section.
See Methods
McNab, Sheila, 7
Mean value, 115, 118
Median, 115
Medical ethics, writing on, 130
Medical Research Council (UK), 66
Medical Subject Headings (MeSH),
Medicine and Science in Sports and
Exercise, 25
MEDLINE, 3, 13, 85
blinding, 66
informed consent, 64
number of subjects, 67
randomization, 65
Middle initials, 86
Moberg, Vilhelm, 16
Multiple publication, 143
N. Engl. J. Med., 29, 34, 61, 129
Name-and-year system, 82, 84
Names of journals, 89
Nature, 130, 136
New Scientist, 8
Newsweek, 4, 11
Normally distributed, 115
Notes added in proof, 138
Numbering system, 83
Number Needed to Treat, 118
Number of subjects, 67
decimal point, 106
enumeration, 110
in tables, 43
percentages, 44, 108
quotients of units, 107
How to Write and Illustrate a Scientific Paper
rounding, 44
spelling out, 105
standard units of measure, 112
with several zeros, 107
Nursing, writing on, 130
Obstetrics & Gynecology, 52, 71
Occupational science, writing on,
Original data, saving, 149
Outcome of study, 72
Pagination, 121, 126
Passive voice, 8
Past tense, 9
Patients, photographs of, 142
Patton, George S., 74
Percentages, 44, 108
charges, 146
copyright material, 143
letter requesting, 148
Personal communication, 83
Pie charts, 36
Plagiarism, 143
Popeye, 137
Present tense, 9
Prevalence, 11
Proofreader’s marks, 140, 141
Prostaglandins, 30
Publishing, dual publication, 143
Publishing ethics, 138, 144
P value, 33, 71, 116, 119
Quotation permission, 146
Quotients of units, 107
R andomized allocation, 65
Randomized clinical trial.
See Clinical trial
Reduction of graphs, 37
Referee, 132, 133
References (bibliographical)
“in manuscript,” 133
“in preparation,” 133
“in press,” 83, 133
citing unpublished work, 83
electronic style, 90
et al., 84
Harvard system, 81, 84
in abstract, 60
journal, abbreviation, 89
name-and-year system, 84
names of journals, 89
updating list in proof, 138
Vancouver system, 81
Rejection of papers, 135
Remote verb, 14
Responding to editors, 132
dropouts, 69
loss to follow-up, 70
participant flow, 68
study outcome, 72
Rounding, of numbers, 44
Running title, 52
Sample size, 67
Scanning, 39
Scientific Style and Format, 86
SD. See Standard deviation
Secic, Michelle, 115
SEM. See Standard error of the mean
Sexist language, 7
Side effects, 73
Significance tests, 33, 71, 115, 116,
Simple pie charts, 36
Spacing, text, 123, 125, 128
Spelling out numbers, 105
Standard deviation, 115, 118
Standard error of the mean, 115
Statistical methods, 114
Statistics, 33, 71, 114
Structured abstract, 59
Strunk, William, Jr., 6
subgroup analysis, 117
choosing the journal, 18
Subtitle, 121
Summary data, 29
Symbols, table footnotes, 47
Synonyms, 14
abbreviation in, 47
where to use, 42
Table typing, 47
Table versus graph, 44
Tense, 9, 49, 53
Thalidomide letter, 101
acknowledgments, 97
aims, 95
at a glance, 93
compilation, 91
contributors, 92
cover illustration, 98
definitions, 100
glossary, 100
length, 99
overview, 92
structure, 99
Three-dimensional graph, 34, 36
Time (weekly), 4, 131
Times (typeface), 128
Times, The (London), 128
abbreviations, 51, 111
capitalization, 126
declarative, 48
for thesis, 51
neutral, 48
subtitle, 121
Tjio, Joe Hin, 59
Tufte, Edward R., 31, 36
Twain, Mark, 128
Typing, 47
12-point Times, 128
block letters versus lowercase
letters, 129
double-spacing, 125, 128
length of the manuscript, 130
pagination, 121, 126
up-and-down-style versus down
style, 129
Under preparation, 83
Units of measure, 112
Unpublished observation, 83, 86,
Vancouver (reference) system, 81,
82, 83
Vancouver Document, 19, 47, 130
Verbosity, 5
Watson, James D., 8, 130
Website, 90
Where and when to write, 16
White, E. B., 6
Withdrawal, 69
Word processing, 17
Working abstract, 58
World Wide Web, 90
How to Write and Illustrate a Scientific Paper

Report this document