Tips for writing
a Scientific ManuSCRIPT
The art
of writing is the art of applying the seat of the pants to the seat of
the chair. — Mary Heaton Vorse
I WHEN
TO STOP DOING EXPERIMENTS AND WRITE 3
II. WRITING
THE MANUSCRIPT 4
The bottom line 4
The results 4
The methods section 5
The discussion 5
The introduction 6
Figures 6
Acknowledgements 7
III. GENERAL TIPS FOR CLEAR WRITING 8
The flow of logic 8
IV. WAYS
TO IMPROVE YOUR WRITING (other than practicing) 9
While in the process of writing 9
In general 10
V. WHERE
TO SEND THE PAPER 10
The pleasures and pitfalls of high-profile
journals 11
What are Nature
and Science looking for? 12
VI. DEALING
WITH EDITORS 13
Points to bear in mind 13
Writing to the editors 13
VII. DEALING WITH COMMENTS FROM REVIEWERS 14
Comments from referees fall into several
classes 14
How (not) to deal with comments 14
PREAMBLE
So,
you’ve invested several years learning to be a scientist, and you finally feel
like a success — you’re ready to write a paper. This should be the easy bit, right?
Wrong. Maybe writing a paper isn’t the hardest
thing you have to learn as a scientist, but it’s one of several skills (writing
grants, managing a lab, being a good boss) that people often find themselves
having to learn on their own. And it’s
a crucial skill. You can be the best
experimentalist in the world, but if you do a poor job of communicating your
results you won’t make as much difference to the course of science as you
should.
Is
this making it worse, are you now even more unwilling to sit down and
write? Does the bench suddenly look
amazingly friendly and inviting by comparison?
In emphasizing the importance and difficulty of writing a good paper, we
are not trying to increase your anxiety level, but to make it clear that this
skill, like any other, needs to be learnt.
Don’t expect to be able to write your first paper in one sitting just
because your PI can. She’s had a lot of
practice. Do understand that this skill
is a basic professional requirement — there is no way out of writing papers in
the long term if you want to be a scientist.
Take
a deep breath and apply your brain to the problem. You can do it.
Very few people find paper-writing easy first time out; graduate
students often get to the stage of writing a paper years after the last time
they had to produce a serious bit of writing, and a paper is very different
from anything you’ve ever had to write before.
So it’s not surprising that it’s hard.
Nearly everyone gets dramatically better with practice.
Writing,
like interpretation of results, is hard to teach except by example. We hope
that your lab head takes his or her responsibility to help you learn
seriously. But because writing skills
are passed on informally, it may be hard for your advisor to articulate exactly
what they think you should change about your style. To provide some additional help and a different angle on the
problem, we prepared the following document (originally for a course on the
Practice of Science at UCSF). Enough
people found it helpful that we decided to distribute it more broadly.
In
reading this document, consider the source: one of us is an ex-editor of Science, the other is an ex-editor of Nature.
This may bias our perspective on occasion, but we’ve done our best.
This document is intended to be freely
shared, in the same spirit as the
ÒCopyleftÓ standard publicized by the GNU free software project (see small
print below). If you find it helpful,
pass it on. If you want to suggest
additions or changes, e-mail Rebecca Ward (becky@hms.harvard.edu).
Good
luck!
Kelly LaMarco
Rebecca Ward
Copyright (c) 2003 Kelly LaMarco
and Rebecca Ward. Permission is granted to copy, distribute and/or modify this
document under the terms of the GNU Free Documentation License, Version 1.2 or
any later version published by the Free Software Foundation; with no Invariant
Sections, no Front-Cover Texts, and no Back-Cover Texts. The GNU Free Documentation License can be
found on line at http://www.gnu.org/copyleft/gpl.html, or obtained by writing
to to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA.
I
WHEN TO STOP DOING EXPERIMENTS AND WRITE
Everyone
who wants to write must learn that something may in itself be the finest piece
of writing one has ever done, and yet have absolutely no place in the
manuscript one hopes to publish. -Thomas Wolfe
+ When do you have what it takes to make a
paper? The only way to answer this is to try to draft the paper. This often exposes holes in the argument
that cause you to go back and do more experiments. Thus drafting is a useful exercise as soon as you know where
you're going.
+ To draft a paper, simply work out what
the figures and tables would look like. Write a sentence about what each of the figures and tables is
intended to convey. When the sentences
look as if they tell a story, it's time to start writing.
+ Deciding what to leave out can be as
important as deciding what to include.
Fascinating little bits of information that are off the point always
cause trouble. At best you have to
revise the paper to leave them out, at worst criticisms of the incidentals can
be used to reject the paper even though the main points are unassailable.
+ Getting started is the hardest
thing. Most people put off starting to
write until the last possible moment.
There is something about a blank page that causes the mind to go just as
blank. Train yourself out of this
terrified-rabbit syndrome by writing early and often. Sketch out the paper in a rough way as soon as you can see what
it might say. Most people find it
easier to edit something, however close to garbage it is, than to start from
scratch.
II. WRITING THE MANUSCRIPT
To write simply is as difficult as to be good.
-W. Somerset Maugham
The bottom line
First
things first: you're writing a paper because you have something important to
tell the scientific community. What is it? Before you lay pen to paper (or
finger to keyboard), decide on the one thing you want to get across in this
paper. It's your bottom line, and it
should come out clearly in the manuscript.
The
introduction should be written with a view to setting up the background for
what we are about to learn (the bottom line) and why it matters.
Repeat
the bottom line over and over again — at the end of the abstract, in the
introduction, in the results, and in the discussion.
The results
Once
you have figured out what you want to say, it's best to lay out the figures
that you need to make your point in a sequence that tells the story. Any result that isn't relevant to the bottom
line should probably be deleted (if the result seems really important to you
yet is irrelevant to the bottom line, you've probably got the bottom line
wrong).
It
often works well to let the story unfold in the way it actually happened. Write down why you did the experiment and
what the conclusions are. It helps to
train your students to do this in their lab notebooks as they go along. It may seem impossible that you will ever
forget why you did an experiment or the conclusions you drew from it, but when
a series of experiments stretches over many years it's quite likely to happen.
In industrial settings, researchers are often absolutely required to document
in their notebooks why an experiment was performed and what was concluded from
it, because of issues to do with patents and fraud. If your students are likely to move into industry, you'll be
doing them a favor by training them to do this early.
Unless
your experiments were done for reasons that turned out to be irrelevant to what
you actually found out, simply writing down what you did and why will be a good
first draft of your results section.
Aside: Teaching your students to explain in
their notebook why they did the experiment and what they concluded from it has
a couple of advantages. It makes writing papers easier, as noted above, and it
therefore makes the prospect of giving the student "first crack" at
writing the manuscript more appealing.
We feel that training a student to write a paper is an integral part of
the job of the lab head, and that mentors who don't allow students to
participate in drafting a manuscript do a great disservice to that student (and
to the scientific community in general). Ideally, the lab head should allow the
student to compose a draft, edit it, and then discuss the changes with
them. Their draft will only be
impossible to use if the organization is completely wrong — in which case, the
problem is one of logic, not writing ability.
The methods section
After
completing the results, list what methods were used to generate the results,
then write down what you did to carry out the experiments. Simple.
But if you're re-using a methods section from elsewhere (like a student
thesis) don't just cut and paste, check that all the methods were indeed used
in this paper. (A surprising number of
people fail to do this). And remember
to define abbreviations that your audience might find unfamiliar (your pet name
for the buffer you used, for example).
The discussion
What
to include in the discussion is, in part, a matter of taste (and of where
you're sending the paper). We think that the discussion should do more than
merely reiterate the results. It's your
chance to put your findings in perspective, propose a model, outline a
direction of investigation, and make the reader think. But be aware that if you go too far from
what you've actually shown in the paper in the discussion, the referees will
challenge you. Make sure that you
clearly distinguish between what you've shown and what you imagine. (After all, you could be wrong.)
Explain
the reasoning that led you to a particular view. If you find yourself giving very long, detailed explanations
about matters that aren't truly relevant to your paper, ask yourself whether
it's really important to make that particular point in this particular
paper. In general, discussions of
interesting side issues should be minimized. Often they detract from the main
message.
The introduction
We've
left the introduction until last because many people prefer to write it last,
when they're clear about exactly what they have to introduce (like characters
in a novel, a paper may turn out to have its own ideas about what it wants to
say).
The
introduction should cover the aspects of the field that raised the question
that you addressed in this series of experiments. In an ideal world, you really did know about all these points
before you started, and the experiments really were designed to elicit the
answer you got. More often, the
background evolved for you as you did the work. This doesn't matter as long as you can give a coherent reason for
thinking that your question is interesting, and for having believed that the
experiments you did would address it.
To end the introduction, briefly summarize what the reader is going to
learn and why it is important.
Figures
Next to torture, art persuades best.
-Ward Beecher
Each
figure should have a clear point/purpose. Describe the point succinctly in the
figure legend; usually you can use this as the first line of the legend. Make
sure the figure is clearly labeled and symbols are defined in the legend.
Avoid
complicated figures if you can.
Consider whether different ways of presenting the data would serve the
same purpose better — do you really need to show twenty binding curves, or
would a table of 20 Kd's
do just as well?
When
constructing a figure with multiple parts, give some thought to the pattern the
reader's eye will follow. Showing a
sequence of events that runs left to right then right to left & back again
is usually not a good idea.
If
you have intricate color photographs, consider the quality of the photos in the
journal you choose. For example, Development
does lovely color photos on high quality paper (and sometimes can be persuaded
not to charge for them), whereas Nature and Science use lower quality paper (not
great for color photos).
Acknowledgements
It
is much worse to miss out someone who deserves to be acknowledged than to
include someone who doesn't deserve it.
Err on the side of inclusion.
III. GENERAL TIPS FOR CLEAR WRITING
I see only one rule: to be clear. If
I am not clear, then my entire world
crumbles into nothing. -Stendhal
The flow of logic
You
have a story to tell, and its logic is clear to you. The question is, how do you make that logic clear to the
reader?
1.
Help the reader all you can with signposts. For example, it sometimes helps to
start the paragraph with a question that indicates where you're going with the
argument that follows. Subheadings can
be very useful, if the journal allows them.
2.
As well as the paper having a 'bottom line', each paragraph should have
one. What are you trying to explain to
the reader in this paragraph? What
should they have learnt by the time they've finished reading it? It is because of the need for a paragraph to
have a bottom line that many teachers advocate outlining an essay before you
write it. For some people, it does work
well to list the points to be made in the order in which they should be made,
then expand each point to a paragraph. For others, it's not necessary or
helpful. But try it before you decide
that you're one of the latter group.
3.
Treat each paragraph as a thought.
There is a point to each paragraph (the bottom line). Starting a new paragraph indicates a new
thought. It should be clear how the new
thought follows from the old thought.
There should be a clear link (or transition) between the end of one
paragraph and the beginning of the next, and the successive bottom lines of the
paragraphs should follow a logical order. If you didn't originally outline your
paper, summarize the finished paper in the form of an outline to check that the
ideas in your paragraphs follow a logical progression.
4. Sentences within a paragraph need to be
connected by an obvious flow of ideas.
Keep sentences fairly short and to the point. Trying to get too many ideas into one sentence (or one paragraph)
will make it hard for the reader (or reviewer) to decipher the meaning. Remember, people are busy and few of your
readers will understand your system as well as you do. Points that you think are blindingly obvious
may well not be.
5. Check that phrases within the same sentence
connect with each other. When you use a pronoun like 'it' or 'they', check that
it's clear what the pronoun refers to.
For example, "there are several differences between microtubules
and actin filaments; first, they are larger" leaves the reader in
confusion over whether "they" refers to the microtubules or the
filaments.
6. If you provide several lines of evidence
that all tend towards the same conclusion, don't simply say "1 is true, 2
is true, 3 is true. The conclusion
is... " Help the readers to
understand why you're giving them all these facts from the start, by pointing
out how all the lines of evidence support each other: "1 is true,
suggesting this conclusion. Similarly,
2 is true, and furthermore 3. Thus it
seems clear that.... "
7. Don't be afraid to say what you think is
going on. But don't claim that it's
proven if it's not. "An obvious
explanation is ... but many other explanations are possible" or "a
plausible explanation is that..." are two ways to show the reader how
you're making sense of the data without misleading them into thinking you
believe you've proved it conclusively.
IV. WAYS TO IMPROVE YOUR WRITING
(other than practicing)
While in the process of writing
1.
Most people explain things better when they're talking than when they're
writing. In part this is because many
people seem to think that scientific writing requires you to use complicated
multi-syllable words and the passive voice.
Try writing the way you'd talk first.
Dictate into a tape-recorder if that helps. Most journals don't insist on the passive voice, and none of them
require you to use five syllables where one will do (utilization instead of
use, for example).
2.
When you're writing your first draft don't worry about minor issues like using
the same word over and over, ending a sentence with a preposition, splitting
infinitives. Worry about organization
of the manuscript, and how best to ensure it makes sense. You can always tidy up later if you need to.
Most of these rules are not worth worrying about in any case, and worrying
about them too early tends to lead to a loss of clarity.
3. Get an intelligent colleague who is as far
from your field as possible to read the draft and mark all the places where
they get lost, or suggest new ways to convey concepts they find difficult. (Outside advice also helps when deciding
where to send your paper. It can be really difficult to judge the "general
interest" or significance of one's own work.)
4. Put the finished paper aside for a week and
do something else. Then read it
again. You'll be surprised how much
easier it is to spot the parts that are difficult to understand when you've been
thinking about something else for a while.
In general
1.
When you read papers in fields other than your own, take note of which ones
leave you feeling positive. Why? Did the authors explain why they were
interested in the question in a clear and convincing way? Were the experiments logically
presented? Was it easy to see why they
reached the conclusions they did? Could
you understand the gist of the figure even before reading the legend? Which stylistic points made it easier for
you to understand, and therefore learn something new? In other words, take note of what worked for you and try to copy
it.
2.
Think about oral presentations that you enjoyed and why you liked them. Some oral presentation styles also work well
in written form. For example, often speakers will start by showing a summary of
the points they are going to make so that you know where they are going as the
story unfolds. (Presenting your work to
other members of your department can also help organize your thoughts before
you write).
3. Buy a book like Strunk and White's
"Elements of Style" and read it from cover to cover. It's not about science writing, but nearly
all of it applies.
V.
WHERE TO SEND THE PAPER
This is a decision you need to make
early on, unless there are several journals in your field that all use the same
format and have similar length requirements.
In most cases it's not too difficult to narrow the choice down to a
couple of journals; think about who you want to see it and why, who is likely
to be interested in it, and which journals publish papers on similar subjects,
at similar levels of interest. Usually
you're looking for a quality journal with the right kind of audience that is
not too painfully slow and whose editors and referees seem to make reasonably
sensible decisions. A couple of issues
to consider here are how easy Web access to the journal isSometimes you're just
looking for a journal that's respectable and won't give you a hard time if all
you want to do is archive something that's not particularly interesting. The
decision is hardest when you have what seems to you to be a really exciting
paper. Should you try one of the
'trendy' journals?
The pleasures and pitfalls of
high-profile journals
Publishing a paper in a high-profile
journal can do you a lot of good, both in the short term (all the conference
chairs looking for someone to fill up their program will see your paper) and in
the long term (when you apply for a job or come up for tenure it doesn't
exactly hurt to have a Nature paper
on your resume). But sending a paper to
Science or Nature has its disadvantages too:
+ There is a huge element of luck in
getting a paper accepted by a journal like Nature
or Science. Both of those journals are selecting about ten papers a week from
over a hundred submissions. Even though
up to 70 of those 100 submissions may fall short of the journal's criteria for
interest or technical quality, there is still some level of subjectivity in
which of the remaining 30 are selected.
Sometimes a referee will show an inadequate level of enthusiasm, while
giving no real criticism of the data; that may just be because the referee had
run out of Prozac, but your paper's rejected just the same.
+ Some papers are very hard to write in a Nature/Science short format. If you really need to show six or more
figures to make your story convincing, or if you need more than about 2,000
words to say what you absolutely have to say, you should probably give in
gracefully and go elsewhere.
+ Since most papers submitted to Nature or Science are rejected, deciding to submit a paper there will, more
often than not, cost you time — usually at least 1–2 months. Half the papers are rejected after 2 weeks,
and you then have to rewrite in another format (another 2 weeks). The rest of the papers that are rejected are
refereed first, which usually takes an extra 3-5 weeks.
If
you submit to one of the trendy journals, do so in the full knowledge that your
paper is quite likely to be rejected even if it's good, and you will then have
to rewrite and submit elsewhere. Don't get suicidal when this happens. Nobel-prize-winning research has been
rejected by these journals, and most Nature-quality
papers aren't published in Nature. If the paper's good, and published in a good
journal, it'll be influential no matter what. But undoubtedly you'll want to run
the gauntlet at some point in your career, so:
What are Nature
and Science looking for?
It's easier to define what turns them
off. Describing your results as
additional confirmation for a well-accepted theory is a certain route to
rejection, for example. The editors are
primarily looking for papers that fall into one of the following categories:
+ an advance whose implications cross
disciplines (i.e., a result of "general interest". For example, the
discovery of cyclins and cyclin-dependent kinases was interesting to
researchers in many fields, including cell cycle regulation, DNA replication,
transcriptional regulation, tumor suppressors and oncogenes, and so on);
+ a result that connects two previously
unrelated areas of research;
+ a really big advance in a trendy field,
even if it isn't all that relevant to anyone outside the field; or
+ supremely important methods (PCR,
differential display).
Other criteria constantly change. For example, the editors will periodically
decide that a particular subject is under-represented and accept papers in this
area to show an interest. Or papers
that use a pioneering technique like gene knockouts may be given an easy time
for a while, until the technique is no longer perceived as pioneering.
VI. DEALING WITH EDITORS
Points to bear in mind
Academic
editors and professional editors are different beasts. Some journals that use professional editors
are Cell, Current Biology, Nature and
Science. These are also the journals that attempt to select papers that
have a poorly-defined characteristic called "general interest" (see
above), measured by the effect of the paper on the jaded palate of the
professional editor. If you want to
publish in one of these journals, make sure that the elements that make your
paper "spicy" are clearly obvious.
Professional
editors will rarely be experts in your field.
This has the disadvantage that they will rarely be able to make their
own judgements about the quality of your work, but the advantage that they are
not often so involved in a field as to take sides on a controversial
issue. Academic editors will be
knowledgeable about your field much more often than professional editors, and
will more often be able to supplement the referees' criticisms with their own
(or dismiss a referee's criticisms as irrelevant).
No
matter who the editor is, he or she is trying to do the best they can to select
the right papers for their journal and to ensure that the papers they publish
are the highest quality possible. There
may be other things going on, but this is the primary focus for any
editor. Don't assume that an editor is
being political in rejecting your paper without good reason. If you think they've made a bad decision,
tell them so.
Writing to the editors
When
you submit a paper, or resubmit a revised paper, make the editor's life easy by
including a cover letter. This should
concisely explain the major conclusions of the paper, as well as why and to
what fields they are important. This is
particularly important for "trendy" journals.
It
is often helpful to list possible reviewers.
But it is more important to list people that you absolutely, positively
do not want to review your paper. Be
reasonable. It is not acceptable to rule out all of the major players in your
field. (I once had an author request
that "no one from Boston, San Francisco, or San Diego" review his
paper!) Also let the editor know if you
have concrete information about competition.
The
cover letter for a revised manuscript MUST include a point-by point rebuttal of
the points raised by the referees. If
you really want the editors to like
you, send them a marked-up copy of the new version showing which bits have
changed.
VII. DEALING WITH COMMENTS FROM REVIEWERS
Comments from referees fall into
several classes
Most comments can be categorized as one
of the following:
1) valid criticisms that are easy to
address
2) valid criticisms that are hard to
address
3) invalid criticisms that you can
easily show to be invalid
4) invalid criticisms that seem valid
unless you know an awful lot about the subject
5) matters of opinion, or
6) deliberate attempts to delay the
paper for no good reason (fortunately rare)
How (not) to deal with comments
Obvious pitfalls in dealing with the
comment types listed above include:
a) not addressing the easy valid
criticisms because you've put too much work into this paper already. If it
improves the paper and it's easy, just do it.
That's what refereeing is all about.
b) rubbing the referee's nose in how
stupid the invalid criticisms were. If a referee, carefully chosen to be expert
in your field, had a problem with your paper, then 99% of the rest of the world
will as well. Try to see where the
misunderstanding came from, and be thankful for the opportunity to fix it
before prime time.
c) complaining about the incompetence of
the referee instead of addressing his/her comments
d) mistaking comment types 3–5 for
comment type 6
e) failing to realise that there really
can be two opinions regarding the interpretation of your data.
The
case of the valid criticism that will take a lot of work to fix is perhaps the
hardest to deal with. Do you go away
and do the work, taking a year and perhaps being scooped in the process, go to
a less good journal (and perhaps run into the same problem), or try to persuade
the editor and referee to let you handwave your way out of it? One argument you can use in this situation
is that one paper doesn't have to solve the whole problem.
When
responding in writing to a reviewer's comments, do respond to every point (even
those you think are ridiculous or incorrect).
List these points, along with the changes you've made to address them
(or the reasons you haven't addressed them), in a letter to the editor. Point out which pages in the manuscript
contain the changes you've made in response to the referees.
In
all dealings with reviewers and editors, you will do better than average if you
work under the assumption that all editors, and all referees, are serious,
conscientious people who are doing their best for science and for the
scientific community — even if the evidence appears to indicate otherwise! Be polite, however provoked. Since editors and referees are conscientious
people, they will probably forgive your rudeness, but why take the chance? Don't fire off a furious e-mail the moment
that you receive the rejection letter. By all means write it and get it out of
your system, but then throw it away and write a more measured response.
Good luck, and happy writing!