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!