Writing a paper must be easy compared with compiling a thesis. It would be like scribing a few brief pages versus hundreds. That's certainly what I thought as a naïve first-time science writer. Having now done both, I'd probably agree with my initial assessment--just barely. A good paper can take a very long time to write. What's more, once written, it can take an even longer time to reach the printed page ...but more on that in my next article.

Many people suggest making each chapter of your thesis read like a scientific paper. This is good advice. But what if, like me, you are faced with tackling your first "first-author" paper well in advance of starting your thesis write-up? If you are lost in the morass of mental anguish that can be paper writing, this article may help you carve a way through.

Long before you ever start writing a paper, you need to figure out what you think it's going to contain. With a bit of luck, you drew up a master list of potential results that would help a skeptical world believe in your hypothesis, even before you'd done most of the experiments. Now's the time to decide the main message you want your paper to get across, and which of your results support it best.

List your selected results in case, to your horror, you find that you don't yet have them all in, or that the quality of your data, images, or statistical analyses is not up to snuff. Then review the boundaries of your paper, but don't spend too long pondering over what to include and what to leave out.

Whatever you ultimately decide, you can pretty much expect the paper's editor and referees to wield the scientific axe, or ask for something extra to be included. I am still amazed by just how few of my amassed results made it into my first couple of papers. So be very selective--only relevant stuff gets in. The fact that you almost killed yourself getting a particular result doesn't qualify it automatically for inclusion.

Before starting to write, I prepare all the figures and tables. These are, after all, the crux of the paper; without them in front of you, you can't really expect to write a thing. Get as much of the incidental text as possible out of the way in the legends. This way, your manuscript will read much more easily.

Remind yourself constantly that writing your results section simply means describing what your figures and tables show, nothing more. Assume almost nothing, explain almost everything. Remember, you are the expert. No one knows as much about your work as you! Write in short simple sentences that someone new to English could understand.

It's up to you whether you write the introduction or discussion first. I prefer doing the introduction first, as it helps me to focus on where my research fits in the grand scheme of things. I also find it extremely useful to keep rereading my introduction when writing the discussion. This helps me avoid leaving gaps in the story or, worse still, contradicting myself. This is not your end-of-year report; your introduction should be terse and to the point. Limit yourself to the literature that is strictly relevant.

Most other scientists will be far more interested in what they think your results mean than in what you think they mean. However, your discussion is a very important part of the paper. It's your chance to argue in favour of your results. If you really believe that what you've written is accurate (if not, I wouldn't want to be in your shoes), you'll want your readers to accept your conclusions as well as your results.

I need to really watch myself when I'm writing discussions. I find it so easy to overstate my conclusions and make unconscious leaps of faith. After stinging criticism from my boss, I've learned that the discussion is not the place to get excited. Sobriety and open-mindedness must be the order of the day. Try to discuss all reasonable explanations and, if in any doubt, err on the side of playing down your conclusions and letting your results speak for themselves. Never forget that in the first instance, you are writing for your editor and referees. These people don't suffer fools gladly.

Once you've managed to write your first draft, you've overcome the hardest part. After all, the remainder of the process is about making the changes that other people suggest (or tell you to do). Assuming you've got your facts straight, the hard graft of thinking from first principles should now be over. Without delay, forward your offering to your boss and encourage him or her to plaster your electronic draft in red font. If you are as inexperienced as I was when I wrote my first paper you'll badly need their critical input. My boss made comments such as "You can't state that; it's just too controversial and anyway isn't really supported by your results". To which a typical reply from me would be "Oh, isn't it? I thought it was!"

Another classic from my boss, all too often repeated, was "Do you really mean: 'this result clearly shows that...', or merely 'this result may suggest that ...' ".

After you've processed the great swathes of red typeface from your boss, and sought approval on the next draft of your manuscript, next in line come your co-authors. Naturally, they must all have the chance to read and make comments on the manuscript before it is submitted. Even minor authors can pick out irregularities that the bigwigs don't spot. Handling distant co-authors can be a lot harder than popping into the next office. Along with my boss, my transatlantic co-author and I spent a very long and exhausting time trafficking half-written drafts between us in what seemed like an eternal triangle. Just exchanging intact figures (as large e-mail attachments or by FTP) added an extra dimension of difficulty. What's more this person actually was a co-author--one who wrote a small chunk of the paper as well as approving my sections.

Finally, before you submit, get as many well-informed people as you can to read the thing. I also find it useful to pick one completely uninformed friend who can usually only understand three words out of every five. Uncluttered by the incomprehensible science, this individual often can spot the glaringly obvious error.

I have to say that writing my thesis was a whole lot easier after having written a paper first. I had already learned to accept the idea that what existed as a whole section in my head might end up as a single sentence on the page. What's more, I had learned just how powerful a communication tool a carefully crafted sentence can be.