How to write a paper

Notes about writing a paper

Below are tips and steps to help you write a paper. As you prepare, start asking yourself questions such as:

What has been done previously?

You should have done a thorough literature review at the beginning, but now is the time to do it again. They are almost always papers that you will miss when you start a project, so now is your chance to find them.

What is it that is new in your work?

Think about what you have done in comparison to previous work. Are the results different? Improved? Contradictory? Are the methods new or better? Etc.

What is the main purpose of your research?

This might sound obvious, but take some time to think about your results, what they mean, and why people will think that they are important. By knowing the main point that your paper should get across, you will more easily be able to come up with a title, an abstract, and figure out the flow of your paper.

What data do I have to support these claims?

You have taken a lot of data, but as you start writing, you will often find that you need to run a few more experiments. Play “devil’s advocate,” that is, try to find any flaws or objections that people might have to your argument.

Now let’s start organizing the paper. I like to proceed as follows.

Step 1. Start with the figures.

The figures are one of most important parts of the paper. When people are deciding whether or not to invest time in reading your paper, they will probably do the following. First, they will ask, “Is the title interesting and relevant enough for me to download the paper?” Second, they will probably skim through the abstract to find your key results. Third, they will scroll through the paper to find your figures. The figures must be clear (as should be the captions!) so that nearly all of the main points can be found and understood from the figures and captions. Then, if they get excited about the paper, they are likely to read it and cite it!

Start by briefly sketching on paper the figures that you think you need for the paper. At this point you may decide that you need more data; this usually happens. You can always start making figures, go take more data, and then get back to the figures.

First, let’s talk about the content of the figures. Everything should be clearly labeled. Here’s an example of what not to do (on the left) and of what to do (on the right).


There are a number of issues here. First, on the left it is very difficult to determine which data is which from the legend. You don’t want the reader to have to struggle with anything. Is the efficiency increasing or decreasing with R? It is not obvious unless you study the figure carefully. If you have to use a legend, make sure that the order that the data is listed makes sense. You can use color (and it often helps), but for someone that is colorblind, red and green will be difficult to distinguish. Sometimes you can use color as well as different line thicknesses or dashing to make it clearer. Think also about what it would look like if printed in black and white because that is the way that some people will print it. Consider the improved figure on the right (no legend is even needed!). Note how the figure is simple, clean, and easier to understand (of course if you are not working in the field or reading the caption, you may have no idea what the axes are or what R is, but that’s ok). As a personal preference, I also like to put the tick marks on the inside of the axes and to enclose the plot with axes on the top, bottom, left, and right (if using Igor Pro (see below) then this can be done using the “Mirror: on” setting in the axes tab when modifying the plot).

Captions should be descriptive and tell you what you should get out of the figure. See the following example.


Bad Figure Caption: Force vs. distance plot of data.

Better Figure Caption: Attractive and repulsive Casimir–Lifshitz force measurements. Blue (orange) circles represent the average of 50 data sets for the force between a gold sphere and a silica (gold) plate in bromobenzene. For clarity, error bars, which represent the standard deviation of the data, are only shown for seven data points.

Note that the first phase of the caption tells you the main message you should get, i.e. that we have measurements of attractive and repulsive forces. The rest gives you more details about the figure.

Making graphs: I use Igor Pro, and I recommend that you use it too. We have a copy from the AFM that can be used (check out the tutorials to learn how to use it). Igor Pro can also be used for scripting code, curve fitting, data analysis, etc. (like Matlab), but it has more plotting options. Plus, you can send me the file so that it is easy for me to modify if needed.

Making Figures: To combine multiple graphs or to make images, I use PowerPoint. You can copy the Igor Pro graph and paste it into PowerPoint, where you can add additional text and images. Then you can save the PowerPoint image as a PDF for importing into MSword (or LaTex). Look to specific journals to determine the appropriate fonts within the figure.

At this point, should make a PowerPoint document that has different versions of your figures and other figures that you may or may not use. This is a good time to make extra figures for presentations or other backup material that supports your data but that you might not include in the paper.

Step 2: First draft of title and abstract

Now you are ready to come up with your title and abstract. Your title should clearly state the main idea of the paper, and the abstract is a mini summary of the problem and how you solved it.  Nature Magazine is very particular about how to write a summary abstract, and I like their suggestions. In most other journals, you will not include the actual references in the abstract; however, the format is similar. Also, the exact word limits will vary depending on the journal.

How to construct a Nature summary paragraph

Annotated example taken from Nature 435, 114-118 (5 May 2005).



Step 3: Writing the body of the paper

First, find about 3-5 really good papers in the journal you want to submit. Analyze these papers carefully. What is the structure? How many paragraphs of intro/background? When is the objective stated? Is there a particular formula for writing papers in that journal?

Depending on the journal, there may or may not be separate sections of the paper. You will usually have a short introduction to the field and the problem, and then you will state briefly what you have done to solve the problem. The figures will guide you through the discussion, and then you will reach your conclusions.

Rather than just writing, it is best to first make a skeleton outline for the paper. What this means is that you first write down a phrase or sentence that describe the paragraph you will later write. Do this for the whole paper! Then talk to me, and we’ll make sure you are going down the right path before writing the whole paper.


The paper will be made up of many paragraphs, and the structure of the paragraphs will almost always be similar. The first sentence of the paragraph will tell you the main idea or message that you should get out of the paragraph, and the following sentences will simply give further details and explanations. Someone should be able to read the first sentence of each paragraph and get all of the main ideas from the paper. Try to do this on your paper! 

Step 4: Revise title and abstract

By the time you finish writing, you will probably find that you need to modify your title and abstract slightly, as the writing process helps focus in on the key points that you want to make.


Step 5: Almost done…

Once you’ve read through your final draft a few times, have one of your labmates take a look. A second (or third) set of eyes will always catch things that you’ve missed. Then it’s time to send the draft to me!


Other notes:

-In addition to the figures that you make for the paper, you should also have a PowerPoint file with related backup figures, i.e. other data that supports the conclusions but might not be included in paper. It’s also good also for talks and your thesis.

-Make figures in Igor unless you really need something else (e.g. contour plots are often easier in Matlab).

-Use EndNote for references when using Word. This will help keep things organized.

-Make sure you define all variables and constants throughout the text.

Other notes:

Here’s a nice virtual issue from ACS called “Mastering the Art of Scientific Publication,” which discusses important things to consider when writing a paper: from cover letters and titles to graphs.