Tips on Writing Papers

http://chaos.utexas.edu/wp-uploads/2014/07/tips-on-writing.pdf
Harry L. Swinney
University of Texas at Austin
swinney@chaos.utexas.edu

Every scientist wants his or her paper to be read, yet many scientific papers are readable only
by specialists in a narrow area. Even papers in specialty journals should be written so non-
specialists can understand something about the concepts and results. The tips presented here
are based on observations of how readers select papers to read and how they decide whether
or not to continue reading. If a paper is not clear, succinct, and interesting, readers will skip to
the next paper.

Title: A scientist interested in your topic will likely find your paper through a key word
search. Sifting through the many titles that come up in a search, a prospective reader will
decide from your paper’s title whether to take the next step and read the abstract. Hence it is
important to think carefully about the choice of words in the title. Make a list of words that
describe your work, and use that list to form a short title, at most ten words long. Avoid
jargon, clever double meanings, and empty words and phrases such as “new, novel, study of,
investigation of, exploration of, precision, high resolution, efficient, powerful.”

Abstract: A carefully written abstract is crucial because most readers will skip on to the next
paper after reading a few sentences of an abstract. Only a small fraction of the readers of the
abstract will then read the paper online or download a pdf.

An abstract should address the obvious questions: What have you done? How did you do it?
What is new here? Why is interesting? What are the ramifications? Do not assume readers are
familiar with your area of work.

Succinctly describe your main results. Be quantitative, as in “We find that the scaling of the
susceptibility X with temperature difference ΔT is given by X ∝ (ΔT)^α with α =0.52±0.03
rather than the predicted value α =5/4.”

Introduction: Almost no one reads a paper continuously through the successive sections –
Introduction, Background, Methods, Results, Discussion, Conclusions. Many persons read a
paragraph or two of the Introduction and then jump to the Conclusions. Others look first at the
figures before reading any of the text. Keep in mind that you rapidly lose readers throughout
this process.

An Introduction should describe the issues and open questions that motivate your work. How
does your work fit in? How does your work differ from previous related work?

Relevant previous research must be reviewed in the Introduction or, if the topic has an
extensive history, the literature review can be part of a separate “Background” section.
Conduct literature searches using Google Scholar, ISI Web of Science, and www.arxiv.org.
Search using key words and names of authors of key papers. Also use search engines to find
recent papers that have cited the key papers on your topic.

Referencing relevant prior work is not optional. Follow the American Physical Society
Guideline on References in Publications: “Proper and complete referencing is an essential
1
part of any physics research publication. Deliberate omission of a pertinent author or
reference is unethical and unacceptable.” See www.aps.org/policy/statements/02_2.cfm

Figures: After deciding on your main point and writing an abstract, create a rough set of
figures. Clear simple figures communicate more effectively than text because our brains
process images much faster than words. Readers are more likely to remember a figure than
text, so time spent preparing figures is a critical part of the communication process. Years
later you will not regret having spent days preparing a single expressive figure that other
authors later reprint in their papers.

Each figure with its caption should tell a self-contained story. Begin the caption with a short
phrase or sentence stating the point illustrated by the figure, and follow with the details. The
caption “The function f(x) plotted as a function of x” is of no help.

A recommended resource on preparing figures is E.R. Tufte’s The Visual Display of

Methods. A shocking number of papers lack information essential for replication of the
research. This omission is unscientific and an indication of sloppy work that will have no
lasting value. Describe the sample preparation method, measurement and computational
procedures, and data analysis. Give all parameter values; do not forget those parameters that
are not varied. Use consistent units; SI units are usually preferred. State the uncertainties in
the measurements and describe how the uncertainties were determined. For numerical
simulations state the boundary conditions and initial conditions, and give the size of the
temporal and spatial meshes; test the convergence of your results as a function of mesh size,
and describe the convergence test. For theoretical analyses state the assumptions,
approximations, and normalizations, and be clear about changes of variables.

Results: Many readers start by reading the Results or Conclusions, so do not assume that
readers know the definitions and other information given earlier in your paper. Present only
material essential for your story. Most research includes steps that turn out to be irrelevant to
the final result. That effort may have taken many months, so there is a temptation to describe
the missteps and blind alleys in the paper, but don’t unless your misstep is one that others are
likely to make.

The Results should unfold in a logical way. Re-read your Results section multiple times,
culling paragraphs and figures that are not needed for the main message. Avoid the temptation
to describe the story in chronological order; your own random walk toward the solution of a
puzzle will confuse rather than inform readers.

Include only essential figures. Readers become overwhelmed and confused by a large number
of figures, and figures with many curves or many panels.

Discussion: Describe how your results solve a long-standing puzzle or extend prior work. Do
not include additional results in the Discussion.

Conclusions: After the Introduction, the Conclusions section is probably the most read.
Some readers even jump from the Abstract to the Conclusions. Thus do not assume that
readers know definitions given in the body of your text. Summarize your key results without
2
repeating phrases or sentences that have appeared in the abstract or previous text. Mention
key figures. Put the results in the context of past work, and point to new questions that your
work raises.

General Comments

Section Headings. Make generous use of brief headings and subheadings to guide the reader.

Paragraphs. Introduce the main idea of a paragraph in the first sentence, and follow that with
clear and simple supporting sentences that focus on the main idea. Avoid digressions from
the paragraph’s focus. Break long paragraphs into several paragraphs. The closing sentence
should reinforce the point of a paragraph.

Linkages. Facilitate the readability of your paper with sentences that link one paragraph to
the next and one section to the next. These linkages make it easier for a reader to understand
where you have been and where you are going.

Word choice. Write for the broadest possible audience. Avoid jargon, acronyms, and
superlatives, and avoid extraneous phrases such as “it is a fact that…” or “it is well known
that”. Understatement is preferable to hyperbole; avoid “unique, ultra, very, novel, innovative,
perfect, super, remarkable, extraordinary.”

Avoid repetition. Repetition is tiresome. Avoid repeating phrases, sentences, and concepts.
Do not repeat or say again what has been said before. Do not include the same information in
the text and in a figure caption. Repetition is tiresome.

Grammar. Use a standard writing style guide such as Strunk and White’s The Elements of
Style or The Chicago Manual of Style; both are available as free pdf downloads.

Most important tip: re-write, re-write, re-write. Become your own severest critic.
Read the text out loud either alone or with another person to check the flow of sentences and
to find ambiguous or irrelevant statements. Ask colleagues and friends to critique your
manuscript, and volunteer to critique theirs. The most helpful critics are often those unfamiliar
with the subject matter.

Writers tend to become attached to their text and reluctant to make revisions. Listen carefully
to the questions your readers raise, and respond by revising your manuscript rather than
becoming defensive.

Cut material not essential for your story. The longer a paper, the fewer readers it will have.
Delete words, phrases, sentences, paragraphs, figures, tables, and sections that are not
necessary. Lengthy details can go into an appendix or supplementary information.

Finally, give the preparation of a manuscript an effort comparable to that put into the
experiment, computations, and analysis. Working through the logic of the arguments is an
essential part of the creative process and often leads to new insights. A well written paper will
garner the attention that the research deserves, while an ingenious result presented poorly may
go unnoticed.
2005; revised 23 July 2014