Writing articles for scholarly journals is the primary means for scientists to communicate the results of their research to other scientists.  And journal articles in large part form the basis of how academic research scientists are evaluated, funded and compensated.  Journals also archive for all time the work you have completed and the knowledge you have produced.

Scientific papers should be written to be exceedingly clear and concise.  Many journals have “page charges”, i.e., there is a per-page fee to publish the article, so there is a clear disincentive for using more words than necessary.  A well-written scientific paper should put the reported results in context, and give the motivation for doing the particular work you are reporting.  It should give the experimental design or methods, even for theoretical and computational research, and how the work was done.  The results should be reported clearly using data displays (tables, charts, figures, graphs, images) that allow the reader to see in the data what the authors see, and certainly what is asserted in the text of the article.  The results should be discussed in the context of previous work.  This discussion should also explicitly state what the reader should understand about the current results, as well as what should be the next step in this line of research.

Grammar is critical in scientific writing.  In fact, scientists try to be so concise that their English should be better than that of workers in other disciplines!

Scientific papers can be broken down to basic units: Title, Authors and Affiliation, Abstract, Introduction, Methods, Results, Discussion, Acknowledgments, and References. These units combine to make a standard format that is used in almost all fields and in all journals.  This format should be adhered to rigidly.  The standard format breeds efficiency and uniformity.  But more important, this standard format allows the paper to be read at several different levels.

For example, many people skim titles to find out what information is available on a subject.  Others may read only titles and abstracts.  Those wanting to go deeper may look at the tables and figures in the results, and so on.  The take home point here is that the scientific format helps to insure that at whatever level a person reads your paper (beyond title skimming), they will likely get the key results and conclusions.  Also journals are indexed in any number of online databases, e.g., Scitation, IOPScience, Google Scholar, Web of Knowledge.  These indexes depend on the standard format, and failure to use it would make your article invisible to online searches.

Examining the sections in detail, the Title should succinctly describe the content of the paper.  The title should contain descriptive words that associate strongly with the content (system, methods, results, conclusions) of your paper.   A majority of readers will find your paper via electronic database searches and those search engines key on words found in the title.

Generally the person most responsible for the work is listed as the first author, while the research mentor, lab director, or supervising professor will be the last listed author.  Most people read major currency into the list order of authors, and being first author is the probably the most important position.

The Abstract summarizes in one paragraph the major aspects of the entire paper.  The Abstract helps readers decide whether they want to read the rest of the paper, or it may be the only part they can obtain via electronic literature searches or in published abstracts.  Save for the results themselves, and maybe the title, the Abstract is the most important part of the paper.  It is what either draws people in to read the details, or to make them skip over it entirely.  Therefore, enough key information (e.g., summary results, observations, trends, etc.) must be included to make the Abstract useful to someone who may to reference your work.

A good strategy for the abstract is to pull the key sentences from the other sections to make the abstract.  In the first sentence the abstract should state the question(s) you investigated in the first sentence, which you can pull from the Introduction.  The methods used in the research, which you can pull from the Methods section.   You should the summarize the key findings, i.e., those that answer directly the question you started to investigate, then any new trendsor surprising results that followed.  End the abstract with a brief summary of your interpretations and conclusions.   Do not be afraid to boldly assert the implications your results.

Limit your abstract to 200-300 words maximum (a typical standard length for journals.)  Limit your statements concerning each segment of the paper (i.e. purpose, methods, results, etc.) to two or three sentences, if possible.

The Introduction establishes the context of the work being reported.  The Introduction should read from the most general to the specific.  This may mean that in writing this section, you move backwards  That is, you start with the purpose of your work in the form of a hypothesis, question, or problem you investigated; a brief explanation your rationale, approach, and the possible outcomes your study can reveal.  Then you put your contribution into a larger context.  To keep the reader focused, lead your introduction with a sentence using key words from the Title.  Do not get too general with the introduction.  Remember you are not writing a review article or a chapter in a textbook.  You have to assume that every reader has the basic knowledge of the field.

You should give the non-specialist readers enough information to understand your results, and at the same time convince specialists that you have a solid and current understanding of the problem you are investigating.  You do this by citing relevant and most recent primary research literature.  This does not have to be a complete literature review, but adequate enough to satisfy referees, and not insult key colleagues by not citing their work.  Learn how to use citation indexes and review articles.  Do not cite general background references like encyclopedias, textbooks, lab manuals, style manuals, etc, because those sources are considered fundamental or “common” knowledge within the discipline.

In the Materials and Methods section you explain clearly and concisely explain how you carried out your study.  In general you want to convince readers that you did a careful job doing the work and that your results can be believed.  And you want to give enough information so other scientists can reproduce your experiments. But you do not have to give a blow by blow description of every move you made in the lab.

Write this section as if you are verbally describing the conduct of your research.  You may use the active voice to a certain extent, although this section requires more use of third person, passive constructions than others.  Avoid use of the first person in this section.  Use the past tense throughout since the work being reported was performed in the past, not the future.

You should specify the experimental design, i.e., how the experimental rig is configured and what instrumentation you used.  Sometimes just giving the vendor and model number of a piece of equipment is enough to tell specialists how you did the study, what capability was available to you, and any inherent limitations in your data collection.  In addition to the variable(s) measured and the protocol for collecting data, you should write about the controls you implemented, how precision was determined, and how the data were analyzed, i.e., quantitative reductions, conversions, normalizations, or transformations; qualitative analyses, and/or statistical procedures used to determine significance.

This holds true for theoretical and computational studies as well.  For a computational study you will want to tell what software package you used, whether a commercial one or one developed in-house.  Telling what hardware platform you used gives insight into the limitations of your study, as well as how to reproduce your results.  If you developed new software or hardware you will want to talk about how your new applications performed on well-known standard problems.  If you have a new theory, track it back, if you can, to a known experimental result.

In the Results section, objectively present your data, computations, or theory, without interpretation.  Do not even provide subtle interpretive language in this section.  Present the results in an orderly and logical sequence using data displays (tables, charts, figures, graphs, images) and text.  A major function of the text is to provide clarifying information, but write the text of the Results section concisely and objectively. The text should guide the reader through your results stressing the answers to the question(s) you raised in the introduction.

The presentation order of the text should correspond to order of the questions/hypotheses that you say you examine in the Introduction.  Make sure the presentation order of the data displays also corresponds to the presentation order of the text.  Inasmuch as possible you want the typesetter to be able put data displays and the text describing them on the same page.

The passive voice will likely dominate here, but use the active voice as much as possible.  Use the past tense.  Avoid repetitive paragraph structures.  Always give the appropriate units when reporting data or summary statistics.  Check and recheck this.  Many embarrassing situations in science and engineering have been traced back to improper units used in reports.

Important negative results should be reported.  If you did not get the anticipated results, it may mean your hypothesis was incorrect and needs to be reformulated, or perhaps you have stumbled upon something unexpected that warrants further study.  In either case, your results may be of importance to others even though they did not support your hypothesis.  Do not fall into the trap of thinking that results contrary to what you expected are necessarily “bad data”.  If you carried out the work well, they are simply your results and need interpretation.  Many important discoveries can be traced to “bad data”.

The story of the Mössbauer effect makes the point.  Prior to 1957 there had been many failed attempts to find gamma-ray resonance in gases.  It was thought that gamma-ray resonance was not possible because of the energy levels involved.  Rudolf Mössbauer, then a Ph.D. student, was at first surprised to observe gamma-ray resonance in solid iridium.  Even once he was sure that he was seeing what he was seeing, no one knew why gamma-ray resonance was possible in solids, but not in gases.  Mössbauer later reasoned that for the case of atoms bound into a solid, under certain circumstances a fraction of the nuclear events could occur essentially without recoil.  He attributed the observed resonance to this recoil-free fraction of nuclear events, and that led to him being awarded the Nobel Prize in Physics in 1961.  The moral of the story is that Mössbauer’s original observations were not bad data, it was new physics.

Summaries of the statistical analyses may appear either in the text (usually parenthetically) or in the relevant data display, e.g., error bars on a graph, or in the legend as a footnote.  When comparing results between different experimental runs, be sure to articulate both the magnitude (% difference, multiples, orders of magnitude, offset, thresholds, etc.) and direction (greater, larger, smaller, offset, etc) of the differences.  Your statistical notes will help justify your assertion of significance.

“Significant” is a loaded word that is context dependent.  For scientific studies, use of the word “significant” means that you have adequate experimental controls, equipment reliability, and data reproducibility that support a statistical test (p-value < 0.05) used to make a decision about the data; and that the differences in the data are more than what you would expect to get by chance alone. Limit the use of the word “significant” to only these situations.  Otherwise “significance” is left open to interpretation.

The Discussion section is where you interpret your results in light of what was already known, and explain the new understandings that you want people to have after taking your results into consideration.  If the Introduction tells the reader what was that state of knowledge before your results came along, the Discussion tells the reader what the state of knowledge should be after your results are understood.  Use sentence structures that connect your interpretation to exactly to the specific result that leads to that interpretation.  If you cannot give a lucid interpretation, maybe you should not be writing the paper.  Close any and all logical gaps.  Avoid “hand-waving” arguments.

Address each experiment for which you presented results.  Discuss them in the same sequence as presented in the Results, providing your interpretation of what they mean in the larger context of the problem. Because the Discussion relates back to the Introduction, and the Introduction puts your work in a larger context by citing work by others, your discussion must relate your results to the findings of other studies.

Use the active voice whenever possible in this section. Make the strongest assertions possible about what your data means.  Using the first person is acceptable, but an over use of the first person, especially in the singular, starts to bring into question the author’s objectivity.

Do not introduce new results in the Discussion.  You might get away with a new data display that compares your results with some prior known results, but the data displays must not contain new data from your study that should have been presented earlier.

The last part of the discussion should state your conclusions, or the conclusions that you want the reader to make from your results.  It should summarize how you went from one state of knowledge to another state of knowledge.  You should also indicate what you think is the logical next step that either you plan to do, or that someone with specialized capabilities should do.  For instance, if you are reporting a theoretical/computational result, you should talk about what experiment should be done to support your calculations.  Conversely if you are reporting an experimental result, suggest a theoretical/computational analysis that can confirm your observations and lead you to a next set of experiments.

In the Acknowledgements section you thank people from whom you received significant help in formulating, designing, or carrying out the work, or interpreting the results.  This would definitely include persons from whom your received critical materials, especially if they provided them as a favor.  Sometimes just though conversations people can give you key insights into your methodology or data that help move your entire project along.  You should acknowledge people who give you that key insight, and also people that proofread and strengthen your manuscript.  However this is not the place to thank people who give you general moral support.

The last part of the paper is the References section.  This section follows automatically from the citations you make in the other sections.  The real trick is knowing when to cite, who to cite and how to cite. The citations in your paper indicate your background knowledge, the quality of your mentoring, and your professional maturity.  Making the right citations actually starts with doing a thorough literature search and properly understanding the historical background and scientific context of your research problem, which you have to do before you even formulate a hypothesis and an experimental design.

Going back to the Introduction, you want to provide readers with a proper context of your work.  You want to cite the relevant sources adequately though not necessarily completely.  Again you want to convince specialists that are reading your article that you understand the exact subject matter by citing the “right” references.  But you have brevity constraints and thus cannot give a full literature review.

In general do not cite textbooks.  They contain fundamental concepts that everyone in the reasonable readership of your paper knows.  Citing a textbook is therefore unnecessary.

Peer-reviewed journal review articles are an excellent place to point new entrants to the field to get some specific background.  You can cite a review article in the very beginning of your Introduction.

The bulk of your citations should be the primary peer-reviewed research journal articles.  Especially when you get down to the finer points of your experiments and comparing your results to those carried out by scientists, you want to make sure you are fairly and faithfully representing those other results, and not someone else’s condensed interpretation of those results.  And you want to give others a dependable intellectual trail to support and validate all your arguments.

Beyond bringing your work into context, and convincing people that you know the field, citing references is also an exercise in professional courtesy.  You should understand that citations are a key metric in the evaluation of scientists.  That is, scientists’ work is judged based in part on how many others cite their work.  You do not want to rudely or naively ignore peoples’ work, especially important people.  But you do not want to be suck-up and cite paper that is off-point, but the authors are your friends or people that you want to cozy up to.  You also want to make sure that the actual paper you cite supports the point that you are trying to make in your paper.

There are several standard rules covering when and how to cite a reference.  For instance, if you are pulling a direct quote from a source that is longer than 2-3 lines, you should set the text off as an indented paragraph.  You should be very careful about citing websites.  Because websits are not necessarily peer-reviewed, they may not be authoritive, and probably not permanent.  (This is a key value of archival journals; permanence of the content for future generations.)  Instead of citing a website, your safer bet is to  to find the original sources of the knowledge on the website.  But if you must cite a website be sure to include the full URL and the date that your retrieved the information.  Refer to the standard and discipline specific writing guides for more grammar and formatting rules.

While all journals use the same format elements discussed here, each journal will have its own author instructions that cover how headings are to appear, what styles are to be used for citing literature, both in the text and in at the end of the article.  There are bibliography and citation management tools like BibTex and EndNote where you can create a library of interesting articles and catalog them according to some classification scheme, i.e, what system they studied, what point they make, what method they use, conclusions they reached, etc, etc.  The good thing about these tools is that they can quickly format your citations and cited reference sections according to whatever style that the journal you are submitting to uses.

To summarize, a scientific paper should follow a standard format that allows people to understand the paper at different levels, and that well serves the paper in the standard indexing schemes. The Title should use hard hitting words that tell people exactly what the paper is about.  The Abstract should in one paragraph draw in readers to the context and results of the article.  Remember it is after reading the abstracts that people decide whether or not to read the entire article.  The Introduction is where you at once let the reader understand the context of your work, and indicate to specialist that you understand what you are doing.  The Methods section is where you give enough information to allow people to believe your results and to possibly reproduce your results.  The Results section is just the facts.  Present your results without interpretation using data displays (tables, figures, charts, graphs) that give the data the most concise and powerful exposition.  The Discussion section is where you tie what was previously known to how your results extend the knowledge.  It is the place where you assert how important your results are, and what should be done next.  Be sure to acknowledge people who helped you produce the results or improve the manuscript.  This is an important professional courtesy as is citing the “right” references.  Be sure to adhere to common grammatical, formatting and citation rules, and by all means follow the author instructions of the specific journal to which you are submitting your paper.

Much of this article was adapted from an article by Prof. Greg Anderson, Bates College.