Some thoughts on writing a scientific paper or thesis
L.B. Railsback, Department of Geology, University of Georgia

First, READ this paper: Mumpton, F.A., 1990, The universal recipe or how to get your manuscript accepted by persnickety editors: Clays and Clay Minerals, v. 38, p. 631-636. The following is not a substitute for Mumpton's paper, but it adds some comments.


(W1) Philosophy.
     The point of writing a scientific paper is to communicate the findings and significance of your research. Always envision yourself writing to a reader who (a) isn't familiar with your study area, samples, or methods, (b) may be (and as a scientist should be) skeptical of the claims you are making, and (c) probably has more pressing things to do with their time and so will skip your article unless you persuade him or her of its clarity and significance. No one will be obligated to read your paper, so you have to persuade them to start reading, and you have to write clearly enough that they keep reading.

(W2) The Introduction.
    The purpose of the introduction is to tell the reader what the paper is about and, more importantly, to justify to the reader why the paper is significant. If a "previous work" section is included here or elsewhere, its goal should be to show how your paper is an advance or improvement on previous work. A section on "previous work" that just shows that the author did his or her time in the library will be skipped or will make the reader quit reading entirely.

     After justifying interest in the topic, don't be afraid to use a sentence that begins "This paper uses/examines/reports . . .", because telling the reader what the paper is about is also one of the main goals of the introduction. This sentence may end with a clause that begins "in order to . . ." or "as a contribution to . . .", so that you also tell the reader what your goal is with respect to refining or increasing scientific knowledge. Such a sentence can, from the reader's perspective, be one of the most important in the whole paper, because the reader then knows what to expect as she or he proceeds. It may be an important sentence for the writer as well, because you're forced to decide what the focus of your paper really will be, and you may realize that you're covering either too many things or not enough things, or not using those things to help refine or increase scientific knowledge.

(W3) Geologic Setting.
     Geologists love to wax poetic about the geologic context of their work. Often such a context is vital to understanding the paper. However, this section of a paper, if present, should not include every detail known about the region or rock unit. It should tell, and should only tell, the information needed to put the results and discussion in context.

(W4) Methods.
    This section of a paper is always boring but almost always necessary. We've all read papers where we've wondered "where did they get this data?" or "what exactly do the data, at the most basic level, mean?". This section should answer those questions. It should not tell what kinds of plots were made, because readers will see that easily enough when they look at the figures.

(W5) Results.
     This is one of the hardest sections for students to write well. Two rules to remember are (1) Be sure to tell what the results are, but (2) Don't interpret the results, or at least make clear where you make an interpretation. For the first, don't tell the reader how you plotted the data, and don't tell the sequence of ways you tried to understand or portray the data. Instead, let the data do the talking, and in fact it's easiest to let the pieces of data be the subjects of sentences.

Consider the following example:

"Compositions of avocados from southern Florida range from xx.xx to yy.yy weight percent QuJO3, with a mean of zz.zz (Fig. 2). In contrast, avocados from central and northeastern Florida have compositions ranging . . . . "
The text above reports the data. On the other hand, consider this text:
"When plotted on an X-Y plot, avocados from the two regions are different."
This tells us what kind of plot was made (which we can figure out, or we can read the figure caption), but it doesn't tell us anything about the results themselves. Many students fall into the trap of reporting how they plotted their data rather than telling what the data are.

     The other trap into which some students fall is to use just a sentence or two in the Results to direct the reader to tables or figures. Sentences like

"Tables 1 and 2 show the data from avacodoes, as do Figures 1, 2, and 3."
do not report the results. See the first example above ("Compositions of avocados . . . .") for a sentence that does report results.

     Reporting the results includes making distinctions between bundles of data and reporting trends and correlations. If statistical analyses or (heaven forbid) mere visual analyses show differences or trends, these are results. Explaining what you think are the geologic reasons for these trends (i.e., making your interpretations) is the business of the discussion.

(W6) Discussion.
     The discussion presents your explanation and interpretation of the data: the geologic reasons for the trends or differences you've reported in the results, and the significance of these findings to larger scientific questions that have been raised in the literature or (less commonly) that you're raising for the first time. The Discussion should not introduce data from your work that haven't already been reported in the results. However, it can have plots or tables showing the results of others, for purposes of comparison with your data.

(W7) Reference to Figures.
     Every figure and table must be cited in the text. If a figure or table isn't sufficiently relevant to be cited in the text, it isn't sufficiently relevant to be included at all. Figures should be numbered so that Figure 1 is mentioned first in the text, then Figure 2, and so on.
     The relationship between figures and text is often redundant. You can't assume that, just because a relationship is shown in a figure, it doesn't need to be stated in the text. A figure is, after all, supposed to be an illustration of a statement in the text. Most editors like text that describes all the data in words, so that the figures are not really essential and are only mentioned parenthetically (literally). You'll find that many reviewers who read manuscripts also want (or need) this kind of text, if only because they're too lazy to leaf back to the figures.
     Some good and bad examples of text/figure linkage are
    Type I: "Sample age and pH are correlative (Fig. 3)."
    Type II: "Figure 3 shows that sample age and pH are correlative."
    Type III (not good): "Figure 3 shows the relationship of sample age and pH"
    Type IV: (worse): "Sample age and pH are shown on Figure 3".
    I is best; II is passable for many readers, but III and IV communicate less and less, and effective efficient communication is our goal.

(W8) Closure.
     Papers commonly end with "Conclusions" or a "Summary". The two are not the same. Conclusions are just that - the inferences that can be drawn from your data, not a reprise of the entire paper. In a sense, this may be a list of the main points you want the reader to take away from reading your paper - the points demonstrated by your research. A summary, on the other hand, is indeed a recapitulation of the entire paper that hopefully incorporates the main conclusions. A summary is usually redundant and unnecessary. Conclusions, on the other hand, may be very useful to drive home the key points, especially for readers who have waded through the discussion but are still asking "So what is the bottom line here?".

(W9) References Cited.
    A paper should have a list of the literature cited in the paper, including those in the tables, in the figures, and on the figures. The list of references should not contain any literature not cited in the paper. In proofreading your paper, put a checkmark next to each reference in the list as you encounter it in the paper; when your're done, references without checkmarks should be deleted.
    For submission to a journal, the format of the references varies with the journal. For a class paper or thesis, internal consistency of format is a minimal requirement.

(W10) Abstract (of a paper) (It comes first, but it is written last).
     The abstract should summarize the paper: it should mention the material and/or study area, it should mention the methods, it should summarize the results, and it should present the main conclusions. The first sentence should probably put forth the most striking result or conclusion, so that the reader will keep reading.
     That all sounds like a lot, but a good first sentence can actually do most of the work, or at least get it started. For example:

Atomic adsorption analyses of 54 avocados sampled from groves in Florida and Georgia, coupled with previously published groundwater analyses, show that concentrations of alkali metals in avocados are proportional to springtime concentration of those elements in shallow groundwater.
     This first sentence gets its material and methods into the subject, with the main conclusion to follow in the predicate. Further sentences can then go into the details and/or elaborate the significance of the conclusions. This formulaic approach to the first sentence can save an author hours of head-pounding to craft the most important ten to thirty words in the entire paper.

     Here are some first sentences that would tell the reader much less:

            Groundwater chemistry is very important.
            The geochemistry of alkali metals has many interesting facets.
            Many people eat avocadoes.

     You may have read, or perhaps will read, abstracts written in a European style that goes like this:

Avocados from groves in Florida and Georgia were studied by means of atomic adsorption analyses, and these results, coupled with previously published groundwater analyses, lead to a number of interesting inferences.
     This tells the reader very little and should not be imitated.

(W11) Title (It gets written early on, but it ought to be seriously reconsidered at the end).
    A good title (1) has all the "key words" relevant to the paper; (2) conveys the key idea(s) of the paper; and (3) is specific. Let's go back to the first sentence that we constructed:

Atomic adsorption analyses of 54 avocados sampled from groves in Florida and Georgia, coupled with previously published groundwater analyses, show that concentrations of alkali metals in avocados are proportional to springtime concentration of those elements in shallow groundwater.
Titles, in increasing order of quality, might be
    "Avocados"
    "Avocados in Florida and Georgia", "Avocados and groundwater", or "Alkali metals in avocados"
    "Linkage of avocado trace elements to groundwater" of "Relationship of fruit chemistry to groundwater"
    "Dependence of alkali metals in avocados from Florida and Georgia on groundwater geochemistry"

(W12) Figures and Tables.
     Figures are photographs, sketches, or drafted illustrations. They contain more than words and numbers. Tables are lists of words or numbers arranged in rows and columns. Figures have captions at the bottom and often have explanatory legends within their fields. Tables have titles at the top, and they may have notes at the bottom providing information necessary to understand the material. Tables do not, however, have captions.

(W13) Figure Captions.
     The caption of a figure should describe the figure. Its first sentence (actually usually its first sentence fragment) should be a general description of the figure.

A rigorous caption would begin with words like
    Plot of C and O isotopic compositions of Pohick Valley carbonates . . . .
    Map of Sr concentration in groundwater of the East Lakedale Quadrangle . . .

Slightly less rigorous but generally acceptable captions would begin with
    Variation in C and O isotopic compositions of Pohick Valley carbonates . . . .
    Sr concentration in groundwater of the East Lakedale Quadrangle . . .

Much less useful captions might begin, and end, with
    Pohick Valley carbonates.
    The East Lake Quadrangle.

Hardly more useful would be
    Isotope analyses of Pohick Valley carbonates show that . . . .
    Groundwater in the East Lake Quadrangle evolved from . . ..

At the bottom of the barrel would be something like these:
    Graph of data.
    Map.

    After this first sentence, subsequent sentences should explain the details of the diagram that aren't immediate apparent from a cursory examination. It's a good idea to look at a figure inch by inch to find the things that are obvious to you but are probably obscure to others.

(W14) Pagination.
     Page numbers on a manuscript are essential. If the reviewer drops your manuscript and can't put the pages back in order, all your work becomes literally meaningless. Page numbers also let reviewers have mileposts, so that the reviewer can say "On page 4, the author states . . .". Such mileposts will make the reviewer's comments much more useful to you.

(W15) Abstract (for a talk or poster session).
     The abstract should summarize the presentation: it should mention the material and/or study area, it should mention the methods, it should summarize the results, and it should present the main conclusions. Abstracts usually don't have references or figures. However, GSA allows authors to put graphics in their allotted abstract space, and figures can accompany extended abstracts.
     The first sentence should probably put forth the most striking result or conclusion, so that the reader will keep reading. That all sounds like a lot, but a good first sentence can actually do most of the work, or at least get it started. For example:

Atomic adsorption analyses of 54 avocados sampled from groves in Florida and Georgia, coupled with previously published groundwater analyses, show that . . . . .
     This first sentence gets its material and methods into the subject, with the main conclusion to follow in the predicate. Further sentences can then go into the details and/or elaborate the significance of the conclusions. This formulaic approach to the first sentence can save an author hours of head-pounding to craft the most important ten to thirty words of the entire paper.

 


There are also Railsback pages on editing scientific papers and theses, a page on scientific papers for Railsback's (and probably others') classes, a page on writing reviews of manuscripts at the request of editors of journals, and a page on writing scientific grant proposals .

There are also pages on writing scientific grant proposals and

Email to Railsback (rlsbk@gly.uga.edu)
Railsback's main web page
UGA Geology Department web page