You are on page 1of 17

Scientific Writing and Scientific Discovery

Author(s): Frederic L. Holmes


Source: Isis, Vol. 78, No. 2 (Jun., 1987), pp. 220-235
Published by: The University of Chicago Press on behalf of The History of Science Society
Stable URL: http://www.jstor.org/stable/231523
Accessed: 07-05-2016 09:49 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

The University of Chicago Press, The History of Science Society are collaborating with JSTOR
to digitize, preserve and extend access to Isis

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
HSS LECTURE

Scientific Writing and


Scientific Discovery
By Frederic L. Holmes*

7W HEN SCIENTISTS REFER to the "literature" of their fields, they have


in mind something very different from what we mean when we talk of
literature in general. The literature of a scientific specialty area is the accumu-
lated corpus of research articles contained in the journals of the field, and it is
regarded as the primary repository of the knowledge that defines the state of that
field. Rarely is that literature examined in the same way as "literature" proper is
examined, for its form as well as its content, its style as well as its meaning. A
scientific paper may be treated as presenting a creative achievement but is not
itself taken as a creative achievement. Works that fall within the traditional
genres of "creative writing" are treated as integral wholes, in which form, mean-
ing, content and style are intimately connected. Scientific papers, however are
generally viewed only as vehicles for conveying information.
Scientific papers, in the form that they have taken since the emergence of
specialized journals during the seventeenth and eighteenth centuries, are var-
iously viewed as reports of completed investigations, public announcements of
discoveries, knowledge claims, the primary locus of the "context of justifica-
tion," the means by which scientists gain recognition, and the end products of a
process of "inscription." They are regarded, however, by historians of science
and others, as very imperfect reflections of creative scientific activity itself. One
of the best known, most trenchant expressions of this view is that of Peter Meda-
war: "What scientists do has never been the subject of scientific ... inquiry. It is
no use looking to scientific 'papers' for they not merely conceal, but actively
distort the reasoning that goes into the work they describe. If scientific papers
are to be accepted for publication, they must be written in the inductive style.
The spirit of John Stuart Mill glares out of every editor of a Learned Journal."'
Medawar's statement succinctly summarizes several characteristics of scientific
papers that are commonly accepted:
* That they are retrospective formulations of work previously completed.
* That they do not accurately represent the work they make public.
* That they are stereotyped according to canons of form dictated by the
authority structure of scientific disciplines.

* Section of the History of Medicine, Yale University School of Medicine, P.O. Box 3333, New
Haven, CT 06510.
This article, as a History of Science Society Lecture, is but a preliminary investigation of its topic.
'P. B. Medawar, The Art of the Soluble (Harmondsworth: Penguin Books, 1969), p. 169.

220

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
SCIENTIFIC WRITING 221

* That they purvey an image of scientific activity that fits an ideology rather
than actual practice.
Most historians of science, I suspect, are sympathetic toward Medawar's
opinion; I have myself assumed a similar view. My own research program-to
recover from unpublished documents, in particular from daily laboratory rec-
ords, the "actual" investigative pathways through which scientists have reached
the conclusions they report in published papers-can be regarded as an effort to
unmask what Medawar claims the scientific paper conceals. I have argued be-
fore, and still maintain, that such documents provide highly favorable opportuni-
ties to follow the intimate interplay of thought and action that constitutes the fine
structure of scientific creativity. Some recent experiences, however, arising par-
ticularly during my study of Antoine Lavoisier, have induced me to modify my
original view of the relation between the activity that can be revealed in this way
and the scientific papers that emerge from it. I hope that, as a consequence, I can
offer some fresh perspectives both on the intrinsic character of scientific papers
and on their place in creative scientific "work." My suggestions are by-products
of my pursuit of other questions, and I have not formulated them systematically,
nor pursued them through a broad examination of research papers other than
those that I have encountered in the course of my particular scholarly and teach-
ing interests. What I have to say is therefore preliminary and impressionistic. I
hope merely to open a conversation with others who may have had similar expe-
riences in their encounters with the literature of science.
In my book Lavoisier and the Chemistry of Life, I followed a method and a
conception that had grown out of my previous study of Claude Bernard.2 I in-
tended to reconstruct Lavoisier's scientific investigation mainly out of his pre-
served laboratory notebooks, which proved to be a rich record of his experimen-
tal activity. I regarded his published scientific papers as the outcomes of these
investigations, containing the arguments that he formulated after their comple-
tion to support conclusions he had actually reached during the course of the work
traced in his notebook record. Although the papers could provide indirect clues
about the state of Lavoisier's thought at earlier stages in the investigation, I kept
always in mind the philosopher's dictum that these were ideas recast within the
context of justification, and not directly a part of the context of discovery.
After finishing my book and submitting it to a publisher, I went back to Paris
for a week to check whether there were in the Lavoisier archives in the Academy
of Sciences any relevant documents to which I had not had access while I was
writing. I did not expect much. I did know that for almost all of the published
papers of Lavoisier that I had analyzed there existed handwritten manuscripts. I
thought I should examine them to see whether they differed in any way from the
printed versions, but I assumed that what was most likely to have been saved
were those final versions that are sent off for publication, with only occasional
variations from what comes out of the press. I found, however, that there was
often more than one draft-usually two or three, but sometimes even more. Not
until I had received the microfilms, transcribed the less legible of them, and
started comparing successive drafts did I begin to realize how crucial they were
as a source of insight into Lavoisier's creative thought. In most cases there was a
likely "first" draft, identifiable by its roughness and by its being written in

2Frederic L. Holmes, Lavoisier and the Chemistry of Life (Madison: Univ. Wisconsin Press, 1985).

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
222 FREDERIC L. HOLMES

Lavoisier's own hand, on the right-hand side of the page, with a wide margin for
revisions to the left. In such drafts one could see two stages of revision occur-
ring: within the right-hand text, changes that Lavoisier made as he went along; to
the left, changes that he made afterward. Successive later drafts could be put in
chronological order, because each consisted of a fair copy of the preceding draft,
and each also had a wide left-hand margin on which Lavoisier made further
revisions in his own hand. As I studied these drafts, I could begin to see Lavoi-
sier not merely substituting a more felicitous phrase here and there, or rearrang-
ing the parts of a text, or adding or omitting pieces to shape those highly polished
scientific memoirs for which he is so well known; the development was not only
in the manner of presenting conclusions already reached, or even in refining the
conclusions themselves. In several of these papers I could watch important ideas
emerging, growing, changing form, or decaying during the evolution of a scien-
tific paper. I shall summarize here briefly just one such example (see Figure 13).
Lavoisier read his memoir "Experiments on the Respiration of Animals and on
the Changes Which Take Place in the Air in Passing Through the Lungs" to the
Academy of Sciences in May 1777, and it was printed in the memoirs of the
Academy for that year (actually published in 1780).4 It is an exemplary scientific
paper in the modern sense, reporting experiments carried out on the calcination
of mercury and on the respiration of birds, phenomena between which Lavoisier
saw a close connection, and the conclusions that he drew from these experi-
ments.
The experiments on the calx of mercury constituted his famous analysis and
synthesis of air. Those on respiration included the historic demonstration that an
animal absorbs a portion of the atmosphere-a component that he then called
"pure air," "eminently respirable air," or sometimes still "dephlogisticated air"
-and returns to the atmosphere a nearly equal portion of another species of air,
the well-defined "fixed air." In his conclusions concerning respiration Lavoisier
suggested two alternative theories: that the respirable portion of the air is con-
verted to fixed air while passing through the lungs; or that the respirable air is
absorbed into the organism and replaced by an almost equal quantity of fixed air.
Historians of science have sometimes misinterpreted these two views as involv-
ing merely the question of the site of the conversion of respirable air to fixed air.
The difference between them is more radical.
We can trace the emergence of these views through a succession of six drafts
of this paper, the writing of which was spread over at least six months. From an
examination of the temporal layers in Lavoisier's formulation we find that, of
these two theories, the one mentioned first-which stands as the historical
"root" of what we know as Lavoisier's theory of respiration-was not present at
all in the first four drafts of the paper. In the third draft (and perhaps also in the
lost portions of earlier ones) he wrote the following:

One cannot doubt, therefore, that the respirable portion of the air becomes fixed in
the lungs, and it is very probable that it combines there with the blood. That probabil-
ity is transformed to a kind of certitude if one considers that it is a recognized prop-

3See also the notebook pages illustrated in C. E. Perrin, "Lavoisier on Calcination and Combus-
tion," Isis, 1986, 77:647-666.
4A. L. Lavoisier, "Expdriences sur la respiration des animaux et sur les changements qui arrivent a
l'air en passant par leur poumon," Memoires de l'Academie des Sciences, 1777 [1780], pp. 185-194.

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
SCIENTIFIC WRITING 223

Figure 1. First page (untitled)


mi^m^ ^ ^Sy^y- ^-^4~~~~~ii-:~'ii8 of the first draft of Lavoisier's
memoir eventually entitled
"Experiences sur la respiration
des animaux et sur les
changements qui arrivent a I'air
en passant par leur poumon,"
/K ~ Lavoisier Papers, Fiche 1349,
Archives of the Academy of
Sciences, Paris.

erty of the air that is better than common air to impart a red color to the bodies with
which it combines. Mercury, lead, and iron provide examples. These metals, com-
bined with the respirable portion of atmospheric air, each form a beautiful red calx.
... It is the same with the blood of animals, it is not bright red except to the extent
that it is nearly continuously in contact with the air, it becomes dark in fixed air, in
every air that is not respirable, in the vacuum of a pneumatic machine. It recovers its
red color, on the other hand, when it is exposed to common air, and especially to the
air that is better than common air: and it is no doubt for this reason that, according to
the observations of many anatomists, the blood is much brighter upon leaving the
lungs than when entering it, when taken from the pulmonary vein than from the
pulmonary artery.
In view of such striking analogies, does it not seem that one is justified in conclud-
ing that the red color of the blood is due to its combination with the air, or rather the
respirable part of the air, and that this is one of the principal objects of the fixation of
the air in the animal economy[?]

In his concluding paragraph Lavoisier wrote: "I believe that the Theory of respi-
ration has been established."5
Not only is this theory of respiration unlike the theory that is familiar to us; it
also ignores one of the two "effects" of respiration that Lavoisier had established
experimentally-the formation of fixed air. Moreover, it gives no way to account
for the steady accumulation of respirable air in the blood that would result from

5A. L. Lavoisier, "Experiences sur la decomposition de l'air dans le poulmon et sur un des princi-
paux usages de la respiration dans 1'economie animale," Archives of the Academie des Sciences,
Lavoisier Papers, fiche 1349, pp. [9-10]. This and all translations are mine.

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
224 FREDERIC L. HOLMES

its continuous absorption. In spite of these critical shortcomings, as the passage


shows, Lavoisier was, at the time he wrote this draft, very enthusiastic about his
idea and believed that with it he had established the theory of respiration. There
is not space to explain in detail here how he was led into such a position, but the
main driving force was the analogy between respiration and the calcination of
mercury. His experiments on respiration were modeled after those he had carried
out on the calx of mercury. Conceptually, the analogy between the two pro-
cesses-that both absorbed the respirable portion of air and produced a red sub-
stance-so captured his imagination that he focused on them to the exclusion of
other aspects of the experimental results that he had described in the very same
manuscript. Another contributing factor was the strong influence upon him of a
recent treatise on respiration by Joseph Priestley which had shown that blood is
bright red only when it is in contact with dephlogisticated air.6
In the next draft of the memoir, which Lavoisier planned to read at a meeting
of the Academy in November 1776, the theory of respiration just quoted ap-
peared unchanged, but various modifications of the language-in the title of the
memoir, in the elimination of the statement that the "theory of respiration is
established," and elsewhere-suggest that he was tempering his initial enthusi-
asm and was growing more circumspect about the status and significance of his
theory (see Figure 2).7
Unable to present his memoir at the meeting because there was not enough
time, Lavoisier afterward reworked portions of this draft again, and in the pro-
cess he somehow came to realize that he must consider an alternative to his
theory. In a portion of the manuscript rewritten in his own hand he put down:
"Of two things, one, either a portion of pure air, dephlogisticated air contained in
the air of the atmosphere, is absorbed during respiration and replaced by an
almost equal quantity of fixed air or mephitic gas which the lungs restore to its
place, or else the effect of the respiration is to change that same portion of
dephlogisticated air into fixed air in the lungs."8 The character of this paragraph
-Lavoisier made several changes in the wording as he went along, only the final
versions of which I have reproduced, and he rewrote the ideas in a different
order in the left-hand margin-suggests that he was here expressing for the first
time the view that dephlogisticated air, or pure air, is converted to fixed air in the
lungs. This alternative appears to have the major advantage that it incorporates
both of the observed "effects" into a single theory.
If this is in fact the first version of his new idea, then we can almost watch the
germ of Lavoisier's second, more significant theory of respiration becoming visi-
ble on this manuscript page. Lavoisier himself, however, did not at the time
perceive the alternative as clearly superior, or even as preferable, to his earlier
one. As he developed the two alternatives, he appeared to be debating with
himself what attitude to take toward them.
In still later revisions of his manuscript we can see Lavoisier gradually shifting
his opinion of the relative strengths of the two theories. It appears evident that he

6Joseph Priestley, "Observations on Respiration, and the Use of Blood," Philosophical Transac-
tions of the Royal Society, 1776, pp. 227-248.
7A. L. Lavoisier, "Experiences sur la d6composition de l'air dans le poulmon des animaux,"
Lavoisier Papers, fiche 1349.
8lbid.

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
SCIENTIFIC WRITING 225

Figure 2. Revised draft of the


memoir on respiration,
intended for presentation at the

Sciences, November 1776.


Lavoisier Papers, Fiche 1349.

was so attached to the earlier one that he at first found the alternative unwel-
come, but he was driven eventually to regard the two as "equally probable" and
to suggest that perhaps both processes occur.
This example, which is representative of the development of several of his
other memoirs, illustrates how Lavoisier apparently acquired some of his crucial
insights in the process of writing out his ideas to prepare a scientific paper. We
cannot always tell whether a thought that led him to modify a passage, recast an
argument, or develop an alternative interpretation occurred while he was still
engaged in writing what he subsequently altered, or immediately afterward, or
after some interval during which he occupied himself with something else; but
the timing is, I believe, less significant than the fact that the new developments
were consequences of the effort to express ideas and marshall supporting infor-
mation on paper.
During my study of the work of Hans Krebs I obtained from that author of
major scientific innovations direct testimony concerning the clarifying effects of
scientific writing on his scientific thought. While discussing with me the paper in
which, in 1932, he had presented his first major discovery-the ornithine cycle of
urea synthesis-Krebs commented:

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
226 FREDERIC L. HOLMES

At that time the papers were short. The writing up didn't take very much time.
Especially as bits and pieces were written up all the time, it was a matter of putting
together the final paper then. I spent a lot of time on writing, but usually while the
work was still going on. And I find in general only when one tries to write it up, then
do I find the gaps. I cannot complete a piece of work and then sit down and write the
paper.9

Other scientists have remarked to me that Krebs's procedure and experience


are not unusual; that experimental scientists commonly begin writing up their
papers during the investigations those papers are intended to report, and that the
writing helps to guide the further course of the investigation.
Although the complete record of Krebs's laboratory experiments is available,
he unfortunately did not save those "bits and pieces" out of which he composed
final drafts. I have been able to utilize his statement about how he worked in a
general way. For example, from the form of his first paper on urea synthesis I
can argue that Krebs may have written about two thirds of the paper before he
arrived at the solution of the experimental phenomenon known as the "ornithine
effect." Occasionally I can identify turns in his experimental pathway that may
be explained as efforts to fill "gaps" identified by writing a paper; but I cannot
fully exploit his revealing description of how he wrote, cannot describe in detail
that close interplay between scientific writing and scientific investigation that
must have taken place.
It will be rare to find scientists who have preserved full records both of their
daily laboratory results and of successive drafts of papers written during the
course of the investigations they report-especially if one is looking for scientists
whose contributions were sufficiently significant and imaginative to justify finely
detailed studies of their research pathways. I believe that it is worth searching for
such situations, however, because they could provide exceptional insights con-
cerning the interactions between writing, thought, and operations in creative sci-
entific activity.
Even without such dense documentation, we can readily see other forms of
interaction between writing and investigation in cases where we have a record of
the laboratory operations to compare with the published papers. Although we
normally think of the scientific paper as a presentation of a completed investiga-
tion, or at least of a coherent phase of an ongoing one, we can see in some
circumstances that the writing of the paper defines, or redefines, the objectives,
the boundaries, and the meaning of the investigations themselves. Lavoisier's
paper on respiration, for example, joined two sets of experiments on seemingly
different problems-the calcination of mercury and respiration, into one broader
inquiry. The connection afterward fell away, when Lavoisier abandoned his first
theory of respiration and switched the analogy to respiration from calcination to
the combustion of charcoal. Historians generally sunder what Lavoisier had
joined in this paper, treating the memoir on respiration as though it contained
two unrelated parts. A more lasting investigative entity created as much by the
writing as by the experiments is Lavoisier's combustion analyses to determine
the composition of plant substances. His memoir "On the Combination of the

9Hans Krebs to F. L. Holmes, conversation, 29 April 1977 (recorded transcript, tape 10-11, p. 21,
author's collection). The paper referred to is Hans Adolf Krebs and Kurt Henseleit, "Untersuchun-
gen uber die Harnstoffbildung im Tierk6rper," Klinische Wochenschrift, 1932, 11:757-759.

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
SCIENTIFIC WRITING 227

Oxygen Principle with Spirit of Wine, Oil, and Different Combustible Bodies"
stands historically as the starting point for the systematic analysis of the elemen-
tary composition of organic compounds.10 The experiments he described in this
memoir, however- on the combustion of alcohol, olive oil, and wax-were not
all originally performed for that purpose. He had carried out the experiments on
wax earlier, as part of his effort to establish the composition of fixed air. He
began the experiments on alcohol as another demonstration of the synthesis of
water, and only during their course came to view the same general procedure as
a means to determine the composition of alcohol. Even after shifting his objec-
tives in this way, Lavoisier saw himself as analyzing further examples of com-
bustible bodies in general. Not until he wrote the drafts of a memoir did he come
to see the method as applicable in particular to plant and animal matter. He did
not set out to perform the first combustion analyses of organic substances; he
only gradually came to appreciate that this was the main significance of what he
had achieved.
In these examples we can see Lavoisier knitting together experiments carried
out for diverse purposes to create coherent investigations on paper. We can find
other examples in which the writing of a scientific paper produced bounded,
discrete investigative units by lifting more limited problems out of the broader
investigative stream within which the experiments involved were originally per-
formed. Claude Bernard, for example, pursued during the years 1843-1848 an
ambitious research program on digestion and nutrition. He aimed to follow the
chemical processes in animals through the digestive tract, into the bloodstream,
and finally to the excretory products that appear in the urine. He sought persis-
tently but 'unsuccessfully to support a general theory of digestion that would
reduce the actions of the various digestive juices-especially saliva, gastric juice,
and pancreatic juice-to a common principle. During this period he published a
number of papers on specific discoveries he had made concerning digestive or
nutritive processes, none of which revealed fully its connections with his broader
investigative objectives. They appeared instead as solutions to more limited,
bounded problems. For example, his 1846 paper on "The Differences Between
the Phenomena of Digestion and Nutrition in Herbivorous and Carnivorous Ani-
mals"-despite its title-presented a compact, self-contained investigation, mov-
ing in a few closely logical steps from a striking observational distinction to a
fundamental theoretical conclusion. Rabbits normally excrete turbid, alkaline
urine, in contrast to carnivorous animals, whose urine is clear and acidic. Ber-
nard asked whether these differences, as well as the differences in the reaction of
the chyme and chyle of carnivores and herbivores, reflect profound differences
in their nutritional processes. He had found, however, that if rabbits are fasting,
their urine becomes clear and acidic. Moreover, it is also clear and acidic if they
are fed meat, whereas dogs nourished on bread begin producing clouded, alkaline
urine. The differences were therefore not fundamental, but merely due to the
nature of the normal diets of the two classes of animals.11

'0A. L. Lavoisier, "M6moire sur la combinaison du principe oxygine avec 1'esprit de vin, l'huile, et
diff6rents corps combustibles," in Oeuvres de Lavoisier, 6 vols. (Paris: Imprimerie Imp6riale,
1862-1893), Vol. II, ed. J. B. Dumas, pp. 586-600.
"Claude Bernard, "Des diff6rences que presentent les ph6nomenes de la digestion et de la nutrition
chez les animaux herbivores et carnivores," Comptes Rendus de 1'Academie des Sciences, 1846,
22:534-537.

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
228 FREDERIC L. HOLMES

From the published paper alone we would reasonably infer that this investiga-
tion formed a discrete episode in Bernard's research career, beginning in an
unexpected event and leading to a definitive conclusion. From his laboratory
record we can show, however, that after he had made the initial observations, he
drew far broader inferences and attempted to integrate the unexpected findings
into the wider experimental and conceptual fabric of his ongoing investigation.
Only after an inclusive theory of nutrition, built hastily upon some of the early
results, had collapsed did he come to limit his objectives in the way portrayed in
his paper. It was in the writing that he detached this investigation from the net-
work of experimental lines constituting his research program, to create a coher-
ent beginning, middle, and ending.12
A more subtle example of the interaction between writing and the shaping of
an investigation can be seen in Hans Krebs's papers describing the ornithine
cycle of urea formation. When he began his investigation of the synthesis of urea,
in July 1931, he did not have an overall guiding hypothesis in mind. His objective
was simply to apply manometric methods, using tissue slices, to a set of ques-
tions that had been well defined in the existing literature of the field. After sev-
eral months it appeared that his investigation would merely confirm the prevail-
ing view of the subject, while refuting several of the more speculative theories
that had previously been proposed. Then, in November of that year, he encoun-
tered the anomalous ornithine effect-that is, the striking rise in the rate at which
a liver slice produced urea when the specific amino acid ornithine, together with
ammonia, was added to its medium. During the following months Krebs concen-
trated his research on an effort to explain that effect, but not to the complete
exclusion of experiments on other aspects of urea synthesis. When he wrote his
first preliminary paper on the subject-which, as I have already suggested, he
probably began before he had solved the problem posed by the ornithine effect-
his "theory of the ornithine cycle" not only appeared late in the text but did not
provide the overall organizing problem for the paper, which he entitled "Investi-
gations Concerning the Formation of Urea in the Animal Body." That is, Krebs
presented his theory only as the "main result" to emerge from a more general
investigation.
A few weeks later Krebs took time to write a longer paper,13 giving full de-
scriptions of the experimental methods and results that he had only summarized
briefly in his two preliminary papers (in the second he had added a step to the
cycle that was not included in the first). Conceptually the definitive paper was
nearly the same as the preceding ones. Krebs took the section directly describing
his "Theory of the Formation of Urea" almost word for word from his initial
paper; but where the first paper had treated various aspects of the formation of
urea in sections that were nearly independent of one another, and of which only
the last few dealt with the ornithine theory, the larger paper began with the three
reactions that made up the theory. The remainder of the paper, which was mainly
a fuller description of the topics covered in the preliminary papers, served as the
evidence on which the theory was based. Thus the consolidation of his solution

12Frederic L. Holmes, Claude Bernard and Animal Chemistry (Cambridge, Mass.: Harvard Univ.
Press, 1974), pp. 278-295.
13Hans Adolf Krebs and Kurt Henseleit, "Untersuchungen uiber die Harnstoffbildung im Tier-
korper, Hoppe-Seyler's Zeitschrift fur physiologische Chemie, 1932, 210:33-66.

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
SCIENTIFIC WRITING 229

to the ornithine effect had changed for Krebs the meaning of the overall investi-
gative enterprise that he had been following for the previous nine months. In the
writing of the definitive article, the investigation became a coherent whole, ori-
ented around the theory of the ornithine cycle.
From the perspective of Medawar's statement, or my own earlier outlook, we
might treat the preceding examples as confirmation that scientific papers distort
the investigations they report. That view assumes, however, that there inheres,
in the experimental activity itself and in the mind of the investigator, a "true"
investigation before he or she begins to "write it up." Once we view writing,
thinking, and performing experiments as mutually interacting, our picture
changes. Whether a scientist is in the habit of beginning papers long before hav-
ing obtained all the results he or she expects to include in it, or waits until
reaching a conclusive point before putting words on paper, the writing is an
integral aspect of the shaping of the investigation. Ongoing research produces
ever-changing clusters of experimental findings, associated ideas, and leads for
further work in various possible directions. Only when the scientist has drawn
from these clusters what he or she is ready to treat as the solution to a particular
set of problems, decided what to include and exclude, and knitted evidence and
inferences into a structured argument for his or her conclusions, has he or she
completed a coherent, bounded investigation. Far from being a distorted version
of something the scientist has already done, the scientific paper is an essential
phase of the investigation itself. Writing the paper is a central act within the
creative process.
Notwithstanding this part that the writing of a scientific paper may play in the
formation of a scientific investigation, the paper does become, when it is itself
completed, a report of that investigation. What form do such reports take? Ac-
cording to Medawar, the paper is a misleading report of the reasoning through
which the scientist has reached the conclusions embodied in it. In one sense his
view is not pertinent. To the audience for which the paper is intended-that is, to
colleagues in the field-the manner in which a scientist has arrived at his or her
conclusions afterward becomes irrelevant. He is expected to present the best
arguments and evidence that he can adduce in support of his claims, regardless of
whether these were the means by which he originally came to them. In that
respect a scientific paper is not intended to be a description of the course of an
investigation, but of the investigator's findings. On the other hand, the paper is
also inescapably a report of past events, because its readers assume that the data
it incorporates derive from experiments that the author has actually performed at
some real time. Peter Dear has recently made the interesting suggestion that the
form of the modern scientific article, as a "research report," originated during
the seventeenth century as a result of the need to substitute a new form of au-
thority for the scholastic mode of textual authority that was rejected during that
period. Basing his case largely on the example of the publications in the early
Transactions of the Royal Society, Dear argues that the members of the Society
expected contributions to knowledge to be validated by their presentation not as
generalized observations, but as accounts of particular experiences that had hap-
pened to the authors or to some other accredited witnesses. In order to lend
verisimilitude to such reports, Fellows of the Society tended to include numerous
circumstantial details concerning what they had observed or the experiments

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
230 FREDERIC L. HOLMES

they had performed. To ensure that they would be accepted as real, such experi-
ences were habitually linked to specific times and places. So powerful was this
rhetorical practice, Dear points out, that some authors, including even Newton,
supplied dates, places, and other contingent circumstances that were fictional.14
Whether Dear's interpretation fully explains the emergence of the research
report as the characteristic form of the scientific articles that appeared in the
Transactions, the Memoirs of the Academy of Science, and later more special-
ized journals-or whether we prefer Steven Shapin's variant on this view, that
highly detailed descriptions of experiments were intended as a form of "virtual
witnessing" to secure assent'5-the reporting of observations and experiments as
historical events did become customary and has, in greater or lesser degree,
remained so ever since. As individual experiments evolved into more extended
research enterprises, it would follow that the report would have evolved into an
account of a series of events-that is, into a connected narrative.
We ought to look, then, at the extent to which the scientific research paper has
been cast in the form of a historical narrative. I have not done so systematically,
and here I can only make some suggestive remarks based on an examination,
from this point of view, of a scattering of papers that have interested me for other
reasons.

The two landmark memoirs On the Digestion of Birds, by Rene Antoine Fer-
chault de Reaumur, published in the Memoirs of the Academy of Sciences in
1752, are exemplary narrative descriptions of a scientific investigation. After an
introductory discussion of the unresolved issue of whether digestion occurs
through trituration or the action of a solvent, Reaumur pointed out that "there
are experiments that no one has thought of" capable of settling the question.
Then he began describing experiments that he had carried out, first on birds with
gizzards, to determine how much force their stomachs could exert on hollow
tubes and spheres he made them swallow. He began his report at the beginning of
his investigation: "The first bird that I forced to swallow these hard pills," he
wrote, "was a turkey." Reaumur described this first experiment, as well as the
subsequent ones, in minute detail. He mentioned the hours at which various
events within each experiment occurred, described the exact changes in shape
undergone by each tube or sphere within the stomachs of the birds, and included
many other circumstances of each trial. Although he made it clear that he was
not describing every experiment he had carried out, his text conveys the strong
impression that he included every one that differed in any significant way from
its predecessors. The reader can follow him step by step as he examines the
results of one experiment and modifies his procedure to overcome its shortcom-
ings in the next. In the second memoir he detailed his experiments on a bird of
prey, whose membranous stomach could not crush hollow objects but contained
a solvent that could dissolve pieces of meat suspended inside them. Reaumur
answers, one by one, potential objections to each experiment by changes incor-
porated into subsequent ones; he moves purposefully through a myriad of contin-
gent obstacles toward a successful conclusion, deepening his definition of the

14Peter Dear, "Totius in verba: Rhetoric and Authority in the Early Royal Society," Isis, 1985,
76:145-161.
15Steven Shapin, "Pump and Circumstance: Robert Boyle's Literary Technology," Social Studies
of Science, 1985, 14:481-520.

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
SCIENTIFIC WRITING 231

problem as he goes. The series of experiments on the buzzard ends with a very
real event-"My buzzard died before the continuation of the experiments for
which I intended it had been carried out"-but Reaumur is able to carry on with
experiments on artificial digestion and on other animals, significantly extending
the scope of his investigation. At the end of the two memoirs the reader feels that
he or she has been with Reaumur through a scientific adventure. They are an
effective argument for his momentous conclusions and, at the same time, a vivid
story.16
Despite the air of reality about this story, there are indications that Reaumur
did not relate it artlessly. There can be no doubt that the experiments described
so carefully were real; but it is less certain that the order in which he described
them (he gave dates for only one or two), and the connections he drew between
them were not in some cases reconstructions made after the completion of the
investigation. Sometimes the language was ambiguous enough to leave in doubt
whether he was recalling the reasoning he had actually followed or was imposing
a logic that was clear only in retrospect. For example, when he wrote, "Fortu-
nately it is easy to conceive of an experiment that shows us what one must think
of the preceding difficulties," and then described an experiment that overcame
these difficulties, he seems deliberately to have avoided the explicit claim that he
had thought of the experiment in that way when he undertook it. In the absence
of the original record of his experiments that Reaumur must have kept, we have
no way to ascertain the extent to which he may have shaped his story in telling it.
Another memorable research paper in the history of physiology that is written
entirely in a narrative mode is Frangois Magendie's initial report on "Experi-
ments on the Functions of the Roots of the Spinal Nerves," published in his own
Journal de Physiologie in 1822. In style it contrasts sharply with Reaumur's
memoirs. Where Reaumur was prolix, repetitive, exhaustive in description, Ma-
gendie was succinct, elegantly spare, including only the minimum of detail neces-
sary to make his point. Yet Magendie was equally successful in drawing the
reader into a compelling story. The clarity, economy, and concentrated focus of
this brief paper, a bare three paragraphs long, make it, I believe, one of the
classics not only of scientific discovery but of literary style. I can do no better
than to quote the first and last paragraphs in translation:

For a long time I have wanted to perform an experiment in which I would cut the
posterior roots of the nerves that arise from the spinal marrow of an animal. I have
attempted it many times without success, because of the difficulty of opening the
vertebral canal without injuring the marrow, and consequently without causing the
animal to perish or at least gravely wounding it. Last month someone brought into my
laboratory a litter of eight puppies, six weeks old. These animals seemed to me very
suitable for a new attempt to open the vertebral canal. With the aid of a very sharp
scalpel I could, in fact, almost in a single cut, expose the posterior half of the spinal
marrow surrounded by its membranes. In order to strip that organ almost bare, there
remained for me only to cut the dura mater that surrounds it. That I did easily. Then I
had before my eyes the posterior roots of the lumbar and sacral nerve pairs, and by

16Ren6 Antoine Ferchault de R6aumur, "Sur la digestion des oiseaux. Premier Memoire. Experi-
ences sur la maniere dont se fait la digestion dans les oiseaux qui vivent principalement de grains et
d'herbes, et dont 1'estomac est un g6sier," Mem. Acad. Sci., 1752 [1756], pp. 266-307; and Reaumur,
"Sur la digestion des oiseaux. Second Memoire: De la maniere dont elle se fait dans 1'estomac des
oiseaux de proie," ibid., pp. 461-495.

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
232 FREDERIC L. HOLMES

lifting them up one by one with the blades of a small scissors, I could cut them on one
side while leaving the spinal cord intact. I did not know what the result of this attempt
would be. I closed the wound with a suture in the skin, and I observed the animal. At
first I believed that the limb corresponding to the severed nerves was totally para-
lyzed. It was insensitive to the strongest pricking and pressures, and it appeared to
me also immobile. Soon, however, to my great surprise, I saw it move in a very
obvious manner, even though it remained always completely insensitive. A second
and a third experiment yielded exactly the same result. I began to consider it probable
that the posterior roots of the spinal nerves might have different functions from the
anterior roots, and that they were intended more especially for sensibility.'7

After describing in similar style how he had performed the more difficult oper-
ation of severing the anterior roots, observing then that the limb became immo-
bile but retained its sensibility, Magendie went on to his concluding paragraph:

I have repeated and varied these experiments on several species of animals. The
results that I have just announced have been confirmed in the most complete fashion,
both on the fore and the hind limbs. I am pursuing this research, and I shall give a
more detailed description in the next issue. It is sufficient for now to be able to
propose that the anterior and posterior roots of the nerves that arise from the spinal
cord have different functions; that the posterior appear more especially intended for
sensations, whereas the anterior seem to be more particularly linked with move-
ment. 18

As was the case with Reaumur, the narrative structure of Magendie's paper
provides in the telling a forceful argument for the conclusions he reached. The
line of reasoning in this paper, written soon after the event, seems indistinguish-
able from the course of the investigation itself. Obviously Magendie did not tell
the whole story. He explicitly omitted both his early unsuccessful experiments
and his later repetitions of the successful ones. We would like to know much
more about why he had "long wanted" to perform these experiments. He has
condensed his narrative to what he regarded as its essential elements, elements
that he could select only in hindsight. Yet there is something so direct about his
account that we feel intuitively that this is not merely a rational reconstruction of
events that might have occurred in a quite different manner.
The structures of Lavoisier's memoirs are typically more ambiguous than the
two examples I have just discussed. Earlier I pointed out that Lavoisier created
his investigation of the composition of three plant substances partly on paper, by
putting together experiments that he had performed at different times for differ-
ent purposes. If we read the final version of his memoir "On the Combustion of
the Oxygen Principle with Spirit of Wine, Oil, and Different Combustible
Bodies" with an awareness of how it was constructed, we find no statements
incompatible with its genesis from disparate elements. Read without such knowl-
edge, however, the work presented appears to be a coherent experimental inves-
tigation. Lavoisier proceeds from one substance to the next, treating all three in a
similar manner, describing for each the experimental procedures, and ending
each discussion with a table giving the quantitative composition that he had de-
termined for the substance.

17Fran9ois Magendie, "Experiences sur les fonctions des racines des nerfs rachidiens," Journal de
Physiologie, 1822, 2:276-279, on p. 276.
18Ibid., p. 279.

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
SCIENTIFIC WRITING 233

The individual accounts of the experiments that Lavoisier described within the
sections devoted to each substance also appear ambiguous. They seem in part
generic descriptions of a type of experiment, in which the use of the impersonal
French pronoun on suggests that these are directions for such an experiment that
others might carry out. The data, on the other hand, are very specific- including
Lavoisier's usual long numerals-and are clearly the results of single experi-
ments previously performed. Moreover, narrative elements appear in the midst
of the more general descriptions, and when they do, Lavoisier switches from on
to je; that is, he tells what he himself had done. For example, in the section on
alcohol he wrote: "I decided at first to operate only in atmospheric air, to avoid
the danger of an explosion; but, since I could not hope to maintain the combus-
tion for a long time in that way, because of the large quantity of vital air that
alcohol consumes in burning, I arranged the apparatus in such a way as to supply
vital air in the measure that it is consumed."19 After this interpolation he re-
turned to an impersonal style to describe the procedures that he had settled on
and the experiments whose results he reported.
The order in which Lavoisier proceeded from alcohol, to olive oil, to wax in
the memoir is systematic. He might have followed that order of investigation, but
as we have seen, the record shows that he did not. Once alerted to that fact, we
can see that in his paper he did not connect the three substances in a narrative
order at all. The sections merely follow one another in the text.
The hybrid form of this paper-Lavoisier's almost imperceptible shifts from a
narrative to an analytical mode of description-is, I believe, typical of scientific
research papers. The duality reflects the dual function of the scientific paper, as a
research report and as the argument for a conclusion. For their own purposes
scientists have no need to sort out these elements in their writing so as to provide
a more homogeneous literary form. To distinguish the narrative elements from
the post facto reconstructions is, however, one of the central challenges to the
historian of scientific activity.
The examples from the eighteenth and early nineteenth centuries that I have so
far mentioned may seem remote from the style of the typical scientific paper of
the twentieth century. The standard format that the editorial imperatives of mod-
ern specialized journals impose on individual authors, the mandated division into
sections called "Introduction," "Methods," "Results," and "Discussion," and
the nearly universal description of investigative operations in the passive voice
may appear to have suppressed the narrative mode visible in scientific papers of
earlier eras. Beneath this format, however, I believe the ambiguity still lurks. For
convenience I will illustrate the point with the same (first) paper by Hans Krebs
on the synthesis of urea that I discussed earlier.
Krebs organized his paper into ten sections, with headings such as "Method,"
"Units of Measurements," "Site of the Formation of Urea from Ammonia," and
"Formation of Urea from Arginine." The individual sections describe the results
of particular aspects of the overall investigation of urea synthesis that he carried
out over nine months, but the order of discussion bears no relation to the histori-
cal order in which he pursued these aspects of the investigation. Nor can the
investigation as it is recorded in his laboratory notebook be broken down into

19Lavoisier, "Sur la combinaison du principe oxygine" (cit. n. 10), p. 589.

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
234 FREDERIC L. HOLMES

discrete segments corresponding to these subheadings. Experiments that he now


grouped in this way had originally been interwoven in more subtle ways within
an ongoing research stream.20 In the sections concerned with "the effect of or-
nithine on the formation of urea from ammonia" he referred explicitly to the
results of only two or three of the dozens of experiments he had carried out on
that topic, just enough to infer the theoretical conclusion that he had drawn.
Then he presented his theoretical "explanation" for these results in a compact,
tightly reasoned argument.
Read in isolation, this paper would appear to have no narrative content at all.
The organization and the mode of discourse are almost entirely analytic; it is a
discussion not of investigative events, but of experimental findings. When we
view the same paper after reconstructing Krebs's investigative pathway from his
laboratory records, however, we can perceive in his text telltale traces of those
events. When he wrote, "We find with ornithine a special, unexpected effect on
the formation of urea from ammonia," he was alluding to the fact that he himself
had not been expecting to find such an effect until he encountered it on 15 No-
vember 1931. When he stated, a little further on, "Closer investigation showed
that the additional urea formed in the presence of ornithine does not derive from
the ornithine," he was making vague reference to a sequence of experimental
events and reasoning that had led him from an initial idea that ornithine may be a
donor of urea nitrogen to the view that it acts like a catalyst. His statement "We
have investigated many substances, but we find none that can replace ornithine,"
sums up experiments that he and his coworker had carried out intermittently
through much of the course of the investigation until the substances tried had
added up to "many." Finally, his statement "It turned out that even traces of
ornithine have large effects" refers to an experimental denouement that Krebs
had just reached, after repeated efforts, at the time he composed that sentence.21
Thus, even though the stylistic norms for modern scientific papers nearly pre-
clude a direct narrative form of presentation, there are inevitably narrative
aspects of a discussion of results, because such papers embody, as they have for
three centuries, both a current argument for certain conclusions and a descrip-
tion of research that the author has carried on through a prior period of time.
How scientific writing helps to structure the investigations that scientific
papers present can be elucidated fully only when we have available independent
perspectives on the other events that enter into the investigations. I have found
that laboratory notebooks can provide particularly illuminating sources for such
perspectives. When we can follow the daily experimental operations that the
scientist who writes the papers has carried out, or is simultaneously carrying out,
then we can discern how he or she reorganizes what has been done, selects and
discards, connects and interprets results, so as to transform portions of ongoing,
open-ended enquiries into discrete investigative units, into solutions for well-de-
fined problems, or into landmark discoveries.
Once I would have said that through such records we can reveal the actual
historical investigations that lie hidden behind the published papers. If, however,
the writing of the papers is itself part of the investigation, then that is not an

20Krebs and Henseleit, "Harnstoffbildung" (cit. n. 9), pp. 757-759. See also Frederic L. Holmes,
"Hans Krebs and the discovery of the Ornithine Cycle," Federation Proceedings, 1980, 39:216-225.
21Krebs and Henseleit, "Harnstoffbildung" (cit. n. 9), pp. 757-759.

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms
SCIENTIFIC WRITING 235

accurate way to express the situation. What we can do is to reconstruct from


such records a narrative story that reveals some of the deep aspects of scientific
investigation that are not visible on the surface of the investigation that normally
becomes public.
There is another, more fundamental reason why we should not regard such
historical reconstructions as depicting the "real" investigation or the "true" pro-
cess of discovery. In agreement with recent defenders of the value of narrative
history, I believe that the stories we can reconstruct from any type of documen-
tary evidence are not inherent in the documents but are in part created by the
imagination of the historian. Laboratory notebooks are in some ways ideal docu-
ments on which to base investigative narratives, because the normal practice of
dating each day's operations provides a natural chronological backbone on which
to build. These records provide, however, only traces of an ordered series of
individual episodes. The historian must imagine the connections between one
experiment and the next, must provide the configurations that unite these daily
episodes into larger units of meaning. The historian defines a starting point and a
stopping point that seem to him to enclose a story with a beginning, an ascending
development, and a resolution. Just as the scientist has once used these docu-
ments to construct the logical, rational investigation set forth in the research
paper, so the historian afterward can reconstruct from the same material an in-
vestigative pathway that may help us to understand some of the springs of the
scientist's creativity.
Not only must the historian interpret laboratory documents imaginatively in
order to elicit a story from them, but the type of story told is determined in part
by his or her own investigative goals. In this respect I am influenced by Hayden
White's view that the story a historian tells is shaped in part by a guiding meta-
phor.22 In the case of my own research program, I have frequently described my
aim as understanding the nature of the "research trails" or the "investigative
pathways" of certain creative scientists. I have also called my studies examina-
tions of the "fine structure of scientific activity." These phrases are, however,
spatial metaphors for processes in time, whose generalized character I cannot
describe in more literal language.23 Undoubtedly I find fine structure in the daily
operations recorded in notebooks, or connect these operations into pathways and
trails, in part because of my belief in the appropriateness of these metaphors.
The historical narratives that I, or others with other guiding ideas, can produce
from such records are imaginative constructions, just as are the scientific writ-
ings that the scientists who kept the records produce from them. Such historical
narratives can, as Medawar put it, help to reveal what scientists do; but they are
not more "correct" descriptions of scientific investigations than are the discus-
sions that appear in the writings of the investigators.

22Hayden White, Tropics of Discourse: Essays in Cultural Criticism (Baltimore: Johns Hopkins
Univ. Press, 1978), pp. 51-80.
23On the prevalence of spatial metaphors see George Lakoff and Mark Johnson, Metaphors We
Live By (Chicago: Univ. Chicago Press, 1980), pp. 14-21.

This content downloaded from 193.54.110.35 on Sat, 07 May 2016 09:49:34 UTC
All use subject to http://about.jstor.org/terms

You might also like