Animals in Experimental Reports: The Rhetoric of Science

Lynda Birke 1
Centre for the Study of Women and Gender,
University of Warwick, United Kingdom

Jane Smith
University of Birmingham, United Kingdom

In this paper, we analyze the ways in which the use of animals is described in the "Methods" sections of scientific papers. We focus particularly on aspects of the language of scientific narrative and what it conveys to the reader about the animals. Scientific writing, for example, tends to omit details of how the animals are cared for. Perhaps more importantly, it is constructed in ways that tend to minimize what is happening to the animal; thus, animal death is obscured by euphemisms, omission, or circumlocutions. What is done to animals is, moreover, often subordinate in the text to the details of experimental procedures and apparatus. We consider how such writing supports a particular kind of image of the "animal" in science, and also creates an impression that what happens to animals is somehow devoid of human agency. This impression, we argue, contributes to the way science is perceived by a wider public.

Amid renewed controversy around animal use in experimental science, both scientists and their critics must respond to public opinion. Since the mid-1970s, there has been such growth in antivivisectionist feeling that we can speak of a new social movement for animal rights, and scientists can no longer dismiss the opposition as merely a lunatic fringe, as they might once have done. Increasingly, they must act to protect their laboratories and animal houses, and engage in public rhetoric. They must, then, "engage in a battle for public opinion, seeking to undermine the animal rights movement by associating it with violence and terrorism, and to improve their own public image by defending the value of their work. And they try to steal back the moral high ground with emotional rhetoric. In effect the scientists have developed a counter-crusade" (Nelkin and Jasper, 1992, p. 38).

As part of that counter-crusade, more than 1,000 people signed a declaration in 1990 on scientists' use of animals, organized by the British Association for the Advancement of Science. The subsequent short report (BAAS; 1993) stressed the benefits to medicine of animal research, and included a special section on the importance of using primates for certain areas of research such as vaccines (AIDS is mentioned here) and Alzheimer's disease.

A key strategy of the report is to emphasize that scientists have clear responsibilities in using animals. "If there were a Hippocratic oath for scientists," the report argues, "it would surely include two primary responsibilities that a research worker must accept when planning animal-based studies. The first is to use animals only for research intended to contribute to the advancement of knowledge. The second is to minimize any possible pain or distress that an animal in the scientist's care may experience. Those are two of the key criteria taken into account when deciding the scientific merit of a research project and before a request is made for funds for the research" (BAAS; 1993, paragraph 5).

The BAAS scientists' rhetorical counter-strategy employs two clear tactics. One is the stress placed on responsibility and on using animals only for certain purposes and under certain conditions. Such rhetoric is designed to reassure waverers in the controversy that animal experiments are not done for trivial reasons (as is often claimed by animal rights protesters).

A second tactic is to attribute lack of understanding of science to the public. Many of those scientists most outspoken in defense of animal use for biomedical research believe this increase in concern to be part of a wider anti-science feeling (or even anti-intellectualism). This complaint is frequently linked to the belief that the general public does not know, or understand, what science is about - and specifically how laboratory animals are kept or what happens to them (Arluke, 1992; Birke and Michael, 1992). In other words, if the public is ill-informed, then it must be better informed. Thus, a letter to the British Medical Journal , defending animal use, remarked that "It is now time for the public to be continually reminded by doctors that the tremendous developments of modern medicine in the past 50 years...would not be available but for animal experimental work" (Drury et al., 1990). Similarly, Mark Matfield, Director of Britain's Research Defense Society, suggested that the medical benefits "are quite apparent to those engaged in biomedical research but are not appreciated by the general public at large. This lack of understanding leaves them prey to the propaganda currently being produced by the animal rights movement" (Matfield, 1989).

Yet if "the public" is being misled, as is claimed, where would it get better information? Clearly, propaganda from either side is one potential source; so too are media stories, lurid or otherwise. In turn, both journalists and the antivivisectionist organizations base much of their information on the written accounts, the scientific papers that emerge in the published journals. Unlike brief conference presentations, published journal accounts give at least some details of the methods employed in animal-based research.

For example, the antivivisectionist organization Animal Aid produced a detailed, and critical, report of animal use in British university laboratories, based on an analysis of over a thousand scientific papers (Wojtas, 1993). Scientific reports in journals are of course written in specialist language, aimed at the small number of scientists who are engaged in similar research. 2 Nevertheless, they do in some ways represent a particular part of the public face of science - a face which is increasingly analyzed by those organizations, such as Animal Aid, who wish to present to the public particular images of both animals and science.

Analyzing the Writing of Science

If written accounts do play a role in communicating science, albeit to a relatively small audience, then how they describe what happens to animals is an important issue. By analyzing written accounts, it is possible to assess what scientists (and journal editors) consider important to communicate, as well as what they have omitted for whatever reasons.

To do this, we decided to analyze a sample of journals publishing research in biomedicine to determine what was said about procedures using animals in the "methods" section. There are two reasons for focusing on the methods: first, this is the part where what is done to the animals used should be set out. Indeed, some journals now specifically issue guidelines to authors indicating what should be spelled out in terms of animal care and procedures. Secondly, it is the methods section that, in principle, sets out the details of methods followed so that others could follow the procedure and perhaps replicate the experiment.

What is published eventually is influenced by journal (and editorial) policies, as well as by the assumptions made by reviewers in the process of review. Yet there is also likely to be some convergence in the way the methods are described: as Gross (1990) has pointed out, scientific papers reflect conversations among scientists, thus revealing much about the assumptions they make. They also reflect traditions in learning how to "write science" - the habitual use of the passive voice, for example. Gross describes scientific papers as "the most visible products of verbal interaction within the community, posed photographs of a continuous activity captured at certain ritually significant moments" (p. 129). They are also, he notes, "photographs" that are written after the event, and may omit or transform many of the events. These "photographs," and their arcane style, can be analyzed not only for the language they use to describe the practices of science, but also for how their language positions the reader: how are these texts supposed to be read by other scientists?

The Study

This paper is based on a qualitative analysis of the language used to describe experimental procedures using animals in a series of scientific papers. We have also analyzed these papers quantitatively, to assess how well codes of practice were adhered to, at least in the way the experiments were reported. These data showed that many papers failed to give important information on the methods, such as the numbers of animals used or the conditions of husbandry (Smith, Birke and Sadler, in press). Here, our concern is with a qualitative analysis of the use of language: 3 what is being stated in these papers about how animals are used, and what impressions do they create about laboratory practice? What we are concerned with, then, is the power of this particular narrative (scientific writing) to construct particular images of what happens to animals.

For these studies, we sampled a total of 152 papers from eight journals published in 1990-91. The papers sampled were from the following journals: Life Science, Journal of Neuroscience, Journal of Immunology, Journal of Comparative Psychology, Journal of Experimental Biology, Journal of Surgical Research, Cardiovascular Surgery, and Laboratory Animals. 4

In doing this, we made the assumption that these papers would be broadly representative of scientific writing as a whole (although there might also be important differences between them); we have no reason to believe they are unrepresentative, and have no wish to single out individual papers. For this reason, we chose not to identify particular papers or authors here. 5

We selected journals for sampling that, in our judgement, provided a range of different kinds of experimental approach in biomedical research. Once a journal was selected, the choice of volumes was random (volume number being selected in advance). Within each journal, we analyzed all articles that reported animal use. The papers sampled originated in several different countries, with rather different controls over the use of animals in research. 6 Some of our material reflects these differences; other material probably does not. Scientists, after all, submit papers to journals based in other countries, whatever the constraints have been on the original practice. This, at least, might be expected to reduce differences between reports.

Descriptions of How Animals Are Used

Recent work in the sociology of science has emphasized the central role of the scientific paper - not as a post hoc report of "what really happened," but as a product, an end in itself. The writing of a paper, moreover, does not follow faithfully what happened, but is constructed to answer particular critics (see Latour, 1987). Yet at the same time, this research emphasizes that "doing technical work in science entails a kind of craft, or embodied expertise, which is not incorporated into the mode of methods reporting that is characteristic of scientific literatures" (Lynch et al., 1983).

Similarly, in a study of scientists working with chimpanzees, Wieder (1980) noted how researchers led a double life, assuming consciousness in the animals when interacting with them but denying it through the detachment of scientific writing. The informal, or craft, knowledge of chimp behavior had to be assumed in the design of experiments: yet these details were "likely to have no place in the standard report of the experiment" (ibid., p. 97).

Several themes emerged from our qualitative analysis of research papers; one is the way in which relevant details about the care of the animals were left out - precisely the kind of "craft expertise" of animal care to which Lynch and his co-workers referred. Perhaps more importantly, the deaths of the animals were minimized or omitted, even when this was a critical detail. Other themes were the ways in which "mistakes" in procedures involving animals were dealt with in reporting, and the ways in which the experiences of the animal become subordinated in scientific texts to the descriptions of the technical procedures.

Omitting Details

One striking point about these papers emerged from the quantitative analyses: relatively few gave full details of numbers of animals used, or how they were housed. We give details of these numerical data elsewhere (Smith, Birke and Sadler, in press). Omission is, rhetorically, one way of obscuring; what it does is to gloss over both the exigencies of what happens to the animals and any methodological deficiencies. It is difficult, for example, to decipher what is happening, and to how many animals, when a paper refers only to "rats ...were decapitated." Sometimes, we could infer numbers from results; in this particular example, that inference was impossible, and results were presented only in terms of quantity of molecular binding. Omitting details matters, too, in terms of scientific quality. Immunological studies that refer vaguely to some animals being infected "in another lab," for example, fail to tell the reader how the animals were transported between labs, and when. Yet transport is likely to be stressful, and therefore to have an effect on immune responses. 7

Obscuring Death

One aspect of scientific narrative that has attracted attention is the use of euphemisms and other devices for minimizing or altering the significance of the death of animals. First, several studies have noted how often scientists (and animal technicians) refer to "sacrifice" rather than killing (at least in written accounts), drawing the parallel with ritual sacrifice (Arluke, 1988; Lynch, 1988; Phillips, 1994). The point here is not only that the word "kill" is avoided, but also that invoking the concept of "sacrifice" changes the meaning. The "sacrifice" of laboratory animals symbolically transforms them; they are sacrificed in order to create scientific data.

In our study, we found that approximately half the papers used the word "killed" if animal death was explicitly mentioned. Where this word did occur, it was usually followed by the method or treatment immediately following death: thus, animals might be "killed by cervical dislocation," "killed by carbon dioxide," or "killed and perfused." In other papers, the method of "sacrifice" was noted - by decapitation or dislocation, for example. More commonly, however, animals were simply "sacrificed." An alternative form of euphemism was simply not to refer directly to death at all. Thus, animals might be "anaesthetized and then exsanguinated," or perhaps "bled and ... sacrificed."

One interesting euphemism was "harvested," used occasionally to describe collection of bodily tissues, such as lymph nodes. Like "sacrifice," it carries particular cultural connotations; "harvesting" seems rather benign, conveying images of rural bounty. Yet this, too, obscures and changes meaning. Consider, for example, the paper that tells us that amniotic fluid "was harvested from CO2-killed donors" near the end of pregnancy. Here, the pregnant female becomes a donor; her fetuses inevitably die in the process, but their deaths are obscured by the language of harvests.

Terms invoking sacrifice were predominantly used to describe experiments in which death was a necessary precondition to experimental procedures (in other words, the experiment was concerned primarily with the tissues that were removed from the animal after death - as in biochemical investigations). Other experiments, however, may involve physiological investigations of the animal under terminal anesthesia (the animal is usually given an overdose of the anesthetic at the end of the experiment); such studies may or may not entail subsequent removal of particular tissues.

Descriptions of these experiments reveal a second form of minimizing death. In describing experiments using terminal anesthesia, the papers were not euphemistic; they simply did not specify that the animal had died. In some cases, it is quite obvious that the animal was dead, because procedures are described that are incompatible with life. One paper, for example, tells the reader, for example, that "following anaesthesia with pentobarbital sodium...heparin was injected into the inferior vena cava. The heart was rapidly excised, and the aorta cannulated." The animal obviously died during these procedures, but its death is hidden by the language used.

In other cases, the death of the animal is less obvious. One, for instance, states that "mice were reanesthetized with methoxyfluorane .. and the soleus muscle with a 4mm length of motor nerve was removed for study." Another experiment reports the use of silicon implants into guinea pigs: at the end of the experiment the "implant and its surrounding tissue" were removed en bloc and examined histologically.

The third device for minimizing the impact of death on the reader is to use circumlocutions; thus one paper informs us, for example, that "the 10-day mortality of this local bowel insult is approximately 50%" - that is, half of the animals (10) were dead by 10 days after the surgical procedures. Another paper reported that 24 hours after surgery, all animals "appeared clinically septic." The results section of the same paper, however, revealed that seven of the animals were dead. Similar - if curious - circumspection is revealed by the paper that referred to "sacrificing" the animals by craniocervical dislocation when "they were moribund."

Sometimes, circumlocutions help to gloss over both the severity of the operative procedures used and the deaths of animals; one way of doing this is to use the passive tense, combined with words of Latin origin. Thus, one paper reporting a new surgical technique referred to 31 dogs, of which four were "lost" through "intraoperative exsanguination." Of the remaining 27, only one survived more than a few days. This animal was then "electively sacrificed" after three years.

A further example of the use of the passive voice is the phrase "animals were allowed to survive." Thus, rats in one study "were allowed to survive" for three to five days after injection of radioactive tracers directly into the pancreas. In another, mice were "allowed to survive" for variable amounts of time following limb amputation.

Mistakes and the Language of Procedures

Using euphemisms for death, not mentioning it at all, and circumlocutions all have the effect of reducing the impact on the reader. The death of the animal is somehow downplayed; it has to be actively read into the text by the reader. On the one hand, this strategy may simply be part of taken-for-granted practice; much of what happens to laboratory animals is part of the "tacit knowledge" of science. But on the other hand, the rhetorical denial of animals' deaths is also a response to an ethical dilemma.

Scientists may use similar devices to describe the potentially painful procedures that are carried out on the living animal. Here however, rather more information may be given: there is less omission because these are procedures that are critical to the experiment itself (contrast the example of terminal procedures where the death of the animal is incidental to the experimental protocol).

Even so, two aspects of the style used in description of experimental methods are striking. First, what is done to the living animal typically merits far less detailed description than what happens to its tissues after removal. Then, the tissues go through various processes of preparation (depending on the experiment) which require technology - they may need to be centrifuged, or fixed and stained, for example. At this point, fine details of the apparatus are given - model, running speed, and so on. 8 Omitting details of what happens to animals may entail the device of referring to previously published methods. Thus, a paper might describe how "single intracerebroventricular injections were performed in conscious rats following the methods of X and Y" before giving all the technical details of the stereotaxic apparatus used to hold the animals' heads.

Secondly, the experience of the animal itself may be obscured by descriptions of methods of measurement and surrounding apparatus. In one example, of a study of analgesics:

...Rats were placed in .. holders and their tail was [sic] secured on the laboratory table with masking tape. The pain threshold was determined by electrical stimulation of the tail via intracutaneous electrodes .... The electrodes were two hypodermic needles ...[which] remained in place for the six hour measurement period ...[the voltage was increased] until a vocalization was elicited.

Another study referred to preparation of rats for chronic recording involving electrical wires that were "exteriorized at the head, and the connector was bonded to the cranium." It is noteworthy that, in both these instances, what is actually happening to the animal is subordinated to the descriptions of what was done in order to make measurements.

The descriptions of procedures also included several instances of carelessness, or mistakes. To become part of data, animals have to die as part of the experimental procedures: animals that merely "upped and died" cannot be counted as contributing towards data (Lynch, 1988). 9 So, animals that cannot contribute to results have to be accounted for. There were several examples of "mistakes" noted in the papers we sampled. One surgical paper, for example, described work involving the use of carbon dioxide injected into blood vessels as a contrast agent (a technique used in clinical angiography). Carbon dioxide is soluble in plasma, and is therefore relatively harmless (under specified conditions of partial pressure and volume). Yet one dog received an injection "of room air into the ascending aorta." Not surprisingly, it died very shortly after.

Another example had less to do with mistakes in procedure than with failures of planning: thus, "the final N was less by 1-2 in some cases due to an inability to assay every sample." Some animals, presumably, were killed but their tissues not used, for reasons unspecified. 10

The difference between reporting and not reporting such events may partly depend upon the design of the experiment, and the stage in its execution at which the mistake occurred. Thus Lynch (1985) described one instance in which a researcher had ruptured an artery while lesioning a rat's brain; because he succeeded in stemming the bleeding and doing the lesion, the animal would later become part of the data and, presumably, the accident not recorded. But in some instances, a mistake may alter the total number in that group (which could unbalance the overall design or alter the statistical analyses). So, in that case, it may be preferable to note the mistake (as in the examples we found).

Implicit ideology may also affect the decision. In two of the examples we looked at, the problem occurred after the animals' deaths (such as "inability to assay"). In these cases, the ethical problems concern wastage rather than potential suffering caused to the animals. But what is important here is not so much the squandering of animals per se, but the emphasis that is placed on wastage of potential data. Lynch has described how biologists make use of two concepts of "animal" in their work (often interchangeably). One recognizes the "naturalistic animal, the animal of common sense" (1988); this is not the "animal" whose waste is at issue in the papers we analyzed. The other is that of the "analytic animal," transformed into data (or into artifacts such as electronmicrographs): this is the animal whose waste is noted. 11 The use of the word "animal" often confounds the two meanings. Consider the impact on a non-scientific reader of the language used in the following extract:

After sacrifice, the injections were verified; animals not showing the trace of the needle inside the ventricle were discarded (5-10%).

We read here that some animals were subsequently "discarded," not used as data (helpfully, a percentage appears: as the number of animals used in the experiment was not given in this paper, however, it is not clear what the percentage represents). What this means is that the scientists examined the slices of brains for needle traces. If some did not show traces (that is, the scientists failed to penetrate the required part of the brain), then all slices from that animal were thrown away. The animal, in short, died in vain.

One characteristic of the animal-as-data is that it is less variable, less messy, than real, naturalistic animals. Gross (1990) notes how the use of tables, figures and so forth in scientific reports imposes invariance on the picture of nature that is being constructed. 12 These form a pictorial representation that is "a triumph of simplification...; not mice, but their brain cells...In tables and figures most of the properties of the actual physical objects, of mice or men, have been discarded, and all that remain have been normalized, ideally through quantification" (pp. 74-5).

Animals and Readers in Scientific Text

In general, animals appear in scientific writing as "analytic animals"; they are described as "preparations," or "animal models," for example. Where they do appear as subjects, it is typically in experiments studying the behavior of the living animal. The contrast between "animal" and "subject" was striking in one paper, whose methods section had separate headings for each. In this case, the investigation involved comparison between different species: "subjects" here were humans, while animals included several species of Macaque monkeys and rats.

The methods section of scientific papers is, in principle, included for two reasons. The first is that it allows readers to check the veracity of any claims made elsewhere in the paper. Secondly, by describing the methods employed, scientists allow the possibility of replication. Complete replication is perhaps rare; but its possibility is always assumed in science.

Yet the methods sections that we analyzed rarely gave enough information for replication to be easily achieved. In particular, too little information was given in many cases about the origin of the animals and how they were maintained: details, for example, of humidity and ambient temperature were frequently not given (Smith, Birke and Sadler, in press). Partly, this reflects an unwritten assumption that animals are maintained in standard conditions (also unwritten, at least in scientific papers); in this sense, it is testimony to the tacit knowledge that underpins science and to the assumption that animal laboratories house standardized, invariant, tools of the trade.

To write of animals as though they are laboratory artifacts is made easier by the temporal and spatial separation of animal and scientist. Once the experiments are completed it is only "useful" data that count; it is these which will be included in the written text. The methods, moreover, are purged of the contexts that give them meaning, and particularly of the process of decision-making that, in practice, govern laboratory activities: all that is left is a residual description of uncontested features (such as the brand of equipment used: see Knorr- Cetina,1983). The animals that gave rise to these data are (usually) long since dead.

There is spatial separation, too. Many laboratories share a centralized area maintained by specialist staff. Scientists themselves may spend little or no time in that facility; instead, animals or their parts may be sent to laboratories or collected from the animal unit. 13 In that context of separation, it is hardly surprising that what emerges in written reports bears little resemblance to the living animals "of common sense." 14

Diminishing Tactics

One characteristic of scientific writing lies in its appeal to authority, typically through extensive citation and appeals to allies (Latour, 1987). In addition, scientific papers comprise a network of defenses against possible criticism. "Transformation of linear prose into... a folded array of successive defense lines," notes Latour, "is the surest sign that a text has become scientific" (1987; p. 48). A clear example of both appeals to authority and the construction of "lines of defense" found in the papers we analyzed was the use of references to scientists adhering to guidelines set down by, for example, animal care committees. These rarely gave further information, but simply stated that guidelines were followed. This line of defense makes implicit reference to ethical questions concerning the use of animals in research. Yet nowhere else in the construction of scientific papers are ethical issues made explicit. Instead, the writing is itself constructed in ways that diminish the significance of the animal.

Gross (1990) suggests that the widespread use of the passive in scientific papers "is a routine means for making physical objects and events the subjects of scientific sentences" (p. 73). Phrases such as "the animals were injected..." are typical. Here, the animals are indeed the subject of the verb: but they do not become "subjects" in the more philosophical sense. What the passive voice does is to remove the scientist from the sentence, so reducing the emotional impact on the reader of what is done to the animal. This technique might, moreover, be combined with language that focuses the reader on some particular bodily part, so shifting attention from the whole animal. Consider, for example, the text of a paper noting that "When the nerve was crushed repeatedly with forceps...it became transparent." Here, we might imagine what the nerve tissue looks like. Only later do we learn (through the phrase "mice were allowed to survive") that these nerves were located in living animals.

Scientists do all kinds of things with the animals they use, and with their tissues; but, in the written account, human agency disappears. It is noteworthy, however, that in an analysis of the content of scientific papers, Bazerman (1988) noted that the uses of active verbs occurred almost entirely in connection with intellectual processes. The scientist may actively think and generate hypotheses; meanwhile, the animals are miraculously anesthetized or injected by an invisible hand.

Another diminishing tactic is to talk in phrases such as "the guinea pig gut." In this case, a small fragment of an animal is isolated, and measurements taken; what constitutes data are the product of the various machines that are linked up to the animal fragment (Latour, 1987, p. 66). The scientific paper must be written in such a way that the central role is given to these machine artifacts - the data. The living animal, whose "sacrifice" provided the fragment of gut, is unimportant.

In our study, similarly, priority was given in many papers to what happened to cells or tissues after scientists had removed them from animals. Few details might be given of animals and their housing; yet detailed accounts typically followed of how solutions were made up, at what temperature, how cultures were maintained, for how long, what type of centrifuge was used, and so on. We did find one entertaining slippage between animals and their cells in one paper. This reported an immunological study, using "nude" mice, from which the scientists took cells after the animals' deaths; typically, descriptions of what happened to the cells took precedence over the whole animals. After stating what animals were used, the report tells the reader that "At three months they were lethally irradiated with 900 rad from an X-ray source and reconstituted 4-6 hours later" - no doubt a feat that many scientists would wish to emulate.

The use of the passive voice implies a missing agent: there is no person who failed to use all the rats, merely "an inability to assay." Another paper similarly noted that "data were not utilized from any heart preparation if more than 15 seconds elapsed from excising the heart." There is no one in this sentence who has failed to excise the heart promptly: 15 seconds merely elapsed.

This kind of construction not only leaves out the "missing agent." It also positions the reader. As Latour has argued (1987), scientific reports are written in ways that carefully hedge the writers' bets about critical comments. The examples we quote above illustrate this; a determined critic might note that the N was less than it should be. So, a comment must be included to ward off that criticism. But there is also the possibility of criticism because of mistakes; to deflect this one, the writer refers to an "inability to assay." The reader is, one assumes, supposed to infer some unforeseen catastrophe that prevented the scientists from carrying on. We are not supposed to infer that they went off for coffee.

Lynch (1985) notes how research reports are written so that "agency is left implicit, and if we read the passage literally we get the sense that a 'lesion' or 'sacrifice' is accomplished through an autonomous technology freed of the vicissitudes of human agency" (pp. 151-2). And not only is it freed of human agency: omitting agency also glosses over the exigencies of particular experimental procedures - the messiness of day-to-day science.

Doing Versus Writing Science

Ethnographic studies of scientists, and their conversations while they carry out procedures, have underlined the disparity between day-to-day science and the subsequent written reports. In one example, a perfusion failed partly because the animal was squirming and trying to bite the experimenter, leading to much swearing at the rat. In cutting open another animal, the experimenter severed a membrane around the heart. During these experiments, researchers made several "stabs" at animals with the syringe, or had to deal with leaks in the apparatus during perfusion. Yet, the research reports were written up to describe animals "sacrificed under nembutal anesthesia by intracardial perfusion utilizing a mixed aldehyde fixative media" (Lynch, 1988 p. 74). Similarly, when scientists leave animals anesthetized overnight, without supervision (in this case, the cat died: Phillips, 1993), such details are unlikely to find their way into research reports.

Gross (1990) has emphasized the construction of scientific papers as fictionalized, idealized accounts in which "style . . is not a window on reality, but the vehicle of an ideology that systematically misdescribes experimental and observational events" (p. 84). These records bear little relation to what is spoken in the laboratory. Conversation, unlike written records, typically makes reference to "what can go wrong."

Many journals do, of course, require that authors explain how many animals they used, of what species, and so on, in some detail. Some now insist that submitted papers adhere to published guidelines regarding how animals are kept or used. Thus journals specializing in the scientific study of laboratory animals, say, or in their behavior, are more likely to give details of animal housing and maintenance; they are also more likely to note the need to follow guidelines (see Smith, Birke and Sadler, in press). Such journals focus on the animal itself, rather than its tissues or organ function, so greater attention to details of animals and their husbandry is perhaps unsurprising. Nevertheless, these descriptions do help to make the animal more present. Few papers in other journals, however, give full details of the housing and husbandry conditions of the animals (even though these might be significant variables in experimental design, influencing the animal's physiology and thus the experimental outcome). Moreover, producing papers according to strict formats - introduction, methods, results, conclusion - may serve further to reduce the significance of the animals.

Bazerman (1988) points out, for instance, how increasingly prescriptive requirements for publication in the American Psychological Association journals may lessen "the likelihood that researchers will consciously consider the exact significance of ...information, whether it and other possible information should be included, and exactly how this information should be placed in the structure of the whole article" (p. 260). In other areas of biological research, where the animal is even less of a "subject," the impact of such unconscious choice of linguistic structure may be considerable. Thus, the centrality of "the gut" rather than the guinea pig is reinforced by the way the paper is written: it is the readout from machines that will dominate the methods section, not the animal who lost its life.

"The animal that lost its life" connotes a furry creature, a guinea pig, similar to the kinds of animals some people keep as pets. Not surprisingly, this kind of image is not likely to be invoked in scientific reports. It is the "common sense" image of an animal, to which most of us might refer when we talk about animals conversationally. It can be found in the practice of science - technicians may use this sense of "animal" in their work, for example - even if it rarely finds its way into journal articles (Arluke, 1988).

Scientific writing has become increasingly codified, particularly since the 19th century (Bazerman, 1988). Although it is not explicitly stated, it seems reasonable to suppose that this formalization - through which the presence of the real, live animal is obscured in the writing - has been in part a response to criticism of the use of animals. This may be partly unconscious, or authors may play down what they did to animals because of their own ambivalence, or their fears of reprisals. It may partly be the result of pressure from journal editors: Lederer (1992) has noted how, even in the first part of this century, when antivivisectionist activity was less obvious, journal editors rephrased submitted papers so as to reduce the emotional impact of what had been done to the animals.

It is not our purpose in conducting these analyses to imply that scientists are somehow bent on obscuring the truth about animal experiments. What sociological studies suggest is that scientists are often well aware of at least some of the dilemmas involved (Arluke, 1988; Michael and Birke, 1994). They may respond to these dilemmas by drawing the line (not using certain species, for example, or avoiding particular techniques), or by assuming that abuses of animals in laboratories always occur elsewhere (Birke and Michael, 1992).

The ways in which concepts of "animals" are used in Western culture are complex and multiple. Scientists, no less than other people, must deal with that complexity. They must, for example, treat pets differently from the "animal models" of the lab. Given these multiple meanings and nuances of "animals," it would be surprising indeed if scientists were not ambivalent.

The Public Face of Science?

The rhetoric of science, and the peculiar and particular construction of the scientific paper, have evolved over centuries. Certainly, in the course of that history, styles of writing have developed that do indeed reduce the significance of the animal (or of the scientist or reader). What is important, then, is to understand how that has happened, and what it signifies. Some of it may, no doubt, be a response to antivivisectionist feeling, as Lederer (1992) has pointed out. Obscuring the animal may, however, be partly incidental to other changes in the practice and narratives of science. Science claims to study nature; yet what enters the laboratory (including the animals) is highly artifactual. Knorr-Cetina (1983) notes that 'raw' materials which enter the laboratory are carefully selected and 'prepared' before they are subject to 'scientific' tests...To the observer from the outside world, the laboratory displays itself as a site of action from which 'nature' is as much as possible excluded rather than included" (p. 119).

What scientists are writing about, then, is a particular construction of reality, founded upon the laboratory and its apparatuses. The animals - their bodily integrity, their sentience, their possible suffering - are in many ways incidental to that construction. Scientists are increasingly invoking counter strategies against the animal rights lobby, by pointing to the medical benefits accruing from animal-based research. They do so, amid rhetoric that describes "the public" as ignorant of the truth.

We doubt that "the public" is so "ignorant" of the claims for medical benefits. What people may be responding to is what can be called the body language of science. Scientific papers create a strange narrative that obscures both animals and people, and that focuses on abstractions and apparatus. Public antagonism to science, insofar as it exists, may well be a response to such narratives. There is no evidence (at least from public opinion polls) that people are "anti-science," but trust in science and scientists may be less than it was some decades ago (La FoIlette, 1990). The public may be now skeptical of the authority of science, or claims that it can solve all our ills. They may, too, be skeptical of assumptions that animals used in laboratories do not suffer unduly. A language that obscures the realities of life in the lab is likely only to fuel that skepticism.

Notes

1. Correspondence should be sent to Lynda Birke, Centre for the Study of Women and Gender, University of Warwick, Coventry, CV4 7AL, United Kingdom.

2. Latour (1987) notes that scientific papers possibly constitute the only form of literature that is designed deliberately to be obscure and intelligible only to a tiny minority. Most non-scientific writing is, by contrast, intended to convey a wider message, perhaps to convince others of the argument. In his analysis of scientific writing, he stresses how arguments in a written paper are constructed so as to deflect potential criticism of the science.

3. This draws on discourse analysis; for example, see Potter and Wetherell, 1987.

4. Specifically, the following numbers of papers were sampled from each journal: Life Science (1990) vol. 46, parts 3,10,17,22 (26 papers); J. Neuroscience (1990) vol 10, no. 5 (20 papers); J. Immunology (1990) vol. 145, no. 8 (20 papers); J. Comparative Psychology (1991) vol. 105, nos. 1 and 2 (26 papers); J Experimental Biology (1990), vol 148 (20 papers); J. Surgical Research (10 papers); Cardiovascular Surgery (10 papers); Laboratory Animals (20 papers).

5. We appreciate the irony of doing this, given our conclusions about omitting details. However, we feel strongly that none of these authors are behaving atypically in the way they write, and it would be invidious to single any one writer out for particular blame.

6. As several papers were jointly authored, we cannot give precise numbers. Approximately 50 per cent of these came from North America; approximately ten percent from Japan; and approximately 25 percent from Europe, and others from Hong Kong, Israel, and Australia. There are, however, large differences within each area; controls in Britain are particularly strict, for example.

7. The possible effects of stress on the variable being measured and how often it is ignored in research reports need further investigation. One paper we examined (ironically, investigating effects of stress) seemed to involve subjecting experimental animals to stress through restraint, then killing them (LSM4). Are controls, in the same room, likely to be unaffected by these procedures?

8. One rather graphic example of this was a paper referring to killing rats by microwave irradiation, giving the precise model of domestic microwave, the number of kilowatts, and the duration of exposure in seconds (paper JL3).

9. There is an interesting parallel in that animals must be killed to become meat; animals that just die are not culturally acceptable. See Fiddes, 1991.

10. There may be many more "mistakes" that are obscured simply by the failure of many reports to give the number of animals used; see Smith et al., in press.

11. There is, by contrast, a striking lack of reference in scientific writing to wastage of laboratory animals that do not make it to become data. Large numbers of animals are culled as part of routine maintenance of animal holding facilities, because they are surplus, too old, etc. But these animals never would become "analytic" in Lynch's sense.

12. Knorr-Cetina (1983) makes the point that scientists work within a highly constructed and artificial environment, from which nature itself is systematically excluded. Even the laboratory animals are a construct.

13. In one case we have encountered, the laboratory (which used tissue samples for cancer research) was some miles from the animal house. Communication of protocols was sent to technicians by fax; animals' tissues were sent back by courier.

14. Wieder (1980) notes, however, that for scientists working closely with some species, such as chimpanzees, there is a chasm between scientists' attitudes and behavior toward the animals they work with, and the more constrained way in which scientific reports are subsequently constructed.

References

Arluke, A. (1988). Sacrificial symbolism in animal experimentation: object or pet? Anthrozoos , 2, 97-116.

Arluke, A. (1992). Trapped in a guilt cage. New Scientist , 134, 1815, 33-35.

Bazerman, C. (1988). Shaping written knowledge: The genre and activity of the experimental article in science . Madison: University of Wisconsin Press.

Birke, L. and Michael, M. (1992). Views from behind the barricade. New Scientist , April 4, 29-32.

B.A.A.S (British Association for the Advancement of Science), (1993). Animals and the advancement of Science . London: British Association.

Drury, M., Wade, O., Kuenssberg, E., Howie, J., Horder, J., Lawson, J., Carne, S., and Hull, R. (1990). Benefits of medical research and the doctor's responsibility. British Medical Journal . 300, 538.

Fiddes, N. (1991). Meat: A natural symbol . London: Routledge.

Gross, A. G. (1990). The rhetoric of science . Cambridge, MA: Harvard University Press.

Jasper, J. and Nelkin, D. (1992). The animal rights crusade . New York: The Free Press.

Knorr-Cetina, K. D. (1981). The manufacture of knowledge: An essay on the constructivist and contextual nature of science . Oxford: Pergamon.

Knorr-Cetina, K. D. (1983). The ethnographic study of scientific work: towards a constructivist interpretation of science. In: K. D. Knorr-Cetina and M. Mulkay (Eds.), Science observed: Perspectives on the social studies of science . London: Sage.

LaFollette, M. (1990). Making science our own: Public images of science 1910-1955 . Chicago: University of Chicago Press.

Latour, B. (1987). Science in action . Milton Keynes, UK: Open University Press.

Lederer, S. (1992). Political animals: the shaping of biomedical research literature in twentieth century America , Isis, 83, 61-79.

Lynch, M. (1985). Art and artifact in laboratory science: A study of shop work and shop talk in a research laboratory . London: Routledge.

Lynch, M. (1988). Sacrifice and the transformation of the animal body into a scientific object: laboratory culture and ritual practice in the neurosciences, Social Studies of Science , 18, 265-89.

Lynch, M., Livingston, E., and Garfinkel, H. (1983). Temporal order in laboratory work. In: K. D. Knorr-Cetina and M. Mulkay (Eds.), Science observed: Perspectives on the social studies of science . London: Sage.

Matfield, M. (1989). The animal rights threat to biomedical research. Biologist , 36, 238-240.

Michael, M. and Birke, L. (1994). Enrolling the core set: the case of the animal experimentation controversy. Social Studies of Science , 24, 81-95.

Nelkin, D. and Jasper, J. (1992). The animal rights controversy. In: D. Nelkin (Ed.), Controversy: Politics of technical decisions . London: Sage.

Phillips, M. T. (1993). Savages, drunks and lab animals: The researcher's perception of pain. Society and Animals , 1, 61-81.

Phillips, M.T. (1994). Proper names and the social construction of biography: the negative case of laboratory animals. Qualitative Sociology , 17, 119-142.

Potter J. and Wetherell, M. (1987). Discourse and Social Psychology , London: Sage.

Smith, J., Birke, L., and Sadler, D. (in press). Reporting animal use in scientific papers. Laboratory Animals .

Wieder, D. L. (1980). Behavioristic operationalism and the life-world: chimpanzees and the chimpanzee researchers in face-to-face interaction. Sociological Inquiry , 50, 75-103.

Wojtas, O. (1993). Researchers pilloried by animal rights group. The Times Higher Educational Supplemen t, Oct. 15, p. 2.

For abstracts of all issues, including the most current, click Article Abstracts

To order Society & Animals Journal, go to our secure online ordering page

You can Search the online issues of Society & Animals, as well as the entire Society & Animals Forum (formerly PSYETA) website,
for topics and keywords of your interest: