animal husbandry research paper

• Search Menu
• Sign in through your institution
• Advance articles
• Themed Issues
• Author Guidelines
• Open Access
• About ILAR Journal
• About the Institute for Laboratory Animal Research
• Editorial Board
• Advertising and Corporate Services
• Self-Archiving Policy
• Dispatch Dates
• Journals on Oxford Academic
• Books on Oxford Academic

Article Contents

Introduction, how husbandry variables can impact study outcome, husbandry-related cycles, in-house transport, physicochemical environment, material-related issues, social aspects, conclusions.

• < Previous

Animal Husbandry and Experimental Design

Timo Nevalainen, DVM, MS, PhD is an Emeritus Professor in the Laboratory Animal Center of the University of Eastern Finland, Kuopio Campus, Finland.

• Article contents
• Figures & tables
• Supplementary Data

Timo Nevalainen, Animal Husbandry and Experimental Design, ILAR Journal , Volume 55, Issue 3, 2014, Pages 392–398, https://doi.org/10.1093/ilar/ilu035

• Permissions Icon Permissions

If the scientist needs to contact the animal facility after any study to inquire about husbandry details, this represents a lost opportunity, which can ultimately interfere with the study results and their interpretation. There is a clear tendency for authors to describe methodological procedures down to the smallest detail, but at the same time to provide minimal information on animals and their husbandry. Controlling all major variables as far as possible is the key issue when establishing an experimental design. The other common mechanism affecting study results is a change in the variation. Factors causing bias or variation changes are also detectable within husbandry. Our lives and the lives of animals are governed by cycles: the seasons, the reproductive cycle, the weekend-working days, the cage change/room sanitation cycle, and the diurnal rhythm. Some of these may be attributable to routine husbandry, and the rest are cycles, which may be affected by husbandry procedures. Other issues to be considered are consequences of in-house transport, restrictions caused by caging, randomization of cage location, the physical environment inside the cage, the acoustic environment audible to animals, olfactory environment, materials in the cage, cage complexity, feeding regimens, kinship, and humans. Laboratory animal husbandry issues are an integral but underappreciated part of investigators' experimental design, which if ignored can cause major interference with the results. All researchers should familiarize themselves with the current routine animal care of the facility serving them, including their capabilities for the monitoring of biological and physicochemical environment.

If the scientist needs to contact the animal facility after the study is completed and inquire about husbandry details to be included in the manuscript, it is obvious that there has not been proper emphasis placed on animal husbandry-related issues in the experimental design. Unless agreed otherwise, all facilities follow their own animal care routines, which may not be suitable or include the best choices for individual studies. It may be possible to obtain information on this routine care afterwards, but study-specific details, e.g., when the cages were changed, the location of cages in the rack, and any aberrations in environment may no longer be available. This is an opportunity lost and can ultimately interfere with the study results and their interpretation. Contacts with the animal facility personnel to include and plan husbandry-related aspects should be established well ahead of the study.

An original article communicates results to the scientific community; therefore, it is absolutely necessary that the article contains a complete description of the study under comparable conditions, preferably following ARRIVE guidelines ( Kilkenny et al. 2010 ) to help fellow scientists design future investigations. An article not revealing all the essential pieces of information becomes impossible to repeat with the same results, yet repeatability is one of the key qualities that all scientists seek. There is a clear tendency for researchers to describe procedures, chemicals, reagents, and statistics down to the smallest detail, yet at the same time provide minimal information on animals and their husbandry ( Alfaro 2005 ; Everitt and Foster 2004 ; Gerdin et al. 2012 ).

The most common animal study type deals with interference with the natural order of events and carefully observing the consequences. The importance stems from the quest for inference, i.e., what is produced, contributed to, or caused without ambiguity. In the end, the authors must be able to rule out all possible alternate causes for the results they have obtained. Husbandry issues should be included as essential parts of this process.

There are several recent guidelines available to describe animals and their husbandry ( Alfaro 2005 ; Brattelid and Smith 2000 ; Hooijmans et al. 2010 ; Kilkenny et al. 2010 ; National Research Council (US) Institute for Laboratory Animal Research 2011 ). Some are based on surveys of the existing situation, while others provide reasons or lists of items to be described. A closer look at the surveys assessing the quality of design reveals that there are common deficiencies in design, such as lack of description of randomization and bias ( Kilkenny et al. 2009 ; McCance 1995 ; Perel et al. 2007 ). Although the primary focus of randomization is on random allocation of the animals to the groups, it is feasible to assume that the same deficiencies are true for husbandry-related experimental design items.

Controlling all major variables as far as possible is the key issue when establishing an experimental design, i.e., the only difference between the study and control groups is the procedure and nothing else. Any remaining variation can be controlled by randomizing the treatments to the animals. This applies to all aspects of the design, e.g., procedures under study and everything influencing the lives of the animals and to what they are exposed.

Bias is any systematic difference between the groups in addition to the procedure in the study design. No one would purposely incorporate bias into their design, yet it may be introduced because several aspects have not been properly considered or understood or the possibility has been totally ignored. A simple example of a bias is a study examining the effects of ethanol on animals through offering 10% ethanol in the drinking water. In reality, this design evaluates the combined effects of ethanol, eating less, and drinking less; the latter because animals tend to drink less fluid and ethanol provides a considerable amount of calories and eating is calorie guided.

The other common mechanism affecting study results is a change in variation; sources of variation in animal study in relation to husbandry and beyond have been analyzed by Howard (2002) . An increase in the background noise makes statistical significance more difficult to attain or it may require more animals. Consequently, a decrease in variation will have the opposite effect. Husbandry-related variation effects are traceable to disturbances in animals' lives, such as disease, transportation or cage change, or simply exposure to a fluctuating or changing physicochemical environment. In principle, a change in variation can be caused by any husbandry variable, though some are more likely candidates than others. While establishing an experimental design for an experiment, it is important to invest adequate time for the identification of such factors and building up effective strategies to cope with them. The aim of this paper is to elucidate selected animal housing and care-related items with examples and discuss their potential to interfere with the experimental design.

Animal life follows many different cycles, including the seasonal cycle, reproductive cycle, weekend-working days cycle, cage change /room sanitation cycle, and diurnal cycle. Some of these are actually established by the routine husbandry, and the rest are cycles that are sensitive to husbandry procedures.

Seasonal Cycle

Animals are capable of sensing the season even when housed in windowless rooms with programmed photoperiods and with a tightly regulated physicochemical environment. For example, there are reports of seasonal variation of immunoreactivity in models of septic shock and immunosuppression induced by chronic stress in 2 strains of mice. Kiank et al. (2007) showed that mice living with a 12:12-hour photoperiod had an enhanced risk to die of peritonitis in the summer or autumn compared with the other seasons. This is a situation where a control group may offer a solution to this problem. Theoretically, there is no other way to counteract seasonal variation other than to always carry out the study during the same season, yet even seasons are not alike and this approach as such may not be practical.

Reproductive Cycle

In rodents, the female reproductive cycle is sensitive to the pheromones present in male urine. Classical examples of this are the Whitten effect; male pheromones synchronizing estrus in females, the Lee–Boot effect; suppression or prolongation of estrous cycles of mature females when they are housed in groups and isolated from male mice, and the Bruce effect, which refers to the tendency for female rodents to terminate their pregnancies following exposure to the scent of an unfamiliar male.

Controlling spread of odors within a room, if not to the whole facility, is problematic; this is possible only if ventilation of the test enclosure is separated from the rest of the animals. In other situations, all persons visiting animal rooms should wash their hands with unscented soap and change protective clothing between the rooms ( Wersinger and Martin 2009 ).

Weekend-Working Days Cycle

During the weekend, fewer people are present in the animal facility; there is less human activity and hence a reduced acoustic environment audible to animals in the animal rooms. This change does not go unnoticed by the animals. Whether this represents a better time window for sampling and recording is a study-specific question.

For instance, this cycle causes change in blood pressure, which is a parameter commonly recorded in research. By using telemetric methods, which allow continuous recording of freely moving animals, it has been shown that spontaneous locomotor activity and blood pressure in the working days in rats are higher than during weekends. Apart from the increased locomotor activity (33%, p < 0.001), the daytime blood pressure differences were small (3.7–4.2 mmHg, p < 0.05) yet large enough to complicate the interpretation of study results ( Schreuder et al. 2007 ).

In most animal facilities, sound levels are low during the weekends, suggesting that human activities are a very important source of sound ( Milligan et al. 1993 ). Every facility has its own weekend environment, and all weekends are not exact replicas of the previous ones. For the purposes of experimental design, this possible confounding effect has to be discussed with the facility personnel to decide whether to conduct samplings all through 7 days a week, or to use either working days or weekends, keeping in mind that there may be other disturbing factors that one may prefer to avoid.

Cage Change and Room Sanitation Cycle

Cage changes and room sanitation represent major exposures for laboratory animals. The key question here is how long the animals are disturbed and what is the magnitude and nature of the disturbance, starting from preparations and in particular after the cage change, when the animals are not suitable for sampling or recording. Another solution is to find and use appropriate husbandry procedures to replicate the previous physiological state.

Rats housed in their home cages display increased locomotive activity, bedding manipulation, and defecation in both the mornings and afternoons of cage change days. This behavioral effect is greater than that of time of day and lasts for several hours after the cage change ( Saibaba et al. 1996 ). Likewise in rats, the cage change results in elevations in heart rate and mean arterial pressure levels lasting for up to 5 hours after the procedure. The reactions observed after cage change were significantly greater than those observed after simple handling, handling being part of the cage change process ( Meller et al. 2011 ).

In mice, postcleaning activity also includes aggression, which can cause serious injuries ( Van Loo et al. 2004 ). In C57BL6/NTac mice, the cage change increased systolic blood pressure, heart rate, and locomotor activity; in females, these changes lasted about 100 minutes, while in males the effect duration was about 20 to 25% shorter, irrespective of change frequency (weekly vs . fortnightly) ( Gerdin et al. 2012 ). It seems safe to recommend that no manipulation, dosing, sampling, or recording of animals should be done during the cage change day or the following day. For sensitive behavioral experiments, an even longer timescale may be necessary.

In studies on social behavior, the routine frequency of cage cleaning can be deleterious by disrupting its environment, including scent marks (pheromones), nests, and hoarded food. Cage cleaning and sanitation processes can also alter a rodent's normal production of pheromones and thus may affect behavior. To prevent pheromonal interference and stress-induced pheromonal release in their research subjects, experimenters should assess their current laboratory protocols regarding cage cleaning processes, housing designs, and behavioral assays. The lowest possible frequency would avoid unnecessary stress to animals ( Wersinger and Martin 2009 ).

There is evidence that transfer of specific olfactory cues during cage cleaning and the provision of nesting material can decrease aggression and stress in group-housed male mice ( Van Loo et al. 2004 ). Removing only the wet soiled bedding from a dirty cage, returning the nest or a portion of soiled bedding, some familiar complexity item, or food hopper with diet to the clean cage might be better options for the animals and the study design ( Bind et al. 2013 ; Meller et al. 2011 ; Wersinger and Martin 2009 ).

Diurnal Cycle

Diurnal rhythm influences many aspects of animal life, and changes seen are more than minute “physiologic” variations around a 24-hour mean. The rhythm has an endogenous nature but is affected by husbandry-related factors such as food, light, stress, and even tinted cage walls, and it also influences in vivo results such as melatonin, total fatty acids, glucose, lactic acid, corticosterone, insulin, and leptin in rats ( Wren et al. 2014 ). In animal care and housing, one should avoid anything that could disrupt the diurnal rhythm, such as malfunctioning or changing photoperiods or practicing common types of restricted feeding ( Chacon et al. 2005 ).

Ideally, samples and recordings should be obtained from freely moving animals, so that they are not aware of the procedure. In most cases, this simply is not possible; hence, samples are ‘contaminated’ with human presence in the room and the handling and sampling method. Moving the animals to another space gives them more time to react compared with the same procedure done in the animal room. For example, when cages of group-housed mice were transported to another room on a wheeled trolley and stored on a mobile ventilated rack during testing, this had no effect on blood glucose, but body temperature increased significantly when compared with nontransported controls. It required one hour of acclimatization for temperature to be restored ( Gerdin et al. 2012 ). Because there are more and less sensitive parameters to be assessed in each study, the study group has to evaluate the optimal location to conduct the procedures.

Current caging systems have been designed primarily with disease control in mind, i.e., materials should be easy to sanitize and sterilize. Complexity items came later; some of them are intended only for single use, some can be used to transfer specific olfactory cues during cage cleaning. Any disease, whether overt or subclinical, can interfere with the investigation in an unpredictable way. Health monitoring is practiced to detect the presence of an animal pathogen as quickly as possible. Depending on the caging system, this may require measures ranging from decontamination of the entire facility to a few cages.

Protection of the animals against pathogenic organisms is crucial to animals, animal center personnel, and investigators. The facility has rules to be followed and obeyed, e.g., restricted access of both animals and humans into the facility, acceptable working practices when inside, and requirements for cleaning of research equipment, instruments, and biological materials, and these must be adhered to while establishing the experimental design.

The investigators should allocate the animals at random into the study groups, and then cages should be placed into the cage rack also at random. An alternative is to incorporate blocking, which would not involve an increase in the number of animals but might provide additional information on analysis. Cages in different locations experience temperature, humidity, and lighting gradients based on cage level, distance to ventilation inlets and exhausts, lighting, and even sound sources. If cages are assigned to cage racks in a systematic manner, these external factors can introduce bias into the statistical analyses. These kinds of effects have been reported, but it is likely that most of these biased results have gone unnoticed ( Herzberg and Lagakos 1992 ).

Animal technicians should be told that cage locations are being randomized and be given the master key to the locations. Unless this is done, the location may accidentally change. There are claims that cage locations should be continuously changed, e.g., at each cage change and that would achieve the same goal as permanent random order. Unfortunately, this is not the case; instead, continuous place change leads to increased variation in parameters sensitive to gradients in the room.

Temperature Humidity and Illumination

All holding and caging systems do not provide similar environments to the animals. In some cages, the physical environment is close to the one in the room, whereas in others the changes can be considerable. As an example, a study by Memarzadeh et al. (2004) compared environmental conditions inside mice cages with 4 different mechanical ventilation designs and a static isolator cage. The static isolator cages were found to have lower air velocity, higher relative humidity, higher NH 3 and CO 2 levels, lower body weight gain, and lower water consumption compared with the mechanically ventilated cages ( Memarzadeh et al. 2004 ). In mice, temperature and humidity variation can affect the age of puberty, i.e., low temperatures and extremely low humidity levels have been shown to delay sexual maturation ( Drickamer 1990 ).

Illumination has been shown to affect the estrous cycle in both in albino and normally pigmented mice. When 2 light intensities (15–20 and 220–290 lux) were used, the estrous cycle of both types of mice was shorter, and the proportion of albino mice from which embryos were recovered was significantly smaller than the proportion from black mice at the lower intensity ( Donnelly and Saibaba 1993 ).

Acoustic Environment

Noise sources are facility specific and include environmental control systems, maintenance and husbandry procedures, cleaning equipment, and other equipment used near to the animals. The sounds cover a wide frequency range, including the ultrasonic ( >20 kHz) frequency that animals but not humans hear. It seems likely that the levels reported can have a negative effect on animal physiology or behavior ( Sales et al. 1999 ). It is rather rare for the facilities to monitor the acoustic environment, especially ultrasounds beyond the range of human hearing. Although background sound levels in undisturbed situations are generally low, marked increases in sound levels often occur during the working days. It is clear that the acoustic environment of laboratory animals is an uncontrolled variable with the potential to interfere with behavioral and physiological experiments ( Milligan et al. 1993 ).

Investigators cannot avoid noise created by HVAC machinery in the facility, but they can try to make best of it. A good starting point is to find out whether the facility has assessed the sounds audible to animals all the time and those created by machinery used intermittently, such as cage and rack washers and autoclaves. Avoiding cage manipulations such as cage change, addition of diet to the cage lid and topping off water bottles ( Voipio et al. 2006 ), all sanitation processes, the presence of other experimenters in the same space, and conducting tests during the weekends are all things to be considered. If there are building construction activities causing noise or vibrations going on nearby, it may be best to avoid any studies at all.

Laboratory animals have a much better developed sense of smell than humans. Therefore, it is no surprise that the olfactory environment is important and any disruption to it carries wide range of consequences to the animals and hence to the study results. For example, changing the cage bedding and/or nesting material at various prescribed intervals results in delayed puberty as compared with nondisrupted control mice ( Drickamer 1990 ). Open-top cages allow the odors to spread from one room to the next unless animals are housed in special conditions, e.g., in an IVC-system or isolators with separate inlet and exhaust air ducts.

Cage Material

Traditionally, the polymers used as cage materials have been considered as being inert. The most common clear material, polycarbonate, has been shown to leach Bisphenol A, a monomer with estrogenic activity. Howdeshell et al. (2003) have shown that this compound becomes a problem once polycarbonate equipment is exposed to high temperatures and alkaline conditions, common procedures in sterilization and washing, and the amount of leaching increases as a function of use. Bisphenol A exposure as a result of being housed in used polycarbonate cages produced a 16% increase in uterine weight in prepubertal female mice relative to females housed in used polypropylene cages; however, it has to be noted that this difference was not statistically significant ( Howdeshell et al. 2003 ). This should be taken as a warning sign; the animals in old cages with bottles or even complexity items made of polycarbonate may be exposed to varying amounts of Bisphenol A, which may interfere with any estrogen-sensitive parameters. The only effective solution is to avoid using polycarbonate cages; other polymer materials are less prone to leaching of this chemical.

Bedding is the material with which the animals have continuous contact. The original observation that softwood bedding contains α- and β-pinene, which are compounds with liver microsomal enzyme induction properties, is quite old but so far it has not been widely implemented ( Vesell 1967 ). This is not the only effect of bedding; several other interfering properties have subsequently been discovered, e.g., measurable amounts of endotoxin and (1– >3)-beta- d -glucan are present in different bedding materials; after 5 weeks' exposure, this evoked moderate inflammatory reactions in the lung of the animals ( Ewaldsson et al. 2002 ).

To exclude bedding interference, one can assay residues in a batch, keep to one batch whenever possible, and, if using softwood bedding, then it is advisable to standardize heat treatments both at the manufacturer's plant and the facility, because heat treatment can decrease the levels of the compounds responsible for enzyme induction ( Nevalainen and Vartiainen 1996 ).

Complexity Items

Adding items and materials into the cage increases the complexity of the intimate environment of the animals. Complexity becomes environmental enrichment once it has been verified to confer a welfare outcome on the animals. However, despite many studies on complexity, there are still major gaps in our understanding, e.g., because of variety of item combinations, lack of applicability beyond one strain or stock of animals, and inappropriate controls.

For the scientist, the most important aspect of complexity is consistency; the added items must be present all of the time and they have to be the same for all animals. In group-housed animals, social hierarchies may influence the use of complexity items. Both the facility and the scientists need to understand that complexity is not an inert object in the cage but a potential variable in the study. Changing the intimate environment of animals has effects on brain structure, physiology, and behavior as well as an influence on which genes are expressed in various organs ( Benefiel et al. 2005 ). It has to be understood that minor complexity changes meant to improve animal welfare may alter the physiology and development of the animals in an unpredictable way. Moreover, there is also the possibility that the complexity preferred by the animals may not enhance laboratory animal welfare and may even interfere with the study ( Benefiel et al. 2005 ; Wersinger and Martin 2009 ). It is important to consider the potential effects of any complexity on welfare of the animal strain in use and, even more importantly, more studies need to be done to obtain more accurate data.

Variability between diet brands is to be expected, but there may be considerable variation between batches of the same diet brand. Consequently, the study parameters will also vary. Keeping to one batch in an experiment is advisable, but in subsequent studies, this may not be possible, and then analysis of at least the critical components affecting the study may be helpful. Because the food deteriorates during storage, the analysis may also be necessary if the duration of the experiment is prolonged.

It is widely known that eating too much is unhealthy, but this is what we routinely offer to our research animals. The drawbacks of ad libitum feeding include an increased variation in food intake and consequently an increased variation in body weight and other variables and increased mortality and shorter lifespan. It is not surprising that ad libitum feeding has been called the least controlled variable in rodent bioassays ( Keenan et al. 1998 ).

Dietary restriction is the solution to these problems, but this can lead to other problems. Dietary restriction in rodents requires single housing in order to provide them with a meal once a day, often during the daytime; this disrupts physiological and behavioral diurnal rhythms and would exert a confounding effect on the investigation ( Chacon et al. 2005 ; Damiola et al. 2000 ; Forestell et al. 2001 ; Nelson 1988 ). For example, in studies with dietary restriction, it is unclear whether the differences observed are due to caloric intake per se or altered diurnal rhythms.

The diet board offers the possibility of combining dietary restriction with group-housing and normal eating rhythms ( Kasanen et al. 2009a , 2009b ). From the point of view of scientific quality, the combination of undisturbed diurnal rhythm and restricted caloric intake in rodents would be a valuable achievement. In rats fed with the diet board, serum ghrelin, blood glucose, and fecal corticosterone and immunoglobulin A have been shown to follow a diurnal rhythm ( Kasanen et al. 2010 ).

Scientists are well aware that with large laboratory animals, family relationship such as sisters and brothers must be evenly distributed to all groups in the study. The same allocation principle is not practiced with small laboratory animals, such as rodents. Although rats and mice look alike and litter data is usually lost at weaning, when the sexes are separated and cages filled to predetermined cage occupancy, nonetheless this may not be the best practice.

Good laboratory practice-guidelines emphasize that the study design should take account of kinship in large animals, but not for rodents. Safety studies with small animals typically include hundreds of animals, whereas in studies with large animals, the numbers are comparable with the numbers of small animals used in basic biomedical research. Rather than tacitly accepting the difference in scale of experimental design, it is worth noticing that it is often logically inconsistent. Indeed, this a question of animal numbers, not of species.

In outbred and other undefined animals, litter effects are large and ignoring them can make replication of the studies difficult, if not impossible. As an example, a recent literature review of the valproic acid model of autism showed that only 9% (3/34) of studies correctly determined that the experimental unit was the litter and therefore had made valid statistical inferences. In fact, litter effects accounted for up to 61% ( p < 0.001) of the variation in behavioral outcomes, a much larger percentage than the treatment effects ( Lazic and Essioux 2013 ).

The need to recognize kinship is clear with outbred animals, but it should not be ignored in inbred and defined animals. Prager et al. (2010) showed that in young outbred rats, maternal care and litter size exerted a profound effect on immune-related parameters. Although inbred animals are genetically identical, their embryonic development and maternal care may not be the same, and hence kinship is an issue to be considered with defined strains of laboratory animals.

Accounting for kinship in rodents becomes a reality only once litter codes are available. In the case of in-house breeding, this should not be too difficult. With major breeders, this must be possible but ordering has to be done earlier than usual. The expected higher cost of animals can be offset by a reduction in the numbers of animals needed and thus less labor. When kinship data are available, animals should be divided into the study groups as evenly as possible following a random block design as presented by Festing et al . (2002) .

Animal technicians, not only the scientists, should be considered as key individuals in the studies. The technicians see and observe the animals at least on a daily basis and consequently have a good understanding of what is normal in these animals and in particular what is not. They are also in charge of daily routines, and the way they deal with the animals can make a profound difference.

There is anecdotal evidence that if rats are moved to a new cage by lifting from the base of the tail, they become aggressive, and that this does not happen if they are lifted by holding the body. Lifting by the base of tail has traditionally been the preferred method for mice, and these animals have been considered aggressive and they do not habituate to handling. A recent article by Hurst and West (2010) has challenged this view using 2 strains and a stock of mice. If the mice were transferred to a new cage in cupped hands or in a transparent tunnel kept in the cage, then their anxiety decreased already after one such transfer. The animals were more tame when receiving administrations using the traditional means of immobilization ( Hurst and West 2010 ).

It was concluded that there is no need to consider handling-induced anxiety in mice, in essence representing an animal model of anxiety ( Hurst and West 2010 ). Investigators are urged to find out the advice provided by the facility on how to handle animals during routine care. Furthermore, they may apply this approach themselves and see the outcome.

Laboratory animal husbandry issues are an integral but underappreciated part of experimental design, which if ignored can cause major interference with the results. All study groups should familiarize themselves with the current routine animal care of the facility serving them and their monitoring capabilities on biological and physicochemical environment. With respect to husbandry issues, it is often easier to say what should not be done rather than what to do, but the final decisions are left to the group and depend largely on specific requirements of the study. While defining the experimental design for an experiment, it is important to invest adequate time in the identification of such issues and to devise effective strategies to deal with them.

Google Scholar

Author notes

Email alerts

Citing articles via.

• Recommend to your Library

Affiliations

• Online ISSN 1930-6180
• Print ISSN 1084-2020
• Copyright © 2024 Institute for Laboratory Animal Research
• About Oxford Academic
• Publish journals with us
• University press partners
• What we publish
• New features  
• Open access
• Institutional account management
• Rights and permissions
• Get help with access
• Accessibility
• Advertising
• Media enquiries
• Oxford University Press
• Oxford Languages
• University of Oxford

Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide

• Copyright © 2024 Oxford University Press
• Cookie settings
• Cookie policy
• Privacy policy
• Legal notice

This Feature Is Available To Subscribers Only

Sign In or Create an Account

This PDF is available to Subscribers Only

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.