Striving for humane deaths for laboratory mice: hypobaric hypoxia provides a potential alternative to carbon dioxide exposure.

Killing is often an unavoidable and necessary procedure for laboratory mice involved in scientific research, and providing a humane death is vital for public acceptance. Exposure to carbon dioxide (CO2) gas is the most widely used methodology despite well proven welfare concerns. Consequently, the continued use of CO2 and its globally permitted status in legislation and guidelines presents an ethical dilemma for users. We investigated whether killing with hypobaric hypoxia via gradual decompression was associated with better welfare outcomes for killing laboratory mice. We compared the spontaneous behaviour of mice exposed to CO2, decompression or sham conditions, and used analgesic or anxiolytic interventions to determine their relative welfare impact. Gradual decompression resulted in longer times to unconsciousness and death and the pharmacological interventions support the notion of a minimally negative animal experience, while providing further evidence for pain and anxiety associated with exposure to CO2. Decompression resulted in moderate ear haemorrhage, but our welfare assessment suggests this may happen when mice are unconscious. Hence, gradual decompression could be the basis of significant refinement for killing laboratory mice. Future work should corroborate behaviour with neurobiological markers of loss of consciousness to verify the conscious phase of concern for animal welfare.

Pathological consequences of low atmospheric pressure stunning in broiler chickens.

Low atmospheric pressure stunning (LAPS) is a novel approach to pre-slaughter stunning of chickens using progressive hypobaric hypoxia by the application of gradual decompression (280s cycle) according to a set of prescribed pressure curves. Low atmospheric pressure stunning produces a non-recovery state. Concerns have been raised relating to the possible pathological and welfare consequences of expansion of air in the body during LAPS. In a randomised trial, we compared the gross pathology of broilers exposed to LAPS with a control group euthanised by intravenous injection of pentobarbital sodium (60 mixed sex broilers per treatment). The birds were exposed to each treatment in triplets and all birds were subject to necropsy examination to detect and score (1 to 5, minimal to severe) haemorrhagic lesions or congestion for all major organs and cavities (e.g. air sacs, joints, ears and heart) as well as external assessment for product quality (e.g. wing tips). Behavioural data (latency to loss of posture and motionless) and chamber cycle data (temperature, humidity, pressure and oxygen availability) confirmed that LAPS had been applied in a manner representative of the commercial process. All of the organs observed were structurally intact for both treatment groups. No lesions were observed in the external ears, oral cavity, tracheal lumen, crop and air sacs of birds from either treatment group. There was no difference between treatments in the wingtips, nasal turbinates, thymus, biceps femoralis and colon. Haemorrhagic lesions were observed in the calvaria, brains, hearts and lungs of both treatment groups, but lesions in these areas were more severe in the LAPS treatment group. It was not possible to distinguish between pathological changes induced by decompression or recompression. In the barbiturate group, more severe haemorrhagic lesions were observed in the superficial pectoral muscles as well as greater congestion of the infraorbital sinuses, liver, spleens, duodenum, kidneys and gonads. These findings provide evidence that LAPS did not result in distension of the intestines and air sacs sufficient to cause changes, which were grossly visible on postmortem examination. There was also no evidence of barotrauma in the ears and sinuses. The pathological changes observed in the barbiturate treatment were as expected based on barbiturate toxicity. Low atmospheric pressure stunning appears to produce pathological changes by a variety of well-established mechanisms, and while these pathological data have limited value as welfare indicators, the results confirm that organ integrity was not compromised by the process.

Evaluation of the potential killing performance of novel percussive and cervical dislocation tools in chicken cadavers.

1. Four mechanical poultry killing devices; modified Armadillo (MARM), modified Rabbit Zinger (MZIN), modified pliers (MPLI) and a novel mechanical cervical dislocation (NMCD) gloved device, were assessed for their killing potential in the cadavers of euthanised birds. 2. A 4 × 4 × 4 factorial design (batch × device × bird type + age) was employed. Ten bird cadavers per bird type and age were tested with each of the 4 devices (N = 160 birds). All cadavers were examined post-mortem to establish the anatomical damage caused. 3. NMCD, MARM and MZIN demonstrated killing potential, as well as consistency in their anatomical effects. NMCD had the highest killing potential, with 100% of birds sustaining the required physical trauma to have caused rapid death. 4. The MPLI was inconsistent, and only performed optimally for 27.5% of birds. Severe crushing injury was seen in >50% of MPLI birds, suggesting that birds would die of asphyxia rather than cerebral ischaemia, a major welfare concern. As a result, the MPLI are not recommended as a humane on-farm killing device for chickens. 5. This experiment provides important data on the killing potential of untried novel percussive and mechanical cervical dislocation methods, informing future studies.

Effects of light on responses to low atmospheric pressure stunning in broilers.

Low atmospheric pressure stunning (LAPS) is a novel approach to poultry stunning involving the application of gradual decompression lasting 280 s according to a prescribed pressure curve. The aim of this study was to determine how behavioural, electroencephalogram (EEG) and electrocardiogram (ECG) responses to LAPS are influenced by illumination of the decompression chamber. A secondary aim was to examine responses to the decompression chamber without LAPS being applied, as such a "sham" control has been absent in previous studies. A two by two factorial design was employed, with LAPS/light, LAPS/dark, sham/light and sham/dark treatments (N = 20 per treatment). Broilers were exposed to each treatment in pairs, in each of which one bird was instrumented for recording EEG and ECG. Illumination was applied at 500 lx, and in sham treatments, birds were identically handled but remained undisturbed in the LAPS chamber without decompression for 280 s. Birds which underwent the sham treatment exhibited behaviours which were also observed in LAPS (e.g. sitting) while those exposed to LAPS exhibited hypoxia-related behaviours (e.g. ataxia, loss of posture). Behavioural latencies and durations were increased in the sham treatments, since the whole cycle time was available (in LAPS; birds were motionless by 186 s). Within the sham treatments, illumination increased active behaviour and darkness induced sleep, but slow-wave EEG was seen in both. The pattern of EEG response to LAPS (steep reduction in median frequency in the first 60 s and increased total power) was similar, irrespective of illumination, though birds in darkness had shorter latencies to loss of consciousness and isoelectric EEG. Cardiac responses to LAPS (pronounced bradycardia) closely matched those reported previously and were not affected by illumination. The effects of LAPS/sham treatment primarily reflected the presence/absence of hypoxia, while illumination affected activity/sleep levels in sham-treated birds and slowed time to unconsciousness in birds undergoing LAPS. Therefore, it is recommended that LAPS be conducted in darkness for poultry.

A preliminary investigation into the moral reasoning abilities of UK veterinarians.

Veterinary medicine is an ethically challenging profession, but the ethical reasoning abilities of practising veterinarians in the UK have never been formally assessed. This study investigated moral reasoning ability in 65 qualified veterinarians (38 practising and 27 academic) and 33 members of the public in the UK using the Defining Issues Test. Academic veterinarians had higher scores than members of the public but practising veterinarians did not. There was large variation in moral reasoning abilities among qualified veterinarians. Moral reasoning score in veterinarians did not improve with years of experience. These results show that despite having a professional degree moral reasoning skills of practising veterinarians may be insufficient to deal with the demands of their profession. This could have implications for animal welfare, client services and veterinarian wellbeing. The results highlight the need for more training in this area.

The effect of magnetic resonance imaging noise on cochlear function in dogs.

Noise produced by magnetic resonance imaging (MRI) scanners (which can peak at a sound pressure level of 131 dB) has been shown to cause noise-induced cochlear dysfunction in people. The aim of this study was to investigate whether noise produced during MRI had a deleterious effect on cochlear function in dogs, using distortion product otoacoustic emission (DPOAE) testing, which allows frequency specific, non-invasive assessment of cochlear function. DPOAE testing was performed before and after MRI in one or both ears under general anaesthesia at 14 frequency pairs (f2 frequency ranging from 0.84 kHz to 8.0 kHz). A control group comprised dogs undergoing anaesthesia of a similar duration for quiet procedures. Thirty-six dogs (66 ears) and 17 dogs (28 ears) were included in the MRI and control groups respectively. There was a reduction in DPOAE at all frequencies tested in the MRI group; a similar effect was not evident in the control group. This reduction in the MRI group was statistically significant in five of the 14 frequencies assessed (P < 0.05). These results demonstrate that exposure to MRI noise results in a significant reduction in frequency-specific cochlear function in dogs, although it is not known whether this is reversible or permanent. This suggests that all dogs undergoing MRI studies should be provided with ear protection as a routine precautionary measure.

Physiological and behavioral responses of poultry exposed to gas-filled high expansion foam.

Disease control measures require poultry to be killed on farms to minimize the risk of disease being transmitted to other poultry and, in some cases, to protect public health. We assessed the welfare implications for poultry of the use of high-expansion gas-filled foam as a potentially humane, emergency killing method. In laboratory trials, broiler chickens, adult laying hens, ducks, and turkeys were exposed to air-, N2-, or CO2-filled high expansion foam (expansion ratio 300:1) under standardized conditions. Birds were equipped with sensors to measure cardiac and brain activity, and measurements of oxygen concentration in the foam were carried out. Initial behavioral responses to foam were not pronounced but included headshakes and brief bouts of wing flapping. Both N2- and CO2-filled foam rapidly induced ataxia/loss of posture and vigorous wing flapping in all species, characteristic of anoxic death. Immersion in air-filled, high expansion foam had little effect on physiology or behavior. Physiological responses to both N2- and CO2-filled foam were characterized by a pronounced bradyarrythymia and a series of consistent changes in the appearance of the electroencephalogram. These were used to determine an unequivocal time to loss of consciousness in relation to submersion. Mean time to loss of consciousness was 30 s in hens and 18 s in broilers exposed to N2-filled foam, and 16 s in broilers, 1 s in ducks, and 15 s in turkeys exposed to CO2-filled foam. Euthanasia achieved with anoxic foam was particularly rapid, which is explained by the very low oxygen concentrations (below 1%) inside the foam. Physiological observations and postmortem examination showed that the mode of action of high expansion, gas-filled foam is anoxia, not occlusion of the airway. These trials provide proof-of-principle that submersion in gas-filled, high expansion foam provides a rapid and highly effective method of euthanasia, which may have potential to provide humane emergency killing or routine depopulation.

Physiological responses to low atmospheric pressure stunning and the implications for welfare.

In low atmospheric pressure stunning (LAPS), poultry are rendered unconscious before slaughter by gradually reducing oxygen tension in the atmosphere to achieve a progressive anoxia. The effects of LAPS are not instantaneous, so there are legitimate welfare concerns around the experience of birds before loss of consciousness. Using self-contained telemetry logging units, high-quality continuous electroencephalogram (EEG) and electrocardiogram (EKG) recordings were obtained from 28 broiler chickens during exposure to LAPS in a commercial poultry processing plant. Application of LAPS was associated with changes in the EEG pattern in the form of increases in total power, decreases in mean frequency, and in particular, increases in slow-wave (delta) activity, indicating a gradual loss of consciousness. Increased delta wave activity was seen within 10 s of LAPS onset and consistently thereafter, peaking at 30 s into LAPS at which point the EEG signal shared characteristics with that of birds in a surgical plane of anesthesia. During LAPS, heart rate consistently decreased, with more pronounced bradycardia and arrhythmia observed after 30 s. No heart rate increases were observed in the period when the birds were potentially conscious. After an initial quiescent period, brief body movements (presumed to be ataxia/loss of posture) were seen on average at 39 s into the LAPS process. Later (after 120 s on average), artifacts related to clonic (wing flapping) and tonic (muscle spasms) convulsions were observed in the EKG recordings. Based on EEG analysis and body movement responses, a conservative estimate of time to loss of consciousness is approximately 40 s. The lack of behavioral responses indicating aversion or escape and absence of heart rate elevation in the conscious period strongly suggest that birds do not find LAPS induction distressing. Collectively, the results suggest that LAPS is a humane approach that has the potential to improve the welfare of poultry at slaughter by gradually inducing unconsciousness without distress, eliminating live shackling and ensuring every bird is adequately stunned before exansguination.

Effects of contact dermatitis on hepatic gene expression in broilers.

1. Severe foot and hock dermatitis in broiler chickens can reduce growth rate and increase susceptibility to bacterial infection, affecting both profitability and welfare. However, little is known about the underlying physiological changes associated with foot and hock lesions. 2. This study compared global hepatic gene expression in control birds and those with ammonia-induced foot and hock lesions using Agilent 44 K chicken oligonucleotide microarrays (8 birds per group). 3. In total, 417 genes were differentially expressed of which 174 could be mapped onto the genome. Genes associated with energy metabolism, thyroid hormone activity and cellular control were affected, while there was also evidence of an up-regulation of genes linked to a pro-inflammatory response. 4. It is conceivable that pain is the underlying cause for the observed changes in energy metabolism genes. 5. Changes in hepatic gene expression provide new information on how a chicken's physiological mechanisms alter to cope with foot and hock lesions. The findings support other data indicating that birds with increased severity of lesions are likely to be in pain and that growth will be compromised. Reduction of the incidence of dermatitis by improved husbandry should therefore benefit both welfare and commercial performance.

Survey of the frequency and perceived stressfulness of ethical dilemmas encountered in UK veterinary practice.

The scale of the ethical challenges faced by veterinary surgeons and their perceived stressful consequences were investigated via a short questionnaire, completed by 58 practising veterinary surgeons. Respondents were asked to report how frequently they faced ethical dilemmas, and to rate on a simple numerical scale (zero to 10) how stressful they found three common scenarios. Fifty seven per cent of respondents reported that they faced one to two dilemmas per week, while 34 per cent stated they typically faced three to five dilemmas per week. The three scenarios provided were all rated as highly stressful with 'client wishing to continue treatment despite poor animal welfare' rated as the most stressful (median 9). The female veterinary surgeons gave two of the scenarios significantly higher stress ratings than the male veterinary surgeons. Stress ratings were not influenced by number of years in practice (which ranged from one to more than 25 years). The results show that veterinary surgeons regularly face ethical dilemmas and that they find these stressful. This has implications for the wellbeing of veterinary surgeons and supports the case for increased provision of training and support, especially for those who entered the profession before undergraduate ethics teaching was widely available.

Physiological responses of laying hens during whole-house killing with carbon dioxide.

1. Poultry on farms are sometimes required to be killed in an emergency, such as during a disease epidemic, yet none of the available methods are ideal. Whole-house carbon dioxide (CO(2)) administration has practical advantages, but gives rise to welfare concerns. 2. The study measured the body temperature, respiration, cardiac and brain activity (EEG) responses of 10 adult hens placed in tiered cages in a deep pit house while the entire flock (28,000 end-of-lay hens) was killed with CO(2). Video and thermographic images were also recorded. Liquid CO(2) was injected into the building producing a gaseous concentration of 45% within 19 min. 3. Those hens nearest the gas delivery site showed delayed respiratory, cardiac and EEG responses compared with those at more distant locations. Although sub-zero temperatures were recorded in the immediate vicinity of some birds, body temperatures indicated that they did not die of hypothermia. 4. EEG characteristics strongly associated with unconsciousness were used to determine an unequivocal time to loss of consciousness; this ranged from 6·0 to 10·5 (average 7·8) min after onset of gas injection. Distinctive cardiac and respiratory responses were seen following gas exposure; in particular, birds responded to inhalation of CO(2) by deep breathing. 5. The primary welfare concern is the duration of unpleasant respiratory effects, such as deep breathing, while the birds were substantively conscious. However, the concentration of CO(2) to which the birds were exposed while conscious would not have stimulated nasal and oral nociceptors. Time to death varied between 12·0 and 22·1 min after gas delivery.

Remote monitoring of electroencephalogram, electrocardiogram, and behavior during controlled atmosphere stunning in broilers: implications for welfare.

This study examined the welfare implications of euthanizing broilers with 3 gas mixtures relevant to the commercial application of controlled atmosphere stunning (CAS). Birds were implanted/equipped with electrodes to measure brain activity (electroencephalogram, EEG) and heart rate. These signals were recorded using a purpose-built telemetry-logging system, small enough to be worn by each bird in a spandex backpack. The birds were euthanized in a scaled-down CAS apparatus consisting of a conveyor belt passing through 2 compartments. Three gas environments were applied (8 birds per treatment): 1) anoxia (N(2) with <2% residual O(2), in both compartments), 2) hypercapnic anoxia (N(2) with 30% CO(2) and <2% residual O(2), in both compartments), and 3) a 2-phase approach with a hypercapnic hyperoxygenated anesthetic phase (40% CO(2), 30% O(2), and 30% N(2), in the first compartment, 80 s) followed by a second euthanasia phase (80% CO(2) in air, in the second compartment). All 3 CAS approaches effectively achieved nonrecovery states, and time to loss of consciousness for each bird was determined by visual determination of isoelectric EEG and by calculation of the correlation dimension of the EEG. Hypercapnic anoxia resulted in rapid unconsciousness and death; both anoxic treatments were associated with early onset prolonged wing flapping and sustained tonic convulsions as displayed in the electrophysiological recordings. These responses were seen in the period when consciousness remained a possibility. Hypercapnic hyperoxygenation (the 2-phase approach) was associated with respiratory disruption, but this treatment eliminated initial clonic convulsions in the stunning process, and tonic convulsions were not seen. These results suggest that the presence of O(2) in the first stage of CAS is associated with an absence of potentially distressing behavioral responses. The respiratory discomfort associated with hypercapnic hyperoxygenation is an issue. We propose that this may be compensated by a more gradual induction to unconsciousness, which eliminates the impact of other potentially negative experiences.

Controlled atmosphere stunning of broiler chickens. I. Effects on behaviour, physiology and meat quality in a pilot scale system at a processing plant.

1. The effects of controlled atmosphere stunning on the behaviour, physiology and carcase and meat quality of broiler chickens were studied experimentally in a pilot scale plant. 2. Gas mixtures tested were: single phase anoxic mixture (90% Ar in air, <2% O(2)); single phase hypercapnic anoxic mixture (60% Ar, 30% CO(2) in air, <2% O(2)); and biphasic hypercapnic hyperoxygenation mixture (anaesthetic phase, 40% CO(2), 30% O(2), 30% N(2); euthanasia phase, 80% CO(2), 5% O(2), 15% N(2)). 3. Anoxic stunning resulted in the least respiratory disruption, mandibulation and motionlessness, but most head shaking, leg paddling and twitching. Loss of posture occurred soonest with hypercapnic anoxia with the earliest and most twitching and wing flapping in individuals and earliest leg paddling. Biphasic birds were most alert, exhibited most respiratory disruption and mandibulation, and had the latest loss of posture and fewest, but longest bouts of wing flapping and least leg paddling and twitching. 4. Significant and sudden bradycardia and arrhythmia were evident with all gas mixtures and were not related solely to anoxia or hypercapnia. Birds stunned by Ar anoxia showed a slightly more gradual decline from baseline rates, compared with hypercapnic mixtures. 5. Few differences were found between gas mixes in terms of carcase and meat quality. Initial bleeding rate was slowest in biphasic-stunned birds, but total blood loss was not affected. Acceleration of post-mortem metabolism in anoxic-stunned birds was not sufficient to allow de-boning within 5 h without the risk of tough meat. 6. On welfare grounds and taking into account other laboratory and field studies, a biphasic method (using consecutive phases of anaesthesia and euthanasia) of controlled atmosphere stunning of broilers is potentially more humane than anoxic or hypercapnic anoxic methods using argon or nitrogen.

Controlled atmosphere stunning of broiler chickens. II. Effects on behaviour, physiology and meat quality in a commercial processing plant.

1. The effects of controlled atmosphere stunning on behavioural and physiological responses, and carcase and meat quality of broiler chickens were studied experimentally in a full scale processing plant. 2. The gas mixtures tested were a single phase hypercapnic anoxic mixture of 60% Ar and 30% CO(2) in air with <2% O(2), and a biphasic hypercapnic hyperoxygenation mixture, comprising an anaesthetic phase, 40% CO(2), 30% O(2), 30% N(2), followed by an euthanasia phase, 80% CO(2), 5% O(2), 15% N(2). 3. Birds stunned with Ar + CO(2) were more often observed to flap their wings earlier, jump, paddle their legs, twitch and lie dorsally (rather than ventrally) than those stunned with CO(2) + O(2). These behaviours indicate a more agitated response with more severe convulsions during hypercapnic anoxia, thereby introducing greater potential for injury. 4. Heart rate during the first 100 s of gas stunning was similar for both gases, after which it remained constant at approximately 230 beats/min for CO(2) + O(2) birds whereas it declined gently for Ar + CO(2) birds. 5. In terms of carcase and meat quality, there appeared to be clear advantages to the processor in using CO(2) + O(2) rather than Ar + CO(2) to stun broiler chickens, for example, a much smaller number of fractured wings (1.6 vs. 6.8%) with fewer haemorrhages of the fillet. 6. This study supports the conclusions of both laboratory and pilot scale experiments that controlled atmosphere stunning of broiler chickens based upon a biphasic hypercapnic hyperoxygenation approach has advantages, in terms of welfare and carcase and meat quality, over a single phase hypercapnic anoxic approach employing 60% Ar and 30% CO(2) in air with <2% O(2).

Evaluation of the effects of infrared beak trimming in broiler breeder chicks.

Detailed behavioural observations were made of broiler breeder chicks after their beaks had been trimmed by an automated infrared treatment at one day of age or by the traditional hot-blade method at one day or seven days, or after they had been sham-trimmed or left untreated. Observations took place immediately after the treatments and at regular intervals until six weeks of age; beak length and bodyweight were also measured regularly. There were no significant effects on the behaviour of the chicks in the first hour after trimming or in the subsequent six weeks. Variability in beak length was low within the treatments and there was significant regrowth, but it was least in the birds that had been hot-blade trimmed at seven days. Both beak-trimming methods were associated with small but significant reductions in bodyweight, with the hot-blade-treated birds being more affected.

Mechanothermal nociceptors in the scaly skin of the chicken leg.

Electrophysiological recordings were made from single sensory mechanothermal nociceptive afferent fibres in dissected nerve filaments of the parafibular nerve innervating the scaly skin of the lower leg of the chicken. Two classes of mechanothermal nociceptors were identified consisting of 34 C fibres (conduction velocities 0.45-1.5 m/s, mean 1.08) and nine A-delta fibres (3-15 m/s, mean 6.34). The C fibre afferents had receptive fields which were circular or elliptical in shape and ranged in size from 1 mm in diameter to 4 x 3 mm. Thresholds to mechanical stimulation in the C fibre afferents ranged from 0.3 to 33 g (median 1.5 g) and thermal thresholds were in the range 39-61 degrees C (median 49.4 degrees C). Stimulus-response curves to thermal and/or mechanical stimulation were recorded from 28 C fibre afferents and subjected to a linear regression analysis to determine whether they were best fitted by a linear, log or power function. The results were variable and no single function provided the best fit for all the responses. Of the fibres tested with both stimulus modalities (n=17), only 12 fibres showed the same best fit for both stimuli; in the others the best fit regression lines differed between stimuli. The response of the A-delta fibres to mechanical and thermal stimulation was very similar to the C fibres but the small number of A-delta fibres precluded any detailed statistical analysis. Comparison of the physiological properties of the C fibres in the leg with those previously identified in the beak showed that those in the leg had significantly lower thermal thresholds, but higher mechanical thresholds. The possible functional significance of these differences is discussed. These findings are also discussed in a comparative context to identify similarities and differences between mechanothermal nociceptors in birds and other vertebrates, relating these to their possible evolutionary and functional significance.

Development of pecking damage in layer pullets in relation to dietary protein source.

1. In recent years, the UK egg industry has become increasingly dependent on plant protein sources, in particular soyabean meal, and it has been suggested that this trend (and/or the concomitant absence of animal protein in layer diets) might be causally related to increased feather pecking and cannibalism. 2. This study examined the development of pecking damage in relation to dietary protein source, by rearing 12 groups of 12 layer pullets to 24 weeks of age on diets based on 'animal' (fishmeal) or 'plant' (soyabean meal) protein. 3. Damaging pecking began at 6 weeks of age, in three groups (one plant and two animal). Injurious pecking began at 18 weeks of age, and affected four groups (two plant and two animal). 4. Greater numbers of vigorous pecks/pulls were observed in plant protein groups throughout the experiment, although they were significantly higher only between 13 to 16 weeks of age. Pecking damage scores did not differ between treatments. 5. Dietary protein source did not affect plasma oestradiol, progesterone or egg production. 6. These results do not support the notion that inclusion of fishmeal in laying hen diets prevents or alleviates feather pecking and cannibalism.

Feather eating in individually caged hens which differ in their propensity to feather peck.

Two experiments examined the responses of 16 individually caged laying hens to the presentation of feathers plucked from dead birds of the same genetic line. In the first experiment, hens known from a previous experiment to be either feather 'peckers' or 'non-peckers' (8 of each) were tested for their propensity to eat feathers in four 10min trials, in which they were offered fresh semiplumes measuring 4-6cm (length), one at a time, in front of their cage. Wide variation between birds was observed in numbers of feathers eaten, pecked, picked-up and manipulated. Fourteen out of 16 birds readily ate presented feathers on one or more occasion and both birds that ate no feathers were non-peckers. Peckers ate, picked-up and manipulated feathers significantly more often than did non-peckers (P<0.05, P<0.01 and P<0.01, respectively). A second experiment investigated the possibility that presence of preen (uropygial) oil might contribute to the attractiveness of feathers to eat. The same group of 16 pecker and non-pecker hens were offered a choice between 20 washed and 20 unwashed semiplumes, presented simultaneously in separate containers, in two 10min trials. Unwashed feathers were eaten, pecked and picked-up in preference to washed feathers by both peckers and non-peckers (P<0.05, P<0.01, and P<0.01, respectively), indicating an attraction towards unwashed feathers, or an avoidance of washed feathers for some reason. Peckers and non-peckers did not differ significantly in their preferences. These results provide evidence of a relationship between feather eating and feather pecking at an individual level. The finding that hens could distinguish between normal feathers and those treated in such a way as to alter their olfactory (but not visual) properties suggests olfactory cues may be of importance in determining the attractiveness of conspecific feathers.