Quick read summary
Royal Society paper: Infrared antenna-like structures in mammalian fur
Infrared antennas in mammalian fur
Mammalian fur consists of many types of hair all with complex and distinct anatomy. Guard hair is a bristle-like hair that often protrudes from the fur and can be likened to the 6000 spines on a hedgehog. These hairs are infrared antenna that provide the animal with all-around infrared threat-warning. The schematic drawings above for a typical rodent, the house mouse, and mouse-like marsupial, the agile antechinus, show two distinct evolutionary paths but the microscopic dimensions are virtually identical. The hairs are in two parts: a tuned antenna that absorbs and propagates radiation in the axis of the hair; and a waveguide that filters and concentrates the radiation. The hairs have the correct topology for an antenna: they are straight, have no rotation or spiraling, and have a wide absorber so that energy can be concentrated into the thermoreceptors at the base of the hair. This is called optical concentration and improves the sensitivity of the sensor. Optical analysis shows the antenna is tuned to a wavelength of 8-12 microns; the best infrared waveband for detecting the heat from a predator. The filter-like structure in the hair shaft can be shown to make the antennas more sensitive and directional.
The sensors are found families that include over 2000 species and are generally the heavily predated animals. They are missing in animals that do not need them, such as, bats, moles, small predators and animals larger than a rabbit. Thermal imaging shows that top predators of small animals have very effective infrared concealment so ambushing snakes, stalking small cats and diving barn owls are virtually invisible in the infrared. Wild mice and rats have been shown to react to infrared sources emulating the heat from the facial disc of an owl. Related species have very similar antenna designs. So the European rabbit and house mouse have evolved hair of different lengths (30 mm and 7 mm respectively) since their evolutionary split some 80 million years ago but the microscopic dimensions remain identical because they are fixed by the optical laws.
Identifying that guard hair performs the function of an infrared antenna is important. Guard hair is the first hair to grow in the embryo of rodents and if this is the evolutionary order then the original purpose of hair was for infrared protection. Hair in therapsids and feathers in archosaurs are thought to have emerged simultaneously in the Triassic period when endothermic animals started to occupy nocturnal niches. Feathers can block infrared far more efficiently than hair. Palaeontologists may be able to pin down the infrared arms race that led to infrared sensitivity in furred therapsids and infrared cloaking in feathered archosaurs. By the start of the Jurassic period (201 million years ago) there were fully-furred mammaliaforms and fully-feathered Coelurosaurian theropods that are believed to have branched into modern birds. Guard hair anatomy appears to change so slowly that it could be a useful supplementary measure of the inter-relatedness of animals. The infrared engineering in hair anatomy operates at a scale similar to the wavelength and much of the detail is not yet understood. This is precisely the photonic engineering needed for optical computing, optical telecoms and sensors, so studying hair anatomy could provide a rich source of bio-inspiration in the field of photonics.
A video version of the research is here and this contains the animal behaviour footage: https://vimeo.com/316626468
All the data and calculations can be found here: Ian Baker – guard hair data and calculations 4th Sept
1 Early evidence for infrared sensitivity in small animals
All objects produce thermal infrared radiation as a natural balance between vibrating atoms and vibrating electro-magnetic waves. At normal ambient temperatures the energy of this radiation peaks at a wavelength of 10 microns and the atmosphere is conveniently transparent between 8 and 12 microns so thermal cameras normally exploit this waveband. Most warm-blooded animals produce a lot of thermal infrared and are very bright in thermal imaging cameras. The striking exceptions are the common predators of small animals. Snakes are intrinsically cold and blend into the infrared background. Small cats produce little infrared from the furred areas and hide their eyes behind a cold nose. Owls in a dive go virtually invisible in the infrared. Infrared concealment hinted that small animals are able to sense the radiant heat of their predators. Eyes cannot function as infrared sensors because of absorption in the vitreous humour. The fur is the only viable source of infrared sensitivity. Figure 1 shows our photomicrograph of hairs from Mus musculus with two zigzag hairs and a much broader guard hair. It illustrates the characteristic dark banding, a ubiquitous feature of the hair of small mammals, with the exception of bats. The band spacing is in the 6–10 μm range and since this matches the wavelength of thermal infrared radiation prompted our efforts to find an infrared interpretation of hair anatomy.
Figure 1. Two hair-types from Mus musculus showing characteristic banding and a scale compatible with an infrared function. The small hairs are so-called zigzag hairs and the large one is the widest part of a spear-shaped guard hair – approximately the same width as a human hair.
2 The surprising resilience of hair anatomy
The oldest hair specimen trapped in amber was dated at 100 million years ago (mya) and shows that modern cuticular features were already present in the early Cretaceous . The oldest fossil with identifiable hair structure (guard hair and underfur) is a rat-like animal (Spinolestes xenarthrosus) dated at 125 mya . Remarkably, our studies have shown that rodents and small marsupials, have equivalent hair-types with almost identical microscopic dimensions, despite an evolutionary split in the early Jurassic, 170 mya . Figure 2 compares the band spacing and cuticle patterns in zigzag and guard hair of Mus musculus and the Australian, mouse-like, marsupial Antechinus agilis, the agile antechinus. Zigzag hair has converged on a common solution with a grating-like structure on the distal-end. Guard hair has diverged but crucially the band spacing in the distal-end and the cuticle periodicity are very similar. The inheritance or convergent evolution of hair anatomy suggests that each hair form has a vital survival function for the animal. In stark contrast, hair forms that are clearly for other purposes, such as whiskers and auditory hairs, have no periodic internal structure.The microscopic dimensions vary little between species. Our measurements show that the 4 g pigmy shrew, Sorex minutus, has similar band spacing to the 2 kg European rabbit, Oryctolagus cuniculus. Understanding the purpose of the banding is crucial to the study. Here we concentrate on guard hair because it has the essential properties required of an infrared sensor.
Figure 2. Schematics and band spacing measurements on zigzag and guard hairs from the rodent Mus musculus and marsupial mouse Antechinus agilis showing similar microscopic dimensions despite an evolutionary split now believed to be in the early Jurassic .
3 Infrared interpretation of hair anatomy
Every physical body continuously emits electromagnetic radiation with a spectrum that depends on its temperature; the so-called black-body or thermal radiation. The sun emits black-body radiation from an outer layer called the photosphere with an average temperature of 5500°C . This temperature provides the familiar solar spectrum with a peak energy in the visible region (Figure 3). At normal ambient temperatures (0 to 30 °C), thermal radiation has an energy peak at a wavelength of 10 μm. The atmosphere is transparent between 8 and 12 μm  and this waveband is exploited by many thermal cameras.
Figure 3. The power spectra for solar radiation and thermal radiation are set by the source temperature. Thermal cameras normally exploit the 8-12 μm band where the atmosphere is most transparent .
There are two infrared adaptations that could provide a vital survival benefit for small mammals. The most obvious given the distribution of guard hairs around the body and the vulnerabilities of these species, is the ability to sense the infrared signal of a potential predator. This would provide an early warning for species that are often ambushed by stealthy and cryptic predators. The role of guard hairs as the antenna part of an infrared sensor is the main focus of this paper. A second potential adaptation, that applies to small, heavily furred mammals with high metabolic demands, is radiative infrared cooling. The ability to stay cool during an energetic escape from a predator is a clear survival advantage. Many features of zigzag hair support an infrared radiator explanation as explained in this website.
Infrared sensory adaptations have been reported widely. Snakes of the Crotalinae sub-family, as well as some members of the Pythonidae and Boidae families, have evolved infrared sensing pit organs . Beetles from the Genus Melanophila, colloquially called fire beetles, can locate active fires from large distances [17,18]. Evidence for infrared sensing adaptions also exist for several hematophagous species, including bats of the sub-family Hemiodontidae (vampire bats) , members of the true-bug subfamily Triatominae (vampire bugs)  and some hard-ticks of the family Ixodidae . Oddly, butterflies of the Genus Triodes , and dogs  have infrared sensitivity. These sensors are thermal detectors that measure the change of temperature of an infrared absorber and it is clear that infrared sensors have evolved many times in Reptilia, Insecta, Arachnida, and Mammalia Classes.
The hypothesis we are testing is that guard hair is the antenna part of an infrared sensor. Receiving antennae in the broadest sense are tuned structures that absorb electromagnetic radiation and concentrate the energy to elicit an electrical signal. Ultra-high-frequency antennae for TV reception are familiar as metallic tuned structures. At shorter wavelengths metal becomes inefficient and at a wavelength of a few centimetres organic polymers are often used, so-called polyrods. These are a class of antennae called dielectric resonator antennae (DRAs) . The rods and cones in the human retina are DRAs that concentrate the electromagnetic energy to initiate a photo-chemical response. Guard hair anatomy has a combination of DRA characteristics and internal features that follow classical first-order optics. Antennae of this nature are directional and sense infrared radiation in line with the axis of the hair. In this respect, the fur behaves like the compound eye of an insect.
Guard hairs in small mammals already have the key requirements of an antenna. They are stiff, straight, have no rotation or spiralling and are uniformly patterned throughout their length to provide transmission of a dominant wavelength. The tip is very fine and long, and this serves to minimise reflection for on-axis infrared photons, a feature that favours the antenna interpretation of guard hair rather than the tactile one.
Optical modelling requires a clear understanding of the physical nature of the dark bands. Through a microscope with back-illumination the bands are black or brown and appear to focus indicating a higher refractive index. This is consistent with eumelanin, a common pigment in mammalian hair that is known to have high refractive index . Some arthropods use alternating layers of eumelanin to produce structural colours by a process called Bragg reflection . Eumelanin absorbs strongly at UV and visible wavelengths but becomes transparent at longer infrared wavelengths . The chemical composition of hair is remarkably conserved between species and over time  and we speculate that this may be driven by infrared transparency because hair keratin in general reaches peak transparency around 10 μm wavelength . We can therefore model the hair shaft as alternating bands of different refractive indices.
There is an almost identical man-made analogy called a Fibre Bragg Grating (FBG) used to filter light of certain wavelengths in an optical fibre  shown in Figure 4. Light guided down the fibre experiences weak reflection at each refractive index change and if this reflected light is out of phase then it forces the light to propagate though the FBG. This is the principle behind anti-reflection coatings.
Figure 4 Fibre Bragg Grating equations can be used to model the periodic bands in hairs
The transmission condition can be defined by simple formulae:
For 2nd order, d=0.75.λ/n equation 1
For 3rd order, d=1.25.λ/n equation 2
Where d is the band spacing, λ is the infrared wavelength and n is the refractive index. The order is the number of wavelengths between adjacent reflections. We have found hair anatomy mainly employs the 2nd and 3rd order equations.
To establish the function of guard hair requires detailed mapping of the shape, internal banding and cuticular features of the whole hair. We have focused the mapping study on three key species from different Orders: Mus musculus, the house mouse, typical of most small rodents, Antechinus agilis, the agile antechinus (a mouse-like marsupial), and Sorex araneus, the common shrew. Published multi-species databases have provided supplementary data for a wider range of animals. The infrared antenna interpretation of guard hair can be tested mathematically because all the morphological detail is bound by first order optics. Mathematical testing on a range of species is crucial to separate infrared adaptations from coincidental mechanical structure. Specifically, we must show that guard hair is tuned to the optimum wavelength for thermal imaging at 8-12 µm. This paper adopts the nomenclature of Teerink  for hair anatomy.
4.1 Evidence from the guard hair of rodents (Rodentia) and rabbits (Leporidae)
Published databases [9,30] have photographs of guard hair from many species of rodents and rabbits that are very similar in structure. The guard hair of the house mouse, Mus musculus, is drawn schematically in Figure 5 with typical dimensions. The distal-end is spear-shaped (often called a shield) consisting of two tubes with uniform, periodic air cavities, connected by a patterned membrane. In the base of the shield the air cavities fill with polymer and merge becoming periodic dark bands in the hair shaft. Lower in the shaft the bands break into a so-called ‘zipper’ consisting of dark hemispheres that rotate around the axis, this section being quite variable between hairs. The proximal end of the hair narrows and has a densely banded section with a prominent cuticle pattern. The shield is interpreted as an infrared absorber and is much wider than the shaft, a structure that promotes absorption in the wide part and concentration of infrared energy into the base of the hair. The ‘zipper’ is interpreted as an optical filter for concentrating the spectral response around 8-12 µm. The ‘zipper’ is present in rodents and rabbits. In fact, Mus musculus, has guard hair with remarkably similar microscopic dimensions to the European rabbit Oryctolagus cuniculus suggesting that the morphology has changed little since the evolutionary split in the late Cretaceous . It is another important test to provide modelling evidence that the hair shaft functions as an effective optical filter.
Figure 5. A schematic of the guard hair of the house mouse Mus musculus is typical on most rodents. The wide section has two rows of regularly spaced air cavities (approximately 600 in total with an average spacing of 10.0 µm) connected by a patterned membrane. The shaft has complex filter-like structures.
As a first step we need some assurance that the FBG equations are appropriate to analyse hair anatomy. With reference to Figure 4, there is an abrupt change of spacing by a factor of 0.6 at the start of the ‘zipper’. For transmission of a single wavelength this must represent an order change. 0.6 exactly matches the transition from 3rd order (1.25.λ/n) to 2nd order (0.75.λ/n). The ‘zipper therefore performs an important function as it only allows transmission of a single tuned wavelength and radiates unwanted wavelengths. The dark hemispheres rotate around the axis in order to cover all possible polarisation angles. The median measured change in band spacing for hairs from the dorsal coat of one mouse (n= 36 hairs) was 0.62, which is very close to ideal considering the variability in the hair samples. The measurement techniques and data are described in . Confirming that the change in band spacing corresponds to an order change improves confidence in the validity of the FBG equations. Furthermore, we now know that the 3rd order equation: λ = n.d/1.25 can be used to determine the tuned wavelength in the shield. We return to the optical analysis of the ‘zipper’ later.
The refractive index of the distal end of the hair, where infrared photons are absorbed, is required to determine the tuned wavelength. The refractive index of α-keratin has not been reported in the infrared at 10 μm. A number of authors, notably , determined the coefficients for the Cauchy dispersion equation , as A = 1.532 and B = 5890 nm2, giving an extrapolated refractive index for α-keratin of 1.53 at 10 μm. The distal part of the hair for all species consists of evenly-spaced air cavities with a keratin-to-air ratio averaging 1:1.22. Ratioing gives an effective refractive index, n = 1.25. Substituting in λ = n.d/1.25, gives d = λ. The tuned wavelength is then simply the measured band spacing in the top of the hair. There may be an absorption benefit of matching the free-space wavelength to the band spacing since it is present in all the species we have studied. The median tuned wavelength of 70 Mus musculus guard hairs from the back of one mouse was found to be 10.0 μm with a standard deviation of 0.30 μm and inter-decile range 9.4 to 10.5 μm . The Teerink database  has well-scaled photographs that can be analysed to extract the band spacing of 16 species of mice, voles, and rabbits providing an average band spacing of 9.9 μm (Table 1). No exceptions have yet been found.
We conclude that a wide range of small mammal species have guard hair with antenna-like features tuned to the infrared window between 8 to 12 μm.
Table 1 The Teerink database  has well-scaled photographs that can be analysed to extract the band spacing (tuned wavelength) for a range of mammals. An average value of 9.9 µm supports our measurements that a wide range of species have guard hair tuned to a wavelength range between 8 and 12 μm.
4.2 Infrared filtering function of the “zipper”
In many species, especially mice and rabbits, there is an unusual topology in the hair shaft (sometimes likened to a “zipper”). Here the banding switches from 3rd order Bragg transmission at 10 microns to 2nd order with a transition zone of complex spiraling features with symmetry around the longitudinal axis. The photo in Figure 6 shows a “zipper” section. The dark features are semi-circular and rotate around the axis.
Figure 6 “Zipper” structure is composed of semi-circular dark bands that rotate around the axis
The shield fulfils the Bragg transmission condition for 10 µm but there are other wavelengths belonging to different orders that are also transmitted. 8-12 µm radiation has good atmospheric transmission but other wavelengths are absorbed by water vapour in the air and so represent the air temperature rather than the subject of interest. The sensitivity of the antenna is improved if these other wavelengths are radiated away. The ‘zipper’ and lower hair shaft can be demonstrated by modelling to fulfil this role. Figure 7a shows a numerical calculation of the infrared spectrum for the antenna with a band spacing of 10 µm showing the unwanted, sideband wavelengths described in detail in . The transmission spectrum is superimposed on the blackbody energy spectrum for 27ºC. Figure 7b shows the transmission spectrum for the ‘zipper’ and lower hair shaft with a band spacing of 6 µm. Note that the peak is still at 10 µm but the sidelobes are shifted. To determine the spectrum of the shield and shaft in combination Figures 7a and 7b are multiplied giving the spectrum in Figure 7c. The spectrum displays strong sideband suppression. Without the filter the signal (from the 8-12 µm band) constitutes 33% of the total energy. In the ideal case presented in Figure 7c, the spectral filter increases the signal to 72% of the total. The ‘zipper’ therefore plays a vital role in enhancing the performance of the antenna. Because the ‘zipper’ functions as a simple disruptive device to stimulate radiation it does not have to be geometrically perfect hence the general variability observed in hair shafts.
Figure 7. An optical analysis of Mus musculus guard hair  shows the benefit of the ‘zipper’ function. Figure 7a shows the transmission spectra of the antenna superimposed on the thermal spectrum for 27°C. Figure 7b shows the transmission spectra of the lower shaft. Figure 7c shows the combined spectra with suppressed sidelobes. The ‘zipper’ enhances the signal-to-noise for sensing infrared radiation at 8-12 μm.
The main heat source of predators is the eye region and the ability to detect such a small source depends on a tight photon collection cone. Because the hair is less than about 10 mm long it will harvest off-axis photons with different wavelengths and this will add noise and reduce the ability to sense on-axis signals. An additional function of the “zipper” may be to expel off-axis (or higher order) photons as illustrated in Figure 8. Photons that are on-axis can propagate through the filter unaffected. Effectively the spectral response becomes tighter around 10 microns and the acceptance angle of the antenna is narrowed.
Figure 8 “Zipper” structure may also serve to expel off-axis photons and tighten the reception angle
The optical filtering function is good evidence for the sensor interpretation because it has no other role but to improve the signal-to-noise of a receiving antenna.
The complementary functions of the absorbing antenna and optical filtering shows the structure to be a highly integrated optical device. At a top-level the optical performance can be modelled quite well but at the microscopic level the fine detail is far from understood. Since the shield and shaft have waveguiding structures with dimensions typically less than the wavelength of infrared radiation there could be useful bio-inspiration in the field of photonics. Whisker antennae cannot match the sensitivity of thermal cameras and we estimate a temperature sensitivity of approximately one degree centigrade. The function appears to be limited to detecting warm-bodied animals from a range of a few metres.
4.2 Evidence from the agile antechinus, Antechinus agilis, a mouse-like marsupial
Antechinus agilis is a mouse-like marsupial from south eastern Australia (weighing 20 g and measuring 100 mm in length). Figure 9 shows the simplest of three antenna-like hair forms found in the fur of Antechinus agilis. As expected from the distant relationship to eutherian mammals, the guard hair differs radically in morphology, but the details that relate to infrared adaptations are similar. The hair cross-section is nearly circular, contains around 250 periodic air-cavities with an average spacing of 10.2 μm and this section has no significant cuticle pattern. The topology changes abruptly into a tapered shaft with a very prominent and repeatable cuticle pattern, especially at the base.
Figure 9. Guard hair of antechinus agilis, a mouse-like marsupial from south eastern Australia. The wide distal end has regularly-spaced air cavities (250 in total with a spacing of 10.2 µm). The cuticle pattern on the shaft begins abruptly at the start of the shaft and has a wave-like nature and repetition distance suggestive of an optical function. The cuticle pattern has a similar scale to that of Mus musculus in Figure 5.
The Antechinus agilis guard hair is an outstanding subject for further study as it is symmetric and cleanly divided into a simple antenna and a tapering shaft. The cuticle pattern on the shaft in Figure 9 is called an elongated diamond pattern and is found in the hair of many animals. The symmetrical, wave-like nature of the pattern and the constant repetition distance (16 μm) hints at an optical function. The repetition distance is 1.5x the tuned wavelength and a possible purpose is to inhibit the radiation of 10 μm photons so they propagate down the tapering shaft whilst allowing other wavelengths to radiate. Preferential propagation at 10 μm may be another way of achieving spectral filtering. Mus musculus has a very similar cuticular pattern but with rotation of the crowns around the shaft. There are analogous man-made optical structures in the field of holography. Holographic patterns can be etched into surfaces to control how light is propagated or reflected by a process called interference . The models employed by computer-generated holography specialists could test the wave-guiding function of the elongated diamond pattern. Confirmation would establish that the whole structure of guard hair has a photonic basis and is an important conclusion for bio-inspiration in photonics.
4.3 Evidence from antenna-like hairs in shrews
Whisker-like guard hairs are reported in shrew fur  but our samples lacked them. Instead the predominant sensor-like hair in shrews is radically different to that of rodents and marsupials with a compact antenna and a zigzag shaft. Shrews provide an opportunity to test if the antenna is tuned to a single wavelength. The analysis concentrated on the common shrew, Sorex araneus, (weighing 10 g and measuring 80 mm in length) supported by data on the pigmy shrew, Sorex minutus, (weighing 4 g and measuring 55 mm in length). The antenna-like hairs of these species are very similar, differing slightly in length but not in proportion to the body size. Figure 10 shows the hair anatomy of Sorex araneus.
Figure 10. The infrared sensor-like hair in the common shrew, Sorex araneus, has a different concept to rodents consisting of a complex and sophisticated-looking antenna on a zigzag shaft. The antenna is crucially tuned to a single wavelength. The shaft has 3 or 4 tapered bends in different directions that may be performing the same spectral filter function as the shaft of Mus musculus in Figure 5.
With reference to Figure 10, the band spacing in the distal air cavity section is 9.2 μm and shrews in general appear to tune to a lower wavelength, but still well within the thermal imaging band. A key feature is the change from air cavities to a solid polymer section at the widest part of the hair. We expect the effective refractive index to change from 1.25 for the air cavity section (as described in Section 4.1) to a higher value in the solid part. If the structure is tuned to a single wavelength the band spacing must adjust to accommodate the change in refractive index. Figure 10 shows that there is indeed a band spacing change and it corresponds to a refractive index of 1.50 in the proximal end, a value close to the 1.53 derived in Section 4.1. The confirmation that the band spacing is compatible with a single wavelength is strong evidence to support the antenna model because there is no obvious reason for this structural accommodation. The measurement techniques and data are presented in .
The Sorex araneus hair studied in Figure 10 has a zigzag shaft and in common with other zigzag hairs each angle projects in a different direction. The narrow parts are fragile and there must be a strong selection pressure (e.g. an optical function) to overcome the mechanical weakness this feature introduces into the hair. We speculate that this may be a strategy for radiating unwanted photons covering all polarisation angles, much as rodents and rabbits employ the ‘zipper’ structure in the shaft. The shrew family appear to have sophisticated antennae with many detailed features that are yet to be explained.
5 Distribution of animals with sensor-like guard hairs
There are numerous hair photographs in papers, books and on-line with enough detail to recognise them as potential infrared antennas. The orders Rodentia and Lagomorpha have many species with very similar sensor-like hairs. Shrews in the family Soricidae have a different antenna approach but within the family are very similar. The marsupial, Antechinus agilis, has a third approach. Guard hair anatomy appears to evolve slowly and could be a measure of the inter-relatedness of animals. Potentially, over 2180 species may have infrared sensitivity as listed in Table 2
Table 2 Family list of animals with identifiable sensor-like guard hairs from published photographs
Bats lack antenna-like guard hairs probably because flight does not allow them to dwell long enough on the scene to collect enough signal. They have a type of guard hair that is modified to wrap over the underfur and present an aerodynamic shape, but it lacks antenna-like features. Notably, the European mole, Talpa europeae, lacks antenna-like guard hair presumably because it has no need for 360-degree infrared threat warning in underground tunnels. Moles also have a type of guard hair to protect the underfur but these are modified underfur hairs with no antenna-like features. We have not found any photographic evidence in the literature of infrared sensors in animals larger than a rabbit or in small ground-based predators, such as weasels. Clearly, a wide species survey is required to confirm this but sensor-like guard hairs appear to be restricted to heavily predated animals with a need for infrared vigilance.
6 Can guard hair anatomy be a measure of inter-relatedness?
Guard hair anatomy could support studies on the inter-relatedness of animals. Although we have analysed relatively few species in detail (10), there is an indication that guard hair anatomy is highly resilient over time. The guard hairs of the house mouse, Mus musculus, and the European rabbit, Oryctolagus cuniculus, are remarkably similar despite an evolutionary split in the late Cretaceous . The slow evolution may arise from the over-riding importance of complying with the optical laws so in the case of the mouse and rabbit the length of the hair (~7 mm and ~20 mm respectively) has changed in the past ~80 million years but the periodic structure has remained identical as shown in Figure 11.
Figure 11 Rabbits and mice have virtually identical guard hair structures despite a probable split 80 million years ago
Likewise within the family Soricidae (shrews) species appear to have identical guard hair structure but their structure differs radically from Glires suggesting a much more distant split. We only have one marsupial representative, agile antechinus, but this has a third radically different approach and the latest literature proposes the Therian split to be in the 165-190 Ma range. From this piece of evidence alone the separation of shrews should be in the same time-frame.
7 Literature evidence for an infrared sensor at the base of the antenna
The paper has concentrated on the antenna part of the sensor and not discussed the thermoreceptor at the base of the hair. Guard hair becomes narrow and angular at the base and loses its internal structure. Our interpretation is that the infrared photon becomes a surface wave with most of the energy outside the shaft. The physics predicts the likely position of a temperature sensor is around the hair shaft just below skin level but above the dermis which should have higher infrared absorption. The guard hair of Mus musculus is reported to have a ring of 10 µm diameter Merkel cells around the shaft at the interface of the epidermis and dermis  as shown in Figure 12. The Merkel cells in guard hairs are enervated by Aβ-fibres that are reported to transmit signals rapidly to the spinal cord and more directly, via the dorsal column nuclei, to the cortex . This is consistent with the rapid response required to an infrared threat. Despite protruding from the fur guard hairs have surprisingly low tactile sensitivity compared with other hair forms  and this also supports the antenna hypothesis. There is convincing evidence that Merkel cells respond to mechanical stimuli but it may be difficult to separate a mechanical transducer from a thermo-mechanical transducer. Specialists in this field may be able to review the research on mechanoreceptors and enervation around guard hairs to test the infrared antenna functionality.
Figure 12 Guard hair has a ring of Merkel cells in the correct position for an infrared sensor 
8 Evolution of guard hair and consequences for feather evolution
It may be significant that guard hair is the first hair to develop in mice embryos  and if this is the evolutionary order that hair emerged it indicates that the original purpose was infrared protection. Benton  argues for concurrent emergence of hair in therapsids and feathers in archosaurs during the Triassic period when endothermic animals started to occupy nocturnal niches. Our thermal imaging has shown that feathers can block infrared far more efficiently than hair, as exemplified by the downy feathers on the legs of owls. With this fragment of information palaeontologists may be able to pin down the infrared arms race that may have led to infrared sensitivity in furred therapsids and infrared cloaking in feathered archosaurs. By the start of the Jurassic period (201 million years ago) there were fully-furred mammaliaforms and fully-feathered Coelurosaurian theropods that are believed to be the ancestors of modern birds .
9 Evidence for infrared countermeasures in top predators of small animals
The Leonardo Merlin camera was used to study the behaviour of the top predators or small animals. The domestic cat, Felix catus, has generally low infrared radiance apart from the eyes. Cats have cold noses when active with very little infrared emission. In the stalking pose the head is still, the eyes squinted and the nose is projected forward to compensate the emission from the eyes (Figure 13). Small animals are not able to resolve fine details in the face so the cat is able to blend into the infrared background.
Figure 13 The cold nose enables cats to hide infrared emission from the eyes
For this strategy to be successful the resolution of fur sensors must blur the eyes and nose area. We can get a crude estimate of the resolution using a pouncing distance of 60 cm and this suggests an acceptance angle for each sensor of about 6 degrees. Over a hemisphere this is equivalent to about 6000 pixels or approximately the visual performance of a house fly.
Cold-blooded snakes can barely be seen on thermal cameras as they blend perfectly with their background. Even during a strike thermal videos of snakes show no infrared emission so they are virtually indetectable by prey animals.
Owls have strong infrared emission from the armpit area but when they dive this is hidden behind the body. The common barn owl, Tyto alba, has a cold centre stripe in the facial disc equivalent to the cold nose of a cat. Figure 14 shows a dive onto a mouse filmed with the Merlin camera set at an exposure of 1 millisecond. Owls can control the temperature of their feet and the downy leg feathers block emission from the upper legs. The eyes become slits and the whole facial disc appears to cool and remains cool for several seconds after landing. Barn owls may be specialists in hunting the most vigilant of prey.
Figure 14 Common barn owls have almost perfect infrared concealment in a dive
10 Animal behaviour experiments to test for infrared sensitivity
Experiments to test for infrared sensitivity in small, wild rodents are at this stage indicative and must be repeated by other workers. The experimental techniques worked very well however and can be easily copied. The experiments were performed at Peter Wheildons Wildlife Photographic Centre in Otterbourne, Hampshire. There are wild tawny owls, Strix aluca, that nest in the centre so local rats are primed for the threat. An infrared source was built that matched the size and infrared emission intensity of the facial disc of a tawny owl. When the source was cool the rats paid no attention. When the source was switched on the rats disappeared and only returned warily as in Figure 15A. When the source was put into an unnatural position the adults studied it nervously (Figure 15B) but younger rats remained terrified (Figure 15C) with a bobbing-head response. All the videos from the animal behaviour trials can be viewed at https://vimeo.com/316626468
Figure 15 Brown rats respond nervously to an IR source emulating the facial heat from an owl. Young rats (C) remain terrified and bob their heads nervously
Another experiment utilised a Bushnell camera as an infrared source. These cameras have an infrared motion sensor that detects a passing animal and starts a video. At night it also switches on a 940 nanometre, light-emitting diode (LED). Figure 16 shows that the illuminator produces strong infrared radiation but this can be blocked very easily with a glass slide. Rodents should not be able to see the illuminator so we can easily test for infrared sensitivity. Coincidently the shape and size of the illuminator is similar to an owls facial disc.
Figure 16 The illuminators of Bushnell cameras produce strong thermal infrared emission and can readily be used as a switchable infrared source. The infrared can be blocked with a glass sheet.
The camera was placed where tawny owls frequently perched. On nights when the infrared was blocked by a glass sheet there was no awareness of the camera from wood mice or rats and feeding was relaxed. When the mice were exposed to the infrared source they froze as in Figure 17A with the tail vertical and twitching (as in a decoy). This is an extraordinary reaction to an owl threat. It might be a fear-reaction from a lengthy co-existence with heat-sensitive snakes since mice are able to shed their tails. Thermal cameras should be able to confirm if the body literally freezes while the tail stays hot. Rats also froze under the infrared source.
Figure 17 Wood mouse (A) and brown rat (B) frozen in response to infrared emission from the illuminator of a Bushnell camera. No reaction when the infrared is blocked by a glass sheet
There is potential for bio-inspiration in photonic engineering – an important technology for future telecoms, optical computing and sensors. The dimensions in hair anatomy are similar and often less than the wavelength of infrared and this is reflected in many detailed features that clearly operate at the electro-magnetic wave level. A good example is the elongated diamond pattern of Antechinus agilis in Figure 6 that we believe provides a wave-guiding function at a wavelength of 10 μm. The key next step is to confirm the role of the cuticular patterns by expert microscopy and modelling. This would establish that every detail of guard hair anatomy has a photonic purpose and it would justify further study by the photonics community for bio-inspiration.
12 Future work
Expert microscopy is needed to improve our knowledge of the guard hair anatomy so we can use electro-magnetic modelling techniques to study the structures. The work reported so far has concentrated on sensors within the fur for the purpose of providing the animal with all-round infrared threat warning. It was noticed in the antechinus study that there were three types of infrared sensor. Our work concentrated on the simplest structure. There were two other longer hairs with more complicated internal structure but identical cuticle patterning. It is possible this animal was performing more sophisticated infrared scene analysis. Much more work is needed to map the details of these hairs so we can interpret the function.
More photographs of guard hair
The common shrew, Sorex araneus, shows distinct banding across the width of the hair and the awn is only 2 mm long in the sample studied. The stitched image in Figure 18 is compressed by 16x in the length of the hair and illustrates the consistent “washboard” effect. It shows a uniform tuning to 9.3 microns in the left hand side (air cavity section) and many detailed features that are not yet understood. The tip of the hair is on the left and beginning of the shaft to the right. Sorex araneus lacks optical filtering in the hair shaft to reject off-axis (or higher order) photons to tighten the reception angle of the antenna. However it employs a zigzag design and it is possible that the sharp angles of the zigzag perform the same function to shed off-axis photons and shrews can employ this because the hair is so short.
Figure 18 The shield of the common shrew transitions from air cavities to keratin bands that are tuned to one wavelength around 9.3 microns. Many of the other features ‘look’ optical but we have yet to resolve the function
Mus musculus has a shield typical of rodents that is 50 microns wide (Figure 19). Scaled up to human dimensions it would 3 millimetres across and is a major physiological investment by the animal.
Figure 19 The shield of the house mouse has two outer buttresses tuned to a wavelength of 10.0 microns and a thinner membrane with more random structure but roughly 10 microns periodicity
Hair is difficult to photograph because it is so optically active but this front illuminated microphotograph of the house mouse in Figure 20 shows the smooth surface of the shield unlike hair shafts that are often deeply sculptured. The pattern that can be resolved is a wave-like pattern almost identical to that found on the hair shaft of the marsupial agile antechinus. The difference is the rotation of the pattern around the shaft whereas antechinus is straight.
Figure 20 The shield of the house mouse has weak surface features that are very similar to the marsupial mouse, antechinus, suggesting that this has a role in propagating the photon down the hair
Figure 21 shows side-by-side comparison of the guard hair of three distantly related species that we have studied to build evidence for an infrared antenna explanation.
Figure 21 Detailed anatomy of most abundant guard hair in the three nominated species
The authors wish to acknowledge the support of Leonardo MW Ltd for use of high-performance thermal imaging cameras and optical microscopes. Thanks to David J Baker for valuable advice in preparing the manuscript. Amongst many contributors we would like to thank Dr Marissa Parrott of Melbourne University, Dr James Gilbert of ANU, Philip Oakley and Kim Lake of Leonardo, Dr Arvind D’Souza of DRS (USA), Professor Mike Benton of Bristol University, Nik Knight, Chairman of Hampshire Bat Group and Peter Whieldon of Otterbourne Wildlife Photographic Centre featuring Dyson.
- Cena K and Monteith JL. 1975 Transfer processes in animal coats. II. Conduction and convection. Proc. R. Soc. London – Biol. Sci. 188, 395–411. (doi: 10.1098/rspb.1975.0027)
- Blanchard DC. 2010 Of lion manes and human beards: Some unusual effects of the interaction between aggression and sociality. Front. Behav. Neurosci. 3, 1–7. (doi: 10.3389/neuro.08.045.2009)
- Dean I and Siva-Jothy MT. 2012 Human fine body hair enhances ectoparasite detection. Biol. Lett. 8, 358–361. (doi: 10.1098/rsbl.2011.0987)
- Caro T. 2005 The adaptive significance of coloration in mammals. Bios 55, 125–136. (doi: 10.1046/j.1095-8312.2003.00190.x)
- Sarko DK, Rice FL and Reep RL. 2011 Mammalian tactile hair: Divergence from a limited distribution. Ann. N. Y. Acad. Sci. 1225, 90–100. (doi: 10.1111/j.1749-6632.2011.05979.x)
- Bauer GB, Reep RL and Marshall CD. 2018 The Tactile Senses of Marine Mammals. International Journal of Comparative Psychology, 31. (https://escholarship.org/uc/item/1vk1c9z)
- Dry FW. 1926 The coat of the mouse (Mus musculus). Journal of Genetics, 16, 287-340
- Hausman LA. 1920 Structural Characteristics of the Hair of Mammals. Am. Nat. 54, 496–523
- Teerink BJ. 2003 Hair of West European mammals: atlas and identification key. Cambridge University Press. ISBN-13 : 978-0521545778
- Zimmerman A, Bai L and Ginty D, 2014 The gentle touch receptors of mammalian skin, Science, 21, Vol 346, Issue 6212, pp. 950-954, (doi: 10.1126/science.1254229)
- Lechner SG and Lewin GR, 2013 Hairy Sensation, Physiology Society, Volume 28, Issue 3, pp 142-150, (doi: 10.1152/physiol.00059.2012)
- Vullo R, Girard V, Azar D and Neradeau D, 2010 Mammalian hairs in early Cretaceous amber, Naturwissenschaten, Vol 97, Issue 7, pp 683-687. (doi: 10.1007/s00114-010-0677-8)
- Martin T, Marugán-Lobón J, Vullo R, Martín-Abad H, Luo Z, Buscalioni AD. 2015 A Cretaceous eutriconodont and integument evolution in early mammals, Nature, 526 (7573): 380 (doi: 10.1038/nature14905)
- Benton MJ, Dhouailly D, Jiang B and McNamara M. 2019 The early origin of feathers, Trends Ecol Evol, Vol 34(9), 856-869, (doi: 10.1016/j.tree.2019.04.018. Epub).
- Mihalas D, Mihalas BW and Arnett W. 1984 Foundations of Radiation Hydrodynamics Oxford University Press. New York (doi: 10.1063/1.2815048)
- Gracheva EO, Ingolia NT, Kelly YM, Cordero-Morales JF et al. 2010 Molecular basis of infrared detection by snakes. Nature 464, 1006–1011. (doi: 10.1038/nature08943)
- Evans WG. 2006 Infrared Radiation Sensors of Melanophila acuminata (Coleoptera: Buprestidae): A Thermopneumatic Model. Ann. Entomol. Soc. Am. 98, 738–746. (doi: 10.1603/0013-8746(2005)098[0738:irsoma]2.0.co;2)
- Schmitz H, Bleckmann H. 1998 The photomechanic infrared receptor for the detection of forest fires in the beetle Melanophila acuminata (Coleoptera: Buprestidae). J. Comp. Physiol. – A Sensory, Neural, Behav. Physiol. 182, 647–657. (doi: 10.1007/s003590050210)
- Kürten L and Schmidt U. 1982 Thermoperception in the common vampire bat (Desmodus rotundus). J. Comp. Physiol. 146, 223–228. (doi: 10.1007/BF00610241)
- Schmitz H, Trenner S, Hofmann MH, Bleckmann H. 2000 The ability of Rhodnius prolixus (Hemiptera; Reduviidae) to approach a thermal source solely by its infrared radiation. J. Insect Physiol. 46, 745–751. (doi: 10.1016/S0022-1910(99)00163-8)
- Mitchell RD, Zhu J, Carr AL, Dhammi A, Cave G, Sonenshine DE and Roe RM. 2017 Infrared light detection by the haller’s organ of adult american dog ticks, Dermacentor variabilis (Ixodida: Ixodidae). Ticks Tick. Borne. Dis. 8, 764–771. (doi: 10.1016/j.ttbdis.2017.06.001)
- Campbell AL, Naik RR, Sowards L and Stone MO. 2002 Biological infrared imaging and sensing. Micron 33, 211–225. (doi: 10.1016/S0968-4328(01)00010-5)
- Bálint A, Andics A, Gácsi M, Gábor A, Czeibert K, Luce CM, Miklósi Á and Kröger RHH. 2020 Dogs can sense weak thermal radiation. Sci. Rep. 10, 1–9. (doi: 10.1038/s41598-020-60439-y)
- Mongia RK and Bhartia P. 1994 Dielectric resonator antennas—A review and general design relations for resonant frequency and bandwidth. Int. J. Microw. Millimeter‐Wave Comput. Eng. 4, 230–247. (doi: 10.1109/MAP.2010.5687510)
- Stavenga DG, Leertouwer HL, Osorio DC and Wilts BD. 2015 High refractive index of melanin in shiny occipital feathers of a bird of paradise. Light Sci. Appl. 4, e243–e243. (doi: 10.1038/lsa.2015.16)
- Neville AC. 2012 Biology of the arthropod cuticle. Springer Science & Business Media. ISBN 978-3-642-80910-1
- Estévez-Martínez Y, Velasco-Santos C, Martínez-Hernández A-L, Delgado G, Cuevas-Yáñez E, Alaníz-Lumbreras D, Duron-Torres S and Castaño VM. 2013 Grafting of multiwalled carbon nanotubes with chicken feather keratin. J. Nanomater. (doi: 10.1155/2013/702157)
- Dhouailly D, Godefroit P, Martin T, Nonchev S, Caraguel F and Oftedal O. 2019 Getting to the root of scales, feather and hair: As deep as odontodes? Exp. Dermatol. 28, 503–508. (doi: 10.1111/exd.13391)
- Othonos AS and Kalli K. 1999 Fiber Bragg gratings fundamentals and applications in telecommunications and sensing. Artech house. (doi: 10.1063/1.883086)
- Debelica A and Thies ML. 2009 Atlas and key to the hair of terrestrial Texas mammals. Museum of Texas Tech University. ISBN13 978-1-929330-17-1. (doi: 10.5962/bhl.title.142652)
- Huchon D, Madsen O, Sibbald MJJB, Ament K, Stanhope MJ, Catzeflis F, De Jong WW and Douzery EJP. 2002 Rodent phylogeny and a timescale for the evolution of Glires: evidence from an extensive taxon sampling using three nuclear genes. Mol. Biol. Evol. 19, 1053–1065. (doi: 10.1093/oxfordjournals.molbev.a004164)
- Baker, Ian (2020), Ian Baker – guard hair data and calculations 4th Sept, Data from: Infrared antenna-like structures in mammalian fur, Dryad, Dataset, https://doi.org/10.5061/dryad.np5hqbzrg
- Freyer P, Witts BD and Stavenga DG. 2018 Reflections on iridescent neck and breast feathers of the peacock, Pavo cristatus, royalsocietypublishing.org/doi/10.1098/rsfs.2018.0043
- Oscurato SL, Salvatore M, Borbone F, Maddalena P and Ambrosio A. 2019 Computer-generated holograms for complex surface reliefs on azopolymer films. Sci Rep 9, 6775. (doi: 10.1038/s41598-019-43256-w)
- Baker, Ian (2020), Infrared sensors in mammalian fur – Ian Baker.WMV, Dryad, Dataset, https://doi.org/10.5061/dryad.2z34tmpkf
- Forni MF, Trombetta-Lima M and Sogayar MC. 2012 Stem cells in embryonic skin development. Biol. Res. 45, 215–222. (doi: 10.4067/S0716-97602012000300003)
- Winter Y and von Helversen O. 1998 The energy cost of flight: do small bats fly more cheaply than birds? J Comp Physiol B 168, 105–111. (doi: 10.1007/s003600050126)
Last updated – 23rd January 2021