Infrared cooling in small mammals

The Leonardo “Merlin” thermal camera can be operated with exposure times as short as 1 millisecond and effectively freeze animals in motion. The left-hand image of a brown long-eared bat, Plecotus auratus, shows warm membranes and limbs that are presumably suffused with blood during rest. The right-hand image is a bat that has returned after flight showing cool extremities and fur that is radiating infrared strongly. This is contrary to a common assumption that bats lose excess heat through the wings. The membranes appear semi-transparent in the infrared and the muscles seem to be adapted to run cold, as in the feet of ducks. The physiology appears to prioritise the avoidance of freezing on a cold night. Bats of this size generate about a Watt of heat according to VO2 measurements, the same as birds of similar weight. There is no obvious thermoregulation mechanism. The fur appears to be the only source of heat but this is counter-intuitive, since fur should be a good insulator and bats are too small to radiate much energy.

Infrared emission from the fur is the only identifiable cooling mechanism yet this will be negligible on hot nights when energetic animals most need cooling and there is little evidence bats are grounded in hot weather. This enigma was the main reason to look more closely at the hair anatomy of small mammals for clues on the physics of thermoregulation.


Characteristic features in the hair of small mammals

Under a microscope the hair anatomy in small animals shows many characteristics that are shared across a broad range of species. In particular the underfur hairs of most small mammals have a characteristic banding as shown in the collage below of common species. The bands are refractive index changes with the dark areas having higher refractive index.

Best photographs

The internal structure of underfur is identical across most small mammals

The L. A. Hausman database published in 1920 (Hausman, 1920) describes the hair anatomy of over 300 mammals and 60% of the ground-based mammals listed have internal bands with dimensions in the infrared wavelength range, typically 7.3 microns for small mammals, rising to 9.0 microns in rabbit-sized mammals. All bat specimens have external sculpturing with similar dimensions (Sessions, 2014). By studying photographs from the literature it is possible to identify the families with specimens that have a banded hair structure – listed in Appendix I below. This includes distantly related animals from the Old World, New World, Australia and Madagascar. A key observation is the similarity between marsupials and placental mammals. The marsupial feathertail glider (Acrobates pygmaeus) has a band spacing of 6.80 microns compared with the average band spacing for small rodentia of 7.3 microns. This is despite a split now thought to be over 150 mya.  


The fossil record

The scanning electron microscope photo shown below is coronal hair found in the fossilised stomach contents of bat Palaeochiropteryx tupaiodon (presumably due to grooming). The photo is by E Haupt (Brussels Museum of Natural History). The fossil was found in the Messel pit near Darmstadt in Germany and dated at 49 million years. It shows that modern hair anatomy  was already well developed in ancestral bats. 

The oldest hair specimen trapped in amber was from 100 mya (early Cretaceous from France) reported by Vullo, 2010. The authors conclude that modern cuticular features were present for most of mammalian history. The oldest fossil hair with identifiable hair structure (compound hair follicles with primary and secondary hair) appears to be a 125 mya rat-like animal (Spinolestes xenarthrosus) reported by Martin, 2015.

From fragments of data we can conclude that hair anatomy has been inherited and sustained in many distantly related species possibly over the whole of mammal evolution. Interpreting the function of the various hair types needs detailed measurements and optical analysis.


Infrared interpretation of mammal hair

To understand the function of this hair type – first some background. Fibre optics are used very effectively to transmit light over long ranges. Once inside the fibre the light is trapped by a combination of two factors: firstly, light travels slower in a medium of higher refractive index so energy is continuously pulled back into the fibre; and secondly, light is reflected off internal surfaces unless a critical angle of incidence is exceeded. For keratin with a refractive index of 1.58 this angle of incidence is 38 degrees so it is virtually impossible for a photon to escape an ideal fibre. 

There are no mechanical reasons for the banding so we can assume the refractive index changes are deliberate and have an optical basis. These structures are well understood in the field of fibre optics and are known as Fibre Bragg Gratings, FBGs. They are used to manipulate light within a fibre. By analogy we can postulate that mammal hair is for manipulating infrared within the hair shaft. FBGs have periodic refractive index changes that can reflect or transmit a given wavelength by interference between the reflecting interfaces (like the blooming on camera lens) as illustrated below.

We can analyse hair anatomy using very simple FBG equations to calculate the tuned wavelength. 

With certain patterns of band spacing infrared photons can be ejected and the hair shaft becomes a radiating dielectric antenna. In order to search for patterns the whole hair should be mapped. This mapping has been performed on a wide range of mammals including bats, mice, shrews, moles and rabbits and a picture has emerged.

To read more: Fibre Bragg gratings


Mapping the band structure of hairs

The European mole (Talpa europaea) has helical underfur hairs that show banding patterns typical of most mammals. The spines presumably create a spring mattress effect.

Website final 2-6-17

Illustration of band spacing in the European mole

When the band spacing is measured there is some variation from hair to hair but the pattern follows a consistent template of oscillations superimposed on a chirp over the range from 10 to 6 microns. Here are two examples for mole hair:Slide1

Two examples of band spacing in European mole

Mole hair has many abrupt changes of spacing with average negative transitions of 0.800 and average positive transitions of 1.226. The chirp range and oscillations match 2nd order Bragg interference. The equation relating the wavelength (λ) with band spacing (d) is λ = nd/0.75 where n is the effective refractive index. For a tuned photon a band spacing change of 0.8 diffracts the photon to the escape angle for a refractive index of 1.25. The abrupt transition shown below is typical of mole hair and this example shows a disturbance for ejecting a photon of approximately 15.4 microns. We can only estimate the wavelength because we do not know the refractive index accurately.

Mole image 1

Typical band spacing change for photon ejection

The banding patterns in mole hair could still be arbitrary perhaps arising from cyclic conditions during growth so a distantly related animal, the European rabbit, Oryctolagus cuniculus, was chosen to compare statistics. The graphs below shows typical rabbit and mole hairs with statistical data on measurements from many hairs. Rabbit hairs are air-filled with a lower effective refractive index leading to a wider average band spacing, however, the average transitions are very similar (0.8 and 1.22) and beyond coincidence. We can propose from the oscillations and chirp that both are radiating dielectric antennas based on 2nd order Bragg interference with the prime function to radiate infrared in the range 13 to 19 microns.

Infrared adaptations in small mammals - master

Rabbit and mole band spacing have virtually identical statistics



The house mouse (Mus musculus) – a single wavelength antenna

Detailed studies of bats, rabbits and mole hair hinted at wavelengths around 13 to 18 microns but there is a wide error margin because we do not know important parameters. In the house mouse, Mus musculus, the antenna design is completely different and much more sophisticated perhaps reflecting the number of generations. The hair type known as zigzag hair (because it has kinked hair shaft) is shown below (reproduced from photographs). It is the dominant hair form in Mus musculus fur. Zigzag hair is complex but highly consistent enabling the infrared properties to be well established. 



The zigzag hair of the house mouse has a reflection grating at the tip to project infrared back into the fur

The house mouse hair shaft starts almost circular and flattens so by the end it is blade-like 18 microns wide and 4 microns thick. The band spacing changes smoothly down the shaft to compensate for the cross-sectional area change which affects the local refractive index. This lends support to a very narrow tuned wavelength. At the end of the shaft is a so-called awn that has teeth within an air-filled cylinder. This can be interpreted as a first-order diffraction grating (shown in yellow). Diffraction gratings work most efficiently with s-polarised photons and the blade-like hair shaft may have evolved to align photons to the grating. The three kinks are narrow (4 microns diameter), shiny and lack banding. They may have a function to expel high order photons into the hair mass or they may be part of the process to align the polarisation. The purpose of zigzag hair appears to be to perform a controlled radiation from the awn back into the hair mass.


Calculating the tuned wavelength

The awn structure is also tightly tuned and comprises air cavities in a cylindrical tube with elegant sculpturing to create a highly efficient, first order, reflective grating. The optical thickness of the teeth is the same as the air gap (assuming a keratin refractive index of 1.58) – a key requisite of a reflecting grating to ensure all the reflections are in phase.

By measuring the path difference between adjacent teeth we can pin down the wavelength without having to estimate a refractive index or assume an order. There is still some uncertainty about the launch angle but we can make a good prediction of 40 degrees for optimum grating efficiency as shown in photograph below. The average predicted wavelength is 16.25 microns with a probable range of 15 to 17 microns.


Assuming the grating is first order we can assign a tuned wavelength of 16 microns

We can use the wavelength predicted from the grating to check the tuning of the hair shaft. The schematic below shows the predicted refractive index for various segments of the hair shaft. It is based on 2nd order Bragg transmission because there is no solution for 1st or 3rd order.

The value is 1.36 for root of the hair with a cross-sectional area of 140 microns squared and 1.24 for the flat end section with a cross-sectional area of 70 microns squared. These are realistic values and reinforce the 16 microns prediction.

The tuned wavelength prediction of 16 microns is difficult to justify because we have no known sources and this wavelength is strongly absorbed by most materials including water vapour in the air.


Thermal imaging evidence for long wavelength infrared

Zigzag hair appears to radiate infrared photons back into the fur. This may be a way of staying warm on cold nights or it may be to reduce the brightness of the animal so that predators, such as snakes, cannot easily tune into the peak wavelength. If the peak wavelength is absorbed and re-radiated as thermal infrared the brightness at 16 microns is reduced by two orders of magnitude. Water vapour absorption should result in local heating and indeed some thermal images of animals show a glow (see the ‘black hot’ photos below) that may be absorption in humid conditions. Some camera effects could also be responsible.


Some thermal images show glowing air around hot animals


Supporting evidence from the common shrew, Sorex araneus

Mus musculus and Sorex araneus are distantly related (97 mya) but the zigzag hairs have the same basic design. Sorex araneus has much shorter hair with only two kinks. The hair shaft has banding with similar spacing to Mus musculus implying a similar tuned wavelength. The awn is different however consisting an air-filled cylinder with cavities spaced at 10.8 microns. This can be interpreted as a tapered end-fire antenna tuned to 16 micron radiation. The awn is kinked 60 degrees so that the radiation is projected sideways which may be a predator adaptation. The radiator is not as sophisticated as that of Mus musculus but achieves a similar function and shows tuning to about 16 microns.

Optical analysis of shrew zigzag hair shows similar tuning to 16 microns


Bats – another antenna design – same result

Bats have tiny bodies yet in flight the fur forms an aerodynamic shape that holds even at flight speeds as shown in the soprano pipistrelle, Pipistrellus pygmaeus, image below. Working with bat hair is difficult because of a high susceptibility to static electricity and hairs readily gather positive charge. Bats may be using electro-static stiffening to generate a resilient pelage. Holding this charge in high humidity may have driven the evolution of bat hair. All bats use external features with sharp edges to hold charge but since the dimensions are small compared with the wavelength of infrared, bat hair still behaves like a periodic refractive index change and we can still use the Bragg interference equations. 


The artists impression of barbastelle bat hair is shown below with a general oscillation and chirping leading into a more aerodynamic function near the tip. Only one phasing section is shown here but there are on average 4-5 per millimetre and these are thought to be the main disturbance features.


Illustration of barbastelle bat hair

A typical phasing section is shown below. Here an extra scale is inserted (on the lower side) so that the average scale spacing is different between top and bottom. The ratio of 5 to 4 is the most common bilateral phasing pattern.     Slide21

Typical bilateral phasing pattern in barbastelle bat hair

The working model is that the scale patterns diffract the photon to the escape angle. The ratio of scale lengths on each side has been logged in detail giving an average ratio of 0.809 which is remarkably similar to the 0.800 ratio for the European mole. Bats have very efficient broadband antennas because of the many interference combinations but, considering only higher order combinations, a likely wavelength range is between 14 and 18 microns. 

It needs to be reiterated that the hair banding patterns vary in detail from hair to hair but generally follows a common template. The individual interference features are probably not crucial. The coherence length of infrared photons is considerably longer than the hair length. The modern view in physics is that photons ‘sample’ all possible routes through a complex medium before selecting the one with NO destructive interference. The dielectric antenna function of mammal hair only requires enough disruptive features to ensure radiation.


Discussion on the hair follicle anatomy 

If zigzag hair is a radiating antenna the infrared photon must originate in the bulb of the hair. Thermal infrared is bound by strong radiation laws and no optical coating or shape can make a surface emit more power than the laws allow. The bulb of the hair is so small that the total radiation is negligible. A further problem is that thermal radiation is spread over a wide wavelength range from 3 to over 100 microns and it is not possible to tune mirror layers or radiating antennae. Crucially, the laws of thermodynamics do not allow optical diodes, i.e. devices that allow photons to travel one way and not the other, so on a hot night when animals most need cooling this mechanism would be completely ineffective. In consequence thermal infrared emission is categorically ruled out. Infrared radiation generated by a chemical reaction then becomes the only known source. This is called chemiluminescence or bioluminescence when applied to animals. In the far-infrared such sources have not been reported and the physics seems highly improbable. We need to keep an open mind in view of the “weight of evidence” elsewhere.  

Follicle recent photosThis drawing comes from a paper by Lako, 2002 and is typical of drawings on-line and in text books. The anatomy can be understood during the growth phase when there is a need to draw in raw materials from the blood stream to build the hair. Yet it needs to be recorded that illustrations in books and on-line of follicle hair anatomy “look” optical perhaps by coincidence.  

The reason it “looks” optical is that different elements can be described in infrared engineering terms. The emitting antenna is the dermal papilla set in an optical cavity of dielectric mirror layers. The half-wavelength, low impedance, termination of the dermal papilla points into a waveguide-like structure called the medulla. The medulla is often reported to contain heavy elements which is precisely what fibre optic engineers use to raise the refractive index of fibre cores. Two more images below highlight the optical nature of the dermal papilla and the relationship with the blood supply. 


The dermal papilla is described as densely packed with capillaries and in the anagen phase is richly supplied with blood. It is often reported that the arteries and veins servicing hair follicles are closely entwined as in the counter-current circulation in ducks legs and this supports a cooling function for the hair follicles. The final act of growth is to lay down a layer of melanocytes to protect the dermal papilla from UV damage and this only has value if the dermal papilla has an important function after the completion of the hair growth.

The thermoregulation physics can be tested. 1 Watt of cooling is achieved by reducing the temperature of 0.05 milli-litres of blood by 5 degrees each second. 0.05 milli-litres is a reasonable total volume for the dermal papillae of a 10 gram bat although much more research will be needed to fill in the details.


Is there any evidence for a non-thermal infrared source?

We do not have any thermal cameras that are sensitive to 16 micron. There are spectrometers that could measure the emission spectrum in future and this should be the defining test. In the meantime an indirect method was attempted.

The radiation laws and the laws of thermodynamics prevent the harvesting of thermal infrared to generate usable energy. An animal radiating thermal infrared cannot warm up another object to radiate more brightly. A soprano pipistrelle bat recovering from a cat attack was flown and after a 20 minute exercise flight was filmed. Standard gauge printer paper was placed close to the fur (shown below) but not pressed into the fur so the only significant heating was from radiation and not conduction. The paper is opaque in the infrared but glowed brightly and constantly, due to radiation from the fur of the bat. This supports an infrared source outside of the spectral range of the camera (3-5 microns).


After flight a bat can heat up a paper sheet brighter than the bat itself

Leonardo have a camera called Condor that images in the 3.7-5 micron and 7.7-10 micron atmospheric windows on alternative frames. Images can be presented side-by-side or as a pseudo-colour image. A range of natural materials were studied including fur and eye emission of mammals, feather and eye emission from birds and various plants. The domestic cat bi-spectral images below are typical of all natural objects with virtually no difference. All the spectra were pure thermal with the perfect ratio.

Condor cats

At wavelengths below 10 microns infrared emission from animals is pure thermal

There was no evidence for any non-thermal sources in the range of the camera wavelength range. This reinforces the view that if there is an infrared source its  wavelength is much longer than 10 microns.

The naked mole rat (Heterocephalus glaber)

The only small mammal to lack fur is the naked mole rat of eastern Africa.


It is unable to control its temperature relying on behavioural methods such as retreating to deeper tunnels when it is hot and moving closer to the surface and huddling when it is cold. Its subterranean habits allow it to dispense with the need for thermo-regulation and it has no need for fur. It is not causative but an interesting coincidence.

Thermal images of moulting animals

Occasionally animals display a stripey thermal image as in this deer from 2017 Autumnwatch and presented with courtesy from the BBC.

It could be due to arteries close to the surface of the skin but usually animals have a very uniform thermal emission. The time of the year coincides with the autumn moult and it is possible that this animal is undergoing zonal moulting. It would be interesting to look at small rodents as they are known to moult in broad bands.


Hints of an infrared arms race

Producing strong infrared emission at a single wavelength is dangerous for small mammals, such as mice and shrews, because snakes can very easily tune-in with heat sensors. The complex radiating antennas in mouse fur are so sophisticated that there must have been constant predator pressure over a long period of time to drive the evolution of these structures. At present we only know of heat-sensing snakes but the geographical distributions of mice and these specialised snakes  do not appear to match up. When operating thermal imaging cameras in woodland it is remarkable how easy it is to locate small mammals in partial concealment. It is surprising no other predators have evolved thermal hot-spot detection. The eyes of owls have optical properties that match the best thermal imaging cameras but the vitreous humor absorbs infrared strongly. We are left with an unanswered question why mice and shrews have evolved complex hair types that attempt to suppress their thermal brightness.  


Last word on infrared antennas

Although the banding is very obvious there appear to be no references to the nature of these bands. The darker regions appear to have the higher refractive index. In larger animals such as rabbits, chinchillas, ringtailed possums and cats the hair appears to contain pores or air pockets and indeed the Hausman data base shows that the band spacing increases with the size of the mammal suggesting more porosity in the hair. More expert microscopy is needed and when the physical and optical structure of the banding is established it will be possible to employ fibre Bragg grating (FBG) analysis techniques to model the behaviour properly. 

In working with hair samples it was noticed that the oscillations in the hair patterns from the same locality are similar and since it is thought fur is replaced zonally in small mammals it implies the oscillations are due to a global bio-rhythm. 

The cuticle of the hair shaft is a strong feature in identification as it varies widely from species to species (Teerink, 1991). Generally there are wave-like petals that rotate and seem to roughly follow the banding pattern. In moles the spines provide an easy-to-measure cuticular feature that can be mapped against the banding pattern. The spines are not locked to the bands but roughly follow the pattern with an average ratio of 3.6. Any infrared adaptation in the cuticle pattern still needs to be proven but it may be a way of helping the radiation process.


Appendix I

List of species with banded hair structure

Species with banded hair anatomy – distantly related animals have almost identical hair anatomy


  1. Hausman L.A, “Structural characteristics of the hair of mammals”, The American Naturalist, Vol 54, No 635, (Nov-Dec, 1920), pp 496-523
  2. Winter Y, von Helversen O, “The energy cost of flight: do small bats fly more cheaply than birds?”, J Comp Physiol B, 1998 Mar;168(2):105-11
  3. Reichard J.D, Fellows S.R, Frank A.J, Kunz T.H,et al, “Thermoregulation during flight: body temperature and sensible heat transfer in free-ranging Brazilian free-tailed bats (Tadarida brasiliensis)”, Physiol Biochem Zool. 2010 Nov-Dec;83(6):885-97. doi: 10.1086/657253. Epub 2010 Oct 29.
  4. Sessions B.D, Nielson C.E, Sowa J.M et al, “Is bat hair morphology exceptional?”, Vespertilio 17: 171–183, 2014, ISSN 1213-6123).
  5. Reichard J.DKunz T.HKeller C, et al, “Vascular contrast enhanced micro-CT imaging of “radiators” in the Brazilian free-tailed bat (Tadarida brasiliensis)”, Anat Rec (Hoboken).2012 Apr;295(4):563-6. doi: 10.1002/ar.22423. Epub 2012 Jan 26
  6. Teerink J; Hair of west-European mammals.(Cambridge University Press); 1991, 224 .
  7. Lako M, Armstrong L, Cairns P.M, Harris S, Hole H and Jahoda C.A.B, “Hair follicle dermal cells repopulate the mouse haematopoietic system” published in Journal of Cell Science 2002 115: 3967-3974; doi: 10.1242/jcs.00060.
  8. Martin T, Marugán-Lobón J, Vullo R, Martín-Abad H, Luo Z, Buscalioni A. D, “A Cretaceous eutriconodont and integument evolution in early mammals”, Nature, 2015; 526 (7573): 380 DOI:10.1038/nature14905
  9. Vullo R, Girard V, Azar D, Neradeau D, “Mammalian hairs in early Cretaceous amber”, Naturwissenschaten, July 2010, Volume 97, Issue 7, pp 683-687

Last updated – 1st July 2018