How Cats Lap: Water Uptake
by Felis catus
Pedro M. Reis/'2* Sunghwan Jung,3* Jeffrey M. Aristoff,4* Roman Stockest
Animals have developed a range of drinking strategies depending on physiological and
environmental constraints. Vertebrates with incomplete cheeks use their tongue to drink; the
most common example is the lapping of cats and dogs. We show that the domestic cat {Felis catus) laps
by a subtle mechanism based on water adhesion to the dorsal side of the tongue. A combined
experimental and theoretical analysis reveals that Felis catus exploits fluid inertia to defeat gravity
and pull liquid into the mouth. This competition between inertia and gravity sets the lapping
frequency and yields a prediction for the dependence of frequency on animal mass. Measurements
of lapping frequency across the family Felidae support this prediction, which suggests that
the lapping mechanism is conserved among felines.
animals have evolved diverse
Terrestrial means to acquire water, including absorp- tion through the skin (1) or extraction of
moisture from food (2), but most rely on drinking
(3-12). Drinking presents a challenge to land vertebrates, because fresh water occurs mainly as
horizontal liquid surfaces, such as puddles, ponds,
lakes, or streams, and animals must displace water
upward against gravity to drink it. Crucial in the
drinking process is die role of the tongue, which in
vertebrates is used in two distinctly different ways.
Vertebrates with complete cheeks, such as pigs,
sheep, and horses, use suction to draw liquid upward and use their tongue to transport it intraorally
(13, 14). In contrast, vertebrates with incomplete
cheeks, including most carnivores, are unable (after weaning) to seal their mouth cavity to generate
suction and must rely on their tongue to move
water into the mouth (13). When the tongue
sweeps the bottom of a shallow puddle, the process
is called licking (4). When the puddle is deeper
than the tongue excursion into the liquid, it is called
lapping (15). Here, we report on the lapping mechanism of the domestic cat (Felis catus).
Almost everyone has observed a domestic cat
lap milk or water. Yet casual observation hardly
captures the elegance and complexity of this act, as
the tongue's motion is too fest to be resolved by the
naked eye. We used high-speed imaging to capture
the motion of both the tongue and liquid during
lapping [Fig. 1 and movie SI (16)]. With the cat's
fece oriented downward, the tongue extends from
the jaw (Fig. 1A) and its tip curls sharply caudally
(Fig. IB). At the lowest position of the tongue's tip,
its dorsal side rests on the liquid surface, without
piercing it (Fig. IB). When the cat Ms the tongue,
liquid adhering to the dorsal side of the tip is drawn
upward, forming a column (Fig. 1C). This liquid
column is further extended by the tongue's upward
motion (Fig. ID), thinning in the process (Fig. IE),
and is finally partially captured upon jaw closure
(Fig. 1, E and F). Inside the mouth, cavities
between the palate's rugae and the tongue act as a
nonreturn device and trap liquid until it is ingested
every 3 to 17 cycles (75).
This sequence of events reveals two important aspects of how F. catus laps. First, lapping
does not rely on "scooping" water into the mouth
as in dogs {Canis lupus familiaris). Although the
dog's tongue also curls caudally in a cuplike
shape, it penetrates the liquid surface and scoops
up the water that fills its ventral side. In contrast,
the cat's tongue does not dip into the liquid (Fig.
IB) and the cavity on its ventral side remains
empty (Fig. 1, B and C), as also recognized in a
1940 Oscar-winning short film (17). Second,
only the tip of the tongue is used for lapping.
The tip is free of filiform papillae (18) (Fig. IG),
the semirigid hairlike structures that give a cat's
tongue its characteristic roughness. Thus, papillae appear to have no function in lapping.
The movies allowed us to quantify the lapping
kinematics. The position of the tip of the tongue
was tracked over one lapping event (Fig. 2A) and
averaged over 1 1 cycles (Fig. 2B). The tongue's
vertical position during upward motion is well
described by an error-function profile, Z(t) =
ViH{' + Grf(UMAxtVñ/H)}, where H= 3.0 cm
(Fig. 2B, red) is the total vertical displacement,
Umax is the maximum speed, and t is the time
measured with respect to Z(t=O) = H/2, the halfheight As a result, the vertical velocity has an approximately Gaussian profile (Fig. 2B, blue): The
tongue accelerates as it leaves the water surface,
attains a remarkable maximum speed of Umax =
78 ± 2 cm s"1, then decelerates as it enters the
mouth. Recordings of 10 adult individuals (16)
yielded a lapping frequency /= 3.5 ± 0.4 s"1 and
an ingested volume per lap V= 0.14 ± 0.04 ml.
To help understand the mechanism of lapping,
we performed physical experiments in which a
glass disk of radius R (representing the tongue's
tip), initially placed on a water surface, was pulled
vertically upward (Fig. 3). The time series of the
disk position was set by a computer-controlled
stage and programmed to have the observed errorfunction profile (Fig. 2B), where we could control
the lapping height H and speed Umax- The hydrophilic nature of glass (static contact angle =
14°) mimicked the wetness of the tongue. Highspeed imaging revealed the formation and extension of a water column, as in lapping (Fig. 3, A
to G and movies S2 and S3) (16). The column's
ascent was eventually interrupted by pinch-off,
leaving a pendantlike drop on the lower surface
ofthe disk (Fig. 3,FtoH).
Estimation of the forces involved suggests
that the fluid dynamics of lapping are governed
by inertia and gravity, whereas viscous and capillary forces are negligible (16). Inerbai entrainment draws liquid upward into a column, while
gravity acts to collapse it. Ultimately, gravity prevails and the column pinches off. Dimensional
analysis reveals that two dimensionless parameters control lapping: the Froude number, Fr =
^MAx/(g#)1/2> measuring the relative importance
of inertia and gravity (g is the gravitational acceleration), and the aspect ratio, HIR. Experiments
were therefore conducted over a range of Fr and
HIR values (16), for a fixed lapping height, H -
3 cm, determined from observations (Fig. 2B).
The latter is possibly dictated by biological constraints, such as the need to keep whiskers dry to
maintain their sensory performance (19) or to
maximize peripheral vision while drinking.
To test the proposition that the column dynamics are set by a competition between inertia
and gravity, we compared the height ofthe disk
(Z) at pinch-off, ZP, with that predicted from our
scaling analysis. We find (16) that ZP/H ~ Fr*
for Fr* < 1 andZp///- 1 otherwise. Here, Fr* =
(R/H)¥t2/3 is the ratio ofthe time scale for the gravitational collapse ofthe column, tP = (R/UmaxW*2'3,
and the time scale for the upward motion ofthe
disk, H/Umax- Experiments confirmed the existence of two regimes (Fig. 4A). For small disks
(R = 2.5 and 5 mm), ZPIH increased linearly with
Fr*, whereas large disks (R = 10 and 12.7 mm)
reached the final height before pinch-off (ZPIH=
1). Theory also successfully predicts that pinchoff occurs close to the disk (Fig. 3).
Consequently, the balance between inertia
and gravity dictates the lapping frequency,/ by
controlling the time of pinch-off. To maximize
ingested volume, which is assumed to be proportional to the column's volume V, the lapping time,
Iff, should match the pinch-off time, tP, because
fester lapping fails to maximize inertial entrainment,
whereas slower lapping results in belated mouth
closure that misses most ofthe column. This predicts/ ~ (gH)m/R or, because/ ~ Umax/H, that
Fr * is of order one. For domestic cats, we find that
Fr* is indeed of order unity (0.4), using the experimentally measured values Umax =78 cm s"1
and H= 3 cm (Fig. 2B) and a tongue size oïR ~
5 mm (Fig. IG).
The growth dynamics ofthe column's volume
(Fig. 4B) (16) provide further evidence that the
lapping frequency is set by the interplay between
inertia and gravity. When the disk is close to the
bath (Z « //), the column is cylindrical and V
increases as V/R3 = kZ/R (Fig. 4B inset). Gravitydriven drainage then reduces the rate of volume
increase, and Fis at a maximum when gravity and
inertia balance (Fig. 4B). A scaling argument predicts a maximum volume when the disk height
reaches ZMaxVoi/#~ Fr* (16), in excellent agreement with observations (Fig. 4A). This also supports our assumption that Fis maximum close to
the time of pinch-off. In fact, F always peaks
shortly before pinch-off (10 to 70 ms) (Fig. 4B),
which suggests that mouth closure should also
occur just before pinch-off to maximize captured
volume. Observations of domestic cats showed
that mouth closure indeed typically preceded
pinch-off (movie Si) (16).
Fig. 1. The lapping process. (A to F)
Snapshots showing the movement of
the tongue of F. catus and the dynamics of the liquid column during
a lapping cycle. Lapping occurs by
fluid adhesion to the dorsal part of
the tongue's tip and by lifting a
liquid column through the tongue's
upward motion, before jaw closure.
Time elapsed from the first frame is
given in the top left corner of each
frame. (G) Photograph of the dorsal
side of the tongue of F catus, acquired under anesthesia (16). Only
the smooth tip is used in lapping.
Fig. 2. The lapping kinematics. (A) Vertical position
of the tip of the tongue of F. catus over 11 lapping
cycles. (B) Vertical position (red circles) and velocity
(blue squares) of the tongue's tip during a typical
lapping cycle, obtained by averaging 11 cycles. The
dashed line (red) represents a fitted error-function
profile for vertical position. The dotted line (blue) is
its temporal derivative and represents upward speed,
which yielded (7,^ = 78 ± 2 cm s"1.
Fig. 3. Physical model of lapping. A disk of radius R is
driven vertically upward, such that the elevation of the disk,
Z(t), from the liquid interface (dashed line, Z = 0), has an
error-function profile (16). Lowercase z denotes the distance
of a generic liquid layer from the disk. The sequence illustrates the formation (A to E) and pinch-off (F) of the liquid
column. After pinch-off, part of the column collapses into the
bath (G), which leaves a pendantlike drop attached to the
disk (H). The final height of the disk is Z = H. Time is measured relative to the moment the disk reaches height Z = H/2 =
1.5 cm. This sequence corresponds to R = 12.7 mm, H = 3 cm,
Umax = 50 cm s"1 (Fr = 1.42; Fr* = 0.60).
Fig. 4. Liquid column dynamics in the physical model and observed lapping
frequency of felines. (A) Position of the disk at pinch-off, ZP, versus Fr*, for
four disk radii R (solid squares), and position of the disk when the column
achieves maximum volume, Z^xvoi/ versus Fr*, for R = 5 mm (open squares).
(B) Time course of the liquid column volume, V, for different speeds (/maxBlack open squares indicate the pinch-off time relative to the moment when
the disk started moving (16). The maximum column volumes (0.12 to 0.26 ml)
are of the same order as the volume per lap ingested by Felis cotus (0.14 ±
0.04 ml). The inset shows the volume as a function of relative disk height
(solid circles) and the prediction from a scaling argument for low heights
(black line, Z/H « 1). At greater heights, drainage reduces the volume and
then pinch-off occurs. (0 Lapping frequency as a function of animal mass for
eight species of felines. Solid symbols represent data acquired at the Zoo New
England; open symbols represent data from YouTube videos (16). The line of
best fit (dashed red) has slope -0.181 ± 0.024. The solid black line has
slope -1/6.
The balance of inertia and gravity yields a
prediction for the lapping frequency of other
felines. Assuming isometry within the Felidae
family (i.e., that lapping height H scales linearly
with tongue width R and animal mass M scales
as R3), the finding that Fr * is of order one translates to the prediction/- R ~m ~ M ~m. Isometry or marginally positive allomety among the
Felidae has been demonstrated for skull (20, 21)
and limb bones (22). Although variability by
function can lead to departures from isometry in
interspecific scalings (23), reported variations within the Felidae (23, 24) only minimally affect the
predicted scaling/- M~m. We tested this -1/6
power-law dependence by measuring the lapping
frequency for eight species of felines, from videos acquired at the Zoo New England or available on YouTube (16). The lapping frequency was
observed to decrease with animal mass as/=
4.6 M^181 ± 0024 (/in s"1, Min kg) (Fig. 4C),
close to the predicted AT176. This close agreement suggests that the domestic cat's inertia- and
gravity-controlled lapping mechanism is conserved
among felines.
The lapping of E catus is part of a wider
class of problems in biology involving gravity and
inertia, sometimes referred to as Froude mechanisms. For example, the water-running ability of
the Basilisk lizard depends on the gravity-driven
collapse of the air cavity it creates upon slapping
the water surface with its feet. The depth to which
the lizard's leg penetrates the surface depends
on the Froude number, which in turn prescribes
the minimum slapping frequency (25). The Froude
number is also relevant to swimming, for example, setting the maximum practical swimming
speed in ducks (26), and to terrestrial legged locomotion. In this respect, it is interesting to note that
the transition from trot to gallop obeys nearly the
same scaling of frequency with mass as lapping,
/= 4.5 M ~~°14 (/in s~' Min kg) (27).
The subtle use of the tongue in the drinking
process of F catus is remarkable, given the
tongue's lack of skeletal support (28). Complex
movement in the absence of rigid components is
a common feature of muscular hydrostats, which
in addition to tongues include elephant trunks and
octopus arms (28, 29). The functional diversity
and high compliance of these structures continue
to inspire the design of soft robots (29), and a
fundamental understanding of their functionality
can lead to new design concepts and is essential
to inform biomechanical models (29, 30).
References and Notes
1. P. ]. Bentley, T. Yorio, ;. Exp. Bioi 79, 41 (1979).
2. D. Attenborough, The Living Planet: A Portrait of the Earth
(Collins and British Broadcasting Corporation, London, 1984).
3. ]. Heidweiller, ]. A. van Loon, G. A. Zweers,
Zoomorphology 111, 141 (1992).
4. ]. A. W. M. Weijnen, Neurosa. Biobehav. Rev. 22,
751 (1998).
5. W. C. Dilger, in Roots of Behavior, E. L. Bliss, Ed.
(Harper, New York, 1962), pp. 35-47.
6. M. Prakash, D. Quéré, ]. W. M. Bush, Science 320,
931 (2008).
7. S. W. S. Gussekloo, R. G. Bout, ;. Exp. Bioi 208, 3395
(2005).
8. T. L. Daniel, ]. G. Kingsolver, E. Meyhò'fer, Oecologia 79,
66 (1989).
9. ]. G. Kingsolver, T. L Daniel Oecologia 60, 214 (1983).
10. D. Cundall, ;. Exp. Bioi. 203, 2171 (2000).
11. ]. Heidweiller, G. A. Zweers, Condor 92, 1 (1990).
12. ]. G. M. Kooloos, G. A. Zweers,;. Morphoi 199, 327 (1989).
13. A. ]. Thexton, A. W. Crompton, R. Z. German,;. Exp. Zool.
280, 327 (1998).
14. K. M. Hiiemae, A. W. Crompton, in Functional
Vertebrate Morphology, M. Hildebrand, D. Bramble,
K. Liem, D. B. Wake, Eds. (Belknap of Harvard Univ.
Press, Cambridge, MA, 1985), pp. 262-290.
15. A. ]. Thexton, ]. D. McGarrick, Arch. Oral Bioi. 33, 331 (1988).
16. Materials and methods are available as supporting
material on Science Online.
17. Quicker 7? a Wink (Metro-Goldwyn-Mayer Studios, Los Angeles,
CA, 1940); www.youtube.com/watchTv^UZwFrGzQGw.
18. K. Ojima, F. Mitsuhashi, M. Nasu, Y. Suzuki, Ann. Anat.
182, 47 (2000).
19. A. S. Ahi, Vet. Res. Commun. 10, 245 (1986).
20. P. Christiansen, ]. S. Adolfssen, ;. Zool. (London) 266,
133 (2005).
21. A. P. Russell, H. N. Bryant, G. L Powell, R. Laroiya,
;. Zool. (London) 236, 161 (1995).
22. ]. Meachen-Samuels, B. Van Valkenburgh, ;. Morphoi.
270, 729 (2009).
23. M. Doube, A. Wiktorowicz-Conroy, P. Christiansen,
]. R. Hutchinson, S. Shefelbine, PLoSONEA, e4742 (2009).
24. L M. Day, B. C. Jayne, ;. Exp. Bioi. 210, 642 (2007).
25. ]. W. Glasheen, T. A. Mcmahon, ;. Exp. Bioi 199,
2611 (1996).
26. H. D. Prange, K. Schmidt-Nielsen, ;. Exp. Bwl. 53, 763
(1970).
27. N. C. Heglund, C. R. Taylor, T. A. McMahon, Science
186, 1112 (1974).
28. K. K. Smith, W. M. Kier, Am. Sci. 77, 28 (1989).
29. D. Trivedi, C. D. Rahn, W. M. Kier, I. D. Walker,
Appi Bionics Biomech. 5, 99 (2008).
30. H. ]. Chiei, P. Crago, ]. M. Mansour, K. Hathi,
Bioi Cybern. 67, 403 (1992).
31. All research with animals complied with and was approved
by MIT's Animal Rights Committee. We thank ]. Piazza
and the staff at the Zoo New England for help with filming
felines; M. Rock, A. Kennedy, and the Massachusetts Society
for the Prevention of Cruelty to Animals (MSPCA) for help
with filming domestic cats; G. McKinley for use of his
FiSER stage; A. Crompton, A. Thexton, W. Kier, ]. Bales,
). Bush, B. Andreotti, ]. Bico, D. Velia, and A. Boudaoud
for discussions and comments; and A. Jones, E. Samaha,
R. Tomas, and W. jarjoui for technical assistance. S.].,
P.M.R., and R.S. designed research; S.]., P.M.R., R.S., and
J.M.A. performed experiments and analyzed data; S.].,
].MA, and R.S. developed scalings and theory; P.M.R.,
R.S., S.J., and J.M.A. wrote the paper.