Experiments with the drinking bird

J. Güémeza) and R. Valienteb)

Departamento de Fı́sica Aplicada, Universidad de Cantabria, E-39005 Santander, Spain
C. Fiolhaisc) and M. Fiolhaisd)
Departamento de Fı́sica and Centro de Fı́sica Computacional, Universidade de Coimbra,
P-3004-516 Coimbra, Portugal
共Received 17 January 2003; accepted 23 June 2003兲
We present a simple model of the dynamics of the drinking bird and relate its period to the
properties of its internal and external liquids. The effect of humidity on the motion is studied and it
is shown that there are two evaporation regimes. The results of the model are in agreement with
observations. © 2003 American Association of Physics Teachers.
关DOI: 10.1119/1.1603272兴

I. INTRODUCTION starting a new cycle.3 From the temperature difference be-

tween head and body, work can be produced so that the
The drinking bird 共dunking duck or dipping bird兲1 is not drinking bird is in fact a thermal engine.
only a toy, but also a demonstration apparatus on liquid– A liquid and its saturated vapor in thermal equilibrium at
vapor equilibrium and evaporation.2 Figure 1 shows a sche- constant temperature and pressure have the same chemical
matic and a photo of this intriguing thermodynamic potential. But when the vapor partial pressure is smaller than
device.3,4 The bird consists of two spherical glass bulbs con- its saturated pressure 共for water, when the humidity is less
nected by a glass tube that enters well inside the lower bulb. than 100%兲, the chemical potential of the liquid is higher
The bottom bulb 共the bird’s body, hereafter simply referred than that of the vapor and spontaneous evaporation occurs.
to as the body兲 is almost filled with a highly volatile liquid, This evaporation can be used to do work. Thus, at the most
normally methylene chloride (CH2 Cl2 ), whose normal boil- fundamental level, the ability to produce work lies in the
ing point is close to room temperature. There is no air inside difference between the chemical potentials of the external
the bird, but only this internal liquid in thermal equilibrium liquid and its nonsaturated vapor. However, the evaporation
with its vapor. from the head is not the only possible mechanism for the
The top bulb 共the bird’s head, hereafter simply called the drinking bird. A temperature gradient between the body and
head兲 is covered with a porous tissue. It has a small plastic the head may simply be obtained by heating the body, for
hat and a long beak which is covered with the same tissue as example, by illuminating a black painted body with a light
bulb or solar light.4 Our experiments with the ‘‘sunbird’’ 共a
the head. The bird can oscillate around a horizontal metallic
drinking bird that does not drink兲 are described in a compan-
bar attached to the tube at the middle. When the bird leans
ion paper.6
completely forward, it ‘‘drinks’’ water from a glass, although
Because we could not find a quantitative description of the
other external liquids may be used as well. We call this mo-
drinking bird, we present a model that relates its period, that
tion a dip.
is, the time between consecutive dips, to the properties of the
The drinking bird undergoes a cycle, which at first sight
internal and external liquids and to the bird’s dimensions. In
might seem to exhibit perpetual motion. At the beginning of
Table I we list the geometrical and physical data for the toy
the cycle, the bird is upright, with all the internal liquid in
used in our experiments.5 We performed various experiments
the lower sphere. The water on the head is in contact with its
with a drinking bird to verify our model. We also developed
vapor at a given 共room兲 temperature. If the vapor pressure is
a computer simulation of the bird’s dynamics.
smaller than its saturation 共or equilibrium兲 value, evapora- In Sec. II we present our model and compare it with ex-
tion occurs spontaneously. The evaporation cools the head periment, using various liquids as cooling agents. In Sec. III
outside, so that the CH2 Cl2 vapor inside also has to cool. The we analyze the influence of the humidity on the period when
vapor in the head condenses in very small drops, remaining water is used as the external liquid. In Sec. IV we report the
in equilibrium with the internal liquid as the temperature results of the numerical integration of the equations of mo-
decreases. The CH2 Cl2 vapor pressure inside the head be- tion. Our conclusions are given in Sec. V. In the Appendix
comes smaller than that in the body according to the we evaluate the drinking bird moment of inertia and torque,
Clausius–Clapeyron equation, and this pressure gradient which are needed in Sec. IV.
forces the internal liquid to rise up in the tube. As the liquid
rises, the center of mass of the system also rises, and the
momentum produced by the weight eventually forces the
II. PERIOD OF OSCILLATION
bird to tip forward and to dip its beak in the glass, keeping its
head wet. When the bird is almost horizontal, the lower end The cooling of the head during one period is directly re-
of the tube emerges above the internal liquid surface and lated to the evaporation of a certain mass of external liquid
some vapor passes from the body to the head 共see Fig. 1兲. outside the head, which we denote by ⌬m E (⌬m E ⬎0). The
While drinking, the bird remains horizontal for a short time. period temperature decrease inside the head during one
Then, part of the liquid drains back into the body and the
cycle, ⌬T, is related to the energy loss in one period:
bird returns to its upright position. As water evaporation con-
tinues from the head, the internal liquid comes up again, C⌬T⫽⫺⌬m E ⌬h E , 共1兲

1257 Am. J. Phys. 71 共12兲, December 2003 http://aapt.org/ajp © 2003 American Association of Physics Teachers 1257
 According to the Clausius–Clapeyron equation,7 the tem-
perature decrease ⌬T is related to the pressure decrease in-
side the head in one period, ⌬ P, by
⌬P
⌬T⫽ , 共2兲
B
with
⌬h I P I 共 T R 兲
B⫽ , 共3兲
RT R2
where ⌬h I is the molar vaporization enthalpy of CH2 Cl2 ,
P I (T R ) is the vapor pressure at room temperature, T R , and R
is the ideal gas constant. Equation 共2兲 requires that ⌬T be
sufficiently small.
The pressure difference between the head and the body 共a
negative quantity兲 is given by ⫺ ␳ I gz, where ␳ I is the inter-
nal liquid density and z is the height of the internal liquid
level in the tube with respect to the surface level in the body.
The total pressure decrease inside the head in one period 共a
negative quantity兲 is given by
⌬ P⫽⫺ ␳ I g⌬z, 共4兲
where ⌬z is the change of z in one period. If we substitute
Eq. 共4兲 in Eq. 共2兲, the temperature change in one period is
␳ I g⌬z
⌬T⫽⫺ . 共5兲
B
A similar relation holds between any small change of z and
the corresponding change of T.
If the evaporation rate of the external liquid 共a negative
quantity兲, ṁ E , is approximately constant, the period is
⌬m E C⌬T
␶ ⫽⫺ ⫽ , 共6兲
ṁ E ṁ E ⌬h E

Fig. 1. Drinking bird scheme and photo. For dimensions see Table I. where we have used Eq. 共1兲. The numerator C⌬T is charac-
teristic of a drinking bird operating with a given external
liquid. Although the evaporation rate does not remain con-
where C is an effective heat capacity of the head, which stant for many external liquids 共it changes due to the struc-
characterizes a given bird, and ⌬h E is the specific evapora- ture of the felt that covers the head兲, it may still be consid-
tion enthalpy of the external liquid. The minus sign in Eq. 共1兲 ered constant for small time intervals 共say, a few periods;
makes ⌬T a negative quantity. see, Fig. 3 and Table II兲.
To verify Eq. 共6兲, we measured periods and evaporation
rates not only for water—the usual liquid drunk by the
Table I. Geometrical and physical data for the drinking bird 共Ref. 5兲, with bird—but also for other liquids whose partial pressures in the
CH2 Cl2 as the internal liquid 共see Fig. 1兲. air are zero. When water is taken as external liquid, the air
humidity directly affects the bird’s dynamics: the period is
Tube length L⫽6.68 cm longer in humid days than in dry ones. For very high humidi-
Tube external diameter d E ⫽0.57 cm ties, the toy does not even work. This problem does not
Tube internal diameter d i ⫽0.40 cm
occur for other liquids.
Head external diameter d HE ⫽1.47 cm
Head internal diameter d Hi ⫽1.41 cm
We observed that the period, excluding the short time
Body external diameter d BE ⫽1.79 cm taken by the dip, is the time needed for the internal liquid to
Body internal diameter d Bi ⫽1.73 cm reach the top of the tube and fill approximately half of the
Height of upper spherical calotte 共empty兲 h u ⬇0.3 cm head: z max⬇L⫹dH H
i /2, where L is the tube length and d i is the
Angle of Fig. 1 ␸ ⬇20° head internal diameter 共see Table I兲. We also observed that
Hat mass m h ⫽0.81 g after a dip, the internal liquid does not completely return to
Beak mass m b ⬇0.2 g the body; the tube is half filled when a new cycle starts. In
Glass density ␳ g ⫽2.10 g/cm3
Fig. 2共a兲 we show the dependence of z with time obtained by
Glass specific heat c g ⫽0.837 J g⫺1 °C⫺1
CH2 Cl2 density ␳ I ⫽1.336 g/cm3
a simulation of the bird’s motion. 共The details of the simu-
CH2 Cl2 normal boiling point T I,b ⫽313.15 K lation will be given in Sec. IV.兲 In the steady regime, the
CH2 Cl2 vaporization enthalpy ⌬h I ⫽28094.50 J/mol minimum height of the internal liquid is z ⬘ ⫽L/2. In Fig.
2共b兲 we show the temperature change 共obtained in the same

1258 Am. J. Phys., Vol. 71, No. 12, December 2003 Güémez et al. 1258
 To measure evaporation rates, we placed the drinking bird
on a digital balance 共whose precision is ⫾0.001 g). After
pouring a few drops of different liquids on the head, we
measured the evolution of both the mass of external liquid
and the period.8 In these experiments the bird does not drink
when it dips, so that the external liquid that evaporates on the
head is not replaced. Figure 3, which presents results for
ethylic alcohol, allows us to confirm the inverse proportion-
ality between period and evaporation rate given by Eq. 共6兲.
Table II displays the evaporation enthalpies for various
external liquids as well as our measured quantities 共time,
evaporation rate, and period兲 using different external liquids.
We note that the evaporation rate is lower for water and,
consequently, the initial period is much larger. But it is in-
teresting that C⌬T⫽ṁ E ␶ ⌬h E is approximately constant for
a given drinking bird operating with a given external liquid.
Moreover, for liquids with lower evaporation enthalpies 共the
organic liquids in Table II, which are also less dense than
water兲, it turns out that the constant is practically the same
for all of them: C⌬T⬇⫺1 J. The partial pressure in the air is
zero for all organic liquids, and the lower the evaporation
enthalpies, the higher the evaporation rates. Because ⌬T is
known from Eq. 共5兲, we may extract the value of C from the
experimental value for C⌬T. On the other hand, for water
C⌬T⬇⫺1.26 J, so that C⫽4.75 J °C⫺1 , a value we will
use in Sec. IV.

III. INFLUENCE OF HUMIDITY

As mentioned, air humidity affects the bird’s period if wa-
ter is the external liquid. At room temperature, when the
vapor partial pressure, P E , is equal to the liquid–vapor equi-
Fig. 2. The evolution of 共a兲 the internal liquid height, z, and 共b兲 the tem-
perature inside the bird’s head, T. These results were obtained by the simu-
librium pressure at that temperature, P E (T R ), the water on
lation described in Sec. IV, with the parameters indicated in Fig. 5共a兲. the head no longer evaporates and the bird stops. If the hu-
midity is close to saturation, the evaporation rate is low and
the periods would be large. On the other hand, low partial
simulation兲, which is related to the change of z through Eq. pressure of water on dry days leads to a large evaporation
共5兲. After the dip the temperature inside the head rises only rate and a short period.
According to Fick’s law, the evaporation rate of water is
to T R ⫺ 兩 ⌬T ⬘ 兩 (⌬T ⬘ is the temperature change corresponding
given by10,11
to z ⬘ —see Fig. 2兲, and along the cycle it drops by an amount
兩 ⌬T 兩 共with 兩 ⌬T 兩 ⬇ 兩 ⌬T ⬘ 兩 ). ṁ E ⫽⫺ ␩ ⬘ 共 100⫺H 兲 , 共8兲
For our bird 共see Table I兲 we have B⫽(2.07⫾0.01) where ␩⬘⬎0 is a diffusion coefficient depending on the sub-
⫻103 Pa K⫺1 . This value is obtained by substituting ⌬h I stance and on the evaporation area and H⫽100P E / P E (T R ) is
⫽28 094.50 J/mol, R⫽8.3145 J K⫺1 mol⫺1 , and T R ⫽(22.5 the relative humidity.12 We substitute Eq. 共8兲, which is valid
⫾0.1)°C into Eq. 共3兲. The liquid–vapor pressure of the in- for normal convection, in Eq. 共6兲 and find that the period
ternal liquid P I (T R ) in Eq. 共3兲 is obtained from the solution becomes ␶ ⫽ ␬ ⬘ (100⫺H) ⫺1 , where ␬ ⬘ ⫽⌬m E / ␩ ⬘ . How-
of the Clausius–Clapeyron equation: ever, due to the bird’s motion, this relation is not exactly

冋 冉 冊册
observed. Our experimental results 共described below兲 sug-
⌬h I T I,b ⫺T R
P I 共 T R 兲 ⫽ P 0 exp ⫺ . 共7兲 gest a different power dependence of the period on the hu-
R T R T I,b midity, namely
For P 0 ⫽1.013⫻105 Pa 共normal atmospheric pressure兲 and ␶ 共 H 兲 ⫽ ␬ 共 100⫺H 兲 ⫺ ␤ , 共9兲
T I,b ⫽313.15 K 共the normal boiling point of the internal
where ␬ and ␤ are phenomenological parameters.
liquid,4 which is close to T R ), we obtain P I (T R )⫽(0.535 To study the evaporation rate we placed the drinking bird
⫾0.002)⫻105 Pa. For our bird we measured ⌬z⫽(4.2 in a closed chamber 共Fig. 4兲 and measured the periods for
⫾0.1) cm, so that from Eqs. 共4兲 and 共5兲, ⌬ P⫽⫺(5.5 various relative humidities. Figure 4 shows the logarithm of
⫾0.1)⫻102 Pa and ⌬T⫽⫺(0.266⫾0.005) °C. This value the period as a function of the logarithm of (100⫺H). The
of ⌬T is much smaller than T I,b ⫺T R and hence Eq. 共2兲 is experimental results clearly show two linear dependencies of
valid. Similarly, ⌬ P is much smaller then P I . These values ln ␶ on ln(100⫺H) in different ranges of the humidity. For
for ⌬T and ⌬ P are specific to a given drinking bird. A small ln(100⫺H)⭐3.2, that is, for up to ⬇75% relative humidity,
⌬T 共and, therefore, a small ⌬ P) facilitates the operation of the linear fit to the data leads to ␤⫽1.82 and ln ␬⫽9.44. We
the bird. interpret this power law behavior as due to forced air con-

1259 Am. J. Phys., Vol. 71, No. 12, December 2003 Güémez et al. 1259
 Fig. 3. 共a兲 The mass m E of ethylic al-
cohol on the bird’s head, 共b兲 the
evaporation rate ṁ E obtained from the
derivative of the fit to m E , and 共c兲 the
period ␶ as a function of the time. For
time intervals less than 1 min 共a few
periods, see Table II兲, the evaporation
rate is approximately constant. Similar
results were obtained for methylic al-
cohol, chloroform, and ethyl acetate;
n-hexane evaporates too quickly. For
water, the evaporation rate is practi-
cally constant in the time scale consid-
ered here.

vection. Above that humidity the linear fit to the data yields dm E
␤⫽1.24 and ln ␬⫽7.69. Normal evaporation 共␤⫽1兲 would ⫽⫺ ␩ 共 100⫺H 兲 ␤ , 共10兲
dt
only occur if the bird were at rest or slowly moving.
where ␩ ⫽⌬m E / ␬ . The values for ␬ and ␤ are found in Sec.
III. From the data of Table II, we find ⌬m E ⫽6.2⫻10⫺4 g,
IV. SIMULATION OF THE DYNAMICS
so that ␩ ⫽4.92⫻10⫺11 kg/s for H⬍75% and ␩ ⫽28.3
We now present a model for the dynamics of the bird ⫻10⫺11 kg/s for H⬎75%.
based on the previous description of water evaporation. We The evaporation of mass dm E leads to a temperature de-
use as the dynamical variable z(t), the column height of the crease dT given by an equation similar to Eq. 共1兲: dT
internal liquid at time t. From Eqs. 共6兲 and 共9兲 the mass of ⫽⫺dm E ⌬h E /C 共the heat capacity C was obtained in Sec.
the water, dm E , evaporated during the time interval dt is II兲. The temperature decrease dT in the time interval dt
given by leads, in turn, to a pressure gradient d P inside the head given

1260 Am. J. Phys., Vol. 71, No. 12, December 2003 Güémez et al. 1260
 Table II. The specific evaporation enthalpies ⌬h E , 共Ref. 9兲, the times, the evaporation rates, the periods, and
C⌬T⫽ṁ E ␶ ⌬h E for several external liquids. From top to bottom, the liquids are: chloroform, n-hexane 共its
evaporation is so quick that the bird stops after 1 min兲, ethyl acetate, ethylic alcohol, methylic alcohol, and
water at 50% humidity.

Liquid ⌬h E 共J/g兲 t 共min兲 ṁ E ⫻104 共g/s兲 ␶ 共s兲 C⌬T 共J兲

CH3 Cl 247.021 0 ⫺21.6 2⫾0.5 ⫺1.1⫾0.3

3 ⫺4.0 10⫾0.5 ⫺0.99⫾0.05
C6 H14 330.757 0 ⫺19.6 1.5⫾0.5 ⫺1.0⫾0.3
C4 H8 O2 368.438 0 ⫺5.67 5⫾0.5 ⫺1.04⫾0.13
6 ⫺2.33 12.5⫾0.5 ⫺1.07⫾0.04
C2 H6 O 841.547 0 ⫺4.40 3⫾0.5 ⫺1.1⫾0.2
10 ⫺0.90 12.5⫾0.5 ⫺0.95⫾0.04
CH4 O 1101.128 0 ⫺5.91 1.5⫾0.5 ⫺1.0⫾0.3
7 ⫺0.96 9.5⫾0.5 ⫺1.00⫾0.05
H2 O 2257.104 0 ⫺0.644 9⫾0.5 ⫺1.30⫾0.07
(H⫽50%) 45 ⫺0.541 10⫾0.5 ⫺1.22⫾0.06

by an equation similar to Eq. 共2兲, that is, d P⫽BdT. As a With no friction, the angular acceleration is ␣
consequence, during the interval dt, the liquid rises in the ⫽M (z)/I(z), where I(z) is the moment of inertia and M (z)
tube by dz⫽⫺d P/ ␳ I g. The evolution of z is given by z(t the torque with respect to the rotation axis; both depend on
⫹dt)⫽z(t)⫹dz with z(0)⫽0. the level z(t). The quantities I(z) and M (z) are evaluated in

Fig. 4. Top: setup for our experiments on the humidity

dependence of the drinking bird period. The bird is
placed inside a closed transparent chamber. A digital
chronometer and a digital hygrometer complete the
setup. The humidity, which started around 50%, in-
creased during the experiment. Bottom: the logarithm
of the period, ln ␶, vs ln(100⫺H). Two different evapo-
ration regimes were observed, one for high 共closed
circles兲 and the other for low 共closed squares兲 humidi-
ties. The change from one regime to the other occurs at
ln(100⫺H)⫽3.2. The least squares fit yields ln ␶
⫽9.44⫺1.82 ln(100⫺H) for low humidities and ln ␶
⫽7.69⫺1.24 ln(100⫺H) for high humidities. The open
circles are the results of the simulation presented in Sec.
IV.

1261 Am. J. Phys., Vol. 71, No. 12, December 2003 Güémez et al. 1261
the Appendix.
Because a realistic description of the bird’s motion re-
quires including frictional effects, we add to the torque a
term proportional to the angular velocity, that is, we define
M(z)⫽M (z)⫺b ␻ , where b is a phenomenological friction
coefficient and ␻ is the angular velocity. In our model we
take b⫽7.5⫻10⫺7 J s to account for the experimental damp-
ing, although the period is not very sensitive to this param-
eter. The angular acceleration ␣ (t) is then given by
M共 z 兲
␣⫽ , 共11兲
I共 z 兲
and the angular velocity and the angle between the tube and
the vertical direction are given by
␻ 共 t⫹⌬t 兲 ⫽ ␻ 共 t 兲 ⫹ ␣ ⌬t, 共12兲
␪ 共 t⫹⌬t 兲 ⫽ ␪ 共 t 兲 ⫹ ␻ 共 t⫹⌬t 兲 ⌬t, 共13兲
according to the Euler–Cromer algorithm with ⌬t a finite but
small time step.13 For the numerical integration of Eq. 共12兲
we used a time step ⌬t⫽0.001 s.
When ␪ ⫽90°, the internal liquid partly returns from the
head to the body, and the angular velocity and height of the
internal liquid are set to ␻ ⫽0 and z⫽L/2, respectively, be-
fore the new cycle starts. Accordingly, the temperature when
a new cycle starts is T R ⫺ 兩 ⌬T ⬘ 兩 , with 兩 ⌬T ⬘ 兩 ⫽0.211 °C cal-
culated from Eq. 共5兲 with ⌬z⫽L/2 共see Fig. 2兲. The value
⌬T⫽⫺0.283 °C found in our simulation is consistent with
that evaluated from Eq. 共5兲 in Sec. II. Also the maximum z
obtained in the simulation agrees with the experimental ob-
servation that the head was half-filled right before the dip.
Figure 5共a兲 shows the angle ␪ as a function of time, for an
initial quasi-vertical and motionless bird and humidity H Fig. 5. Simulation of the evolution of the angle for a drinking bird with
⫽65%. The period is 19.0 s, in agreement with the experi- m E ⫽0.4 g and m wb ⫽0.1 g. 共See the Appendix for the use of these param-
mental value 19.5⫾0.5 s obtained from the data fit in Fig. 4. eters in the moment of inertia and torque.兲 共a兲 Humidity H⫽65% ( ␶
Figure 5共b兲 shows the same quantity for 85% humidity. The ⫽19.0 s, the experimental value is 19.5 s兲 and 共b兲 H⫽85% ( ␶ ⫽74.5 s, the
experimental value is 76.1 s兲. In both cases the initial conditions are ␪
period is 74.5 s, close to the experimental value, 76.1 ⫽0.1 rad and ␻ ⫽0 rad/s. Note that the time scale is not the same in 共a兲 and
⫾0.5 s, also obtained from the data fit in Fig. 4. We con- 共b兲. The period is almost four times longer for the higher humidity.
clude that our model reproduces the data very well.
The drinking bird is sometimes incorrectly presented as a
perpetual motion machine and the graphs of Fig. 5 might
cause the same misleading impression that the motion con-
tinues forever. In our model we assume that the head is wet, cycles兲 or chemical nonequilibrium 共internal combustion en-
that is, there is external liquid in the head. For our drinking gines兲. But there are not many examples in which a chemical
bird the water reservoir was large enough to keep it moving potential difference produces work without a chemical
for days, so that the results of a model that takes into account reaction.14
the exhaustion of water in the reservoir should be presented Although the drinking bird is a well-known device, only
using a very different time scale. There are other effects that qualitative descriptions of its operation are available in the
are not accounted for by our simple model. The cooling literature. In this work we presented a quantitative model of
mechanism inside the head is admittedly naive, probably the
the bird’s motion. We studied the influence of different ex-
friction varies with angular velocity in a more complicated
ternal liquids on the bird’s behavior, confirming that the pe-
way than we have assumed, the evaporation rate changes
riod was smaller for more volatile liquids such as some or-
with the angular velocity, and the internal liquid cannot be
modeled as a solid. ganic liquids. For water as the external liquid, we observed
how the period depends on the relative humidity and found
two evaporation regimes: one at low humidities (H⬍75%)
V. SUMMARY and the other at high humidities (H⬎75%). Finally, we pre-
The drinking bird is a thermal engine because it can be sented a numerical integration of the equations of motion for
used to produce work from a temperature difference. The a bird that drinks water. Our results reproduce well the mea-
ability to produce work has its origin in the difference be- sured periods. The simulation is pedagogically interesting for
tween the chemical potentials of the external liquid and its checking the importance of different parameters and for see-
vapor. There are many thermal engines that operate directly ing the role of initial conditions 共horizontal or vertical initial
by a temperature difference 共for example, Carnot and Stirling position兲.

1262 Am. J. Phys., Vol. 71, No. 12, December 2003 Güémez et al. 1262
ACKNOWLEDGMENTS

We thank Rui Ferreira Marques for useful discussions.

M 共 z 兲 ⫽M 0 ⫺ 再冋 V I ⫺z ␲ 冉 冊册
di
2
2
␳ I ⫺m IH 共 z 兲 冎 L
2
g sin ␪

This work was partially supported by the Spanish Ministerio

de Ciencia y Tecnologı́a 共Grant No. BFM2000-1150兲 and by
the program ‘‘Portuguese-Spanish Integrated Actions.’’
⫺␳I z␲冋 冉 冊册 di
2
2
1
2
共 L⫺z 兲 g sin ␪

L
⫹m IH 共 z 兲 g sin ␪ , 共A6兲
2
APPENDIX: BIRD’S MOMENT OF INERTIA AND
TORQUE

When the internal liquid reaches the height z inside the with ␪ the angle between the vertical and the tube. In Eq.
internal tube, the moment of inertia I(z) with respect to the 共A6兲 the moment of the ‘‘fixed’’ parts is
rotation axis may be approximated by 共for notation, see
Table I and Fig. 1兲

I 共 z 兲 ⫽I 0 ⫹ 再冋 V I ⫺z ␲ 冉 冊册
di
2
2
␳ I ⫺m IH 共 z 兲 冎冉 冊
L
2
2 L
M 0 ⫽⫺m Bg g sin ␪ ⫹ 共 m H
2
L
g ⫹m h ⫹m E 兲 g sin ␪
2

⫹z ␲␳ I 冉 冊冋
di
2
2
1 2
12
L z
z ⫹ ⫺
2 2 冉 冊册 2
⫹m IH 共 z 兲 冉冊
L
2
2
,
a兲
⫹ 共 m b ⫹m wb 兲 L b g sin共 ␪ ⫹ ␸ 兲 .

Electronic mail: [email protected]

共A7兲

共A1兲 b兲
Electronic mail: [email protected]
c兲
Electronic mail: [email protected]
where we have used the parallel axis theorem. In Eq. 共A1兲, I 0 d兲
Electronic mail: [email protected]
is the moment of inertia for everything but the internal liq- 1
M. V. Sullivan, U. S. Patent 2,402,463 共1946兲; K. B. Kolb, ‘‘Drinking duck
uid, V I is the internal liquid volume, shutter,’’ Phys. Teach. 5, 342 共1967兲; L. M. Ng and Y. S. Ng, ‘‘The Ther-
modynamics of the Drinking Bird Toy,’’ Phys. Educ. 28, 320–324 共1993兲;
4
V I⫽ ␲
3
d Bi
2
冉 冊 3
1
冉
d Bi
⫺ ␲ h 2u 3 ⫺h u ,
3 2
冊 共A2兲
H. E. Stockman, ‘‘Dunking duck without liquid,’’ Am. J. Phys. 29, 335–
336 共1961兲; H. E. Stockman, ‘‘Secret of the dunking duck,’’ ibid. 29,
374 –375 共1961兲; E. Guarner and A. M. Sánchez, ‘‘The superconducting
bird: A didactical toy,’’ Phys. Teach. 30, 176 –177 共1992兲; C. F. Bohren,
and m IH (z) is the mass of the internal liquid in the head, Clouds in a Glass of Beer 共Wiley, New York, 1987兲.
2
which depends on z(t): J. S. Miller, ‘‘Physics of the dunking duck,’’ Am. J. Phys. 26, 42– 43

再 冉冊
共1958兲; J. L. Gaines, ‘‘Dunking duck,’’ ibid. 27, 189–190 共1959兲; D. L.
2
di Frank, ‘‘The drinking bird and the scientific method,’’ J. Chem. Educ. 50,
关 z 共 t 兲 ⫺L 兴 ␲ ␳I if z 共 t 兲 ⭓L 211 共1973兲.
m IH 共 z 兲 ⫽ 2 共A3兲 3
R. Plumb, ‘‘Physical chemistry of the drinking duck,’’ J. Chem. Educ. 50,
0 if z 共 t 兲 ⬍L. 213 共1973兲; K. B. Kolb, ‘‘Reciprocating engine,’’ Phys. Teach. 4, 121–122
共1966兲.
The second term of Eq. 共A1兲 refers to the liquid that remains 4
R. Mentzer, ‘‘The drinking bird. The little heat engine that could,’’ Phys.
in the body, the third term describes the liquid in the tube, Teach. 31, 126 –127 共1993兲.
and the last term the liquid in the head. 5
Drinking Bird Arora ©1997. Patent 100616. Tobar Ltd., Norfolk, UK.
6
The moment of inertia I 0 may be written as J. Güémez, R. Valiente, C. Fiolhais, and M. Fiolhais, ‘‘Experiments with a

冉冊 冉 冊
sunbird,’’ Am. J. Phys. 71, 1264 –1267 共2003兲, following paper.
L 2
2 d Bi 2
1 7
D. A. McQuarrie and J. D. Simon, Molecular Thermodynamics 共University
I 0 ⫽m Bg ⫹ m Bg ⫹ m L 2⫹共 m H
g ⫹m h ⫹m E 兲 Science Books, Sausalito, CA, 1999兲; D. V. Schroeder, Introduction to
2 3 2 12 t
Thermal Physics 共Addison–Wesley, Reading, 1999兲; M. W. Zemansky and

⫻
L
2冉冊 2
2 H dH
⫹ mg
3 2
i
冉 冊 2
⫹ 共 m b ⫹m wb 兲 L 2b . 共A4兲
8
R. H. Dittman, Heat and Thermodynamics 共McGraw–Hill, Singapore,
1997兲, 7th ed.
The time interval between consecutive dips which is experimentally mea-
sured should be corrected for the time elapsed while the bird is at rest in
The masses in Eq. 共A4兲 are m Bg ⫽(4 ␲␳ g /3) 关 (d EB /2) 3 the horizontal position 共approximately 1 s兲 in order to be compared with
⫺(d Bi /2) 3 兴 , the mass of glass in body; m H g ⫽(4 ␲␳ g /3) Eq. 共6兲.
⫻关 (d E /2) ⫺(d i /2) 兴 , the mass of the glass in the head;
H 3 H 3 9
K. Raznjevic, Tables et Diagrammes Thermodynamiques 共Editions Ey-
m t ⫽ ␲␳ g 关 (d E /2) 2 ⫺(d i /2) 2 兴 , the mass of the tube; m h , the 10
rolles, Paris, 1970兲.
A. J. Abraham and C. P. Bean, ‘‘A simple method for measurement of the
mass of the hat; m E , the mass of the water on the head diffusion of vapors,’’ Am. J. Phys. 57, 325–329 共1989兲.
except for the beak; m b , the mass of the beak; and m wb , the 11
J. I. Zubizarreta and G. Pinto, ‘‘An ancient method for cooling water
mass of the water in the beak. In the last term of Eq. 共A4兲, explained by mass and heat transfer,’’ Chem. Eng. Educ. 29, 96 –99
共1995兲.
L 2b ⫽l 2b 共 1⫹tg2 ␸ 兲 , 共A5兲 12
For water at room temperature, the value P E (T R )⫽3.26⫻103 Pa is found
from the Clausius–Clapeyron equation, inserting T E,b ⫽373.15 K and ⌬h E
with l b ⬇(d EH ⫹L)/2, and ␸ the angle shown in Fig. 1. The 共given in Table II兲 in Eq. 共7兲 instead of T I,b and ⌬h I , respectively.
parallel axis theorem was used and, for the glass tube, the 13
H. Gould and J. Tobochnik, An Introduction to Computer Simulation
rotation axis was assumed to be at the tube’s center. Methods 共Addison–Wesley, Reading, 1996兲, 2nd ed.
The torque with respect to the fixed rotation axis is given 14
C. Bachhuber, ‘‘Energy from the evaporation of water,’’ Am. J. Phys. 51,
by 259–264 共1983兲.

1263 Am. J. Phys., Vol. 71, No. 12, December 2003 Güémez et al. 1263