Modeling a falling slinky

Abstract

A slinky is an example of a tension spring: in an unstretched state a slinky
is collapsed, with turns touching, and a finite tension is required to separate
the turns from this state. If a slinky is suspended from its top and stretched
under gravity and then released, the bottom of the slinky does not begin to
fall until the top section of the slinky, which collapses turn by turn from the
top, collides with the bottom. The total collapse time t_c (typically ~0.3 s
for real slinkies) corresponds to the time required for a wave front to
propagate down the slinky to communicate the release of the top end. We present
a modification to an existing model for a falling tension spring (Calkin 1993)
and apply it to data from filmed drops of two real slinkies. The modification
of the model is the inclusion of a finite time for collapse of the turns of the
slinky behind the collapse front propagating down the slinky during the fall.
The new finite-collapse time model achieves a good qualitative fit to the
observed positions of the top of the real slinkies during the measured drops.
The spring constant k for each slinky is taken to be a free parameter in the
model. The best-fit model values for k for each slinky are approximately
consistent with values obtained from measured periods of oscillation of the
slinkies.

Full text

arXiv:1208.4629v1 [physics.pop-ph] 22 Aug 2012
The following article has been accepted by the American Journal of Physics. After it is published, it will b e found at http://scitation.aip.org/ajp.
Modeling a falling slinky
R. C. Cross∗and M. S. Wheatland†
School of Physics, University of Sydney, NSW 2006, Australia
Abstract
A slinky is an example of a tension spring: in an unstretched state a slinky is collapsed, with
turns touching, and a finite tension is required to separate the turns from this state. If a slinky
is suspended from its top and stretched under gravity and then released, the bottom of the slinky
does not begin to fall until the top section of the slinky, which collapses turn by turn from the top,
collides with the bottom. The total collapse time tc(typically ∼0.3 s for real slinkies) corresponds
to the time required for a wave front to propagate down the slinky to communicate the release of
the top end. We present a modification to an existing model for a falling tension spring9and apply
it to data from filmed drops of two real slinkies. The modification of the model is the inclusion
of a finite time for collapse of the turns of the slinky behind the collapse front propagating down
the slinky during the fall. The new finite-collapse time model achieves a good qualitative fit to the
observed positions of the top of the real slinkies during the measured drops. The spring constant
kfor each slinky is taken to be a free parameter in the model. The best-fit model values for kfor
each slinky are approximately consistent with values obtained from measured periods of oscillation
of the slinkies.
1

I. INTRODUCTION
The physics of slinkies has attracted attention since their invention in 1943. Topics of
studies include the hanging configuration of the slinky,1,2 the ability of a slinky to walk down
stairs,3the modes of oscillation of a vertically suspended slinky,4,5 the dispersion of waves
propagating along slinkies,6–8 and the behavior of a vertically stretched slinky when it is
dropped.9–11
Slinkies are examples of tension springs, i.e. springs which may be under tension according
to Hooke’s law, but not compression. Unstretched slinkies have a length ℓ1at which the turns
are in contact, and a finite tension f1is needed to separate the turns from this state. They
collapse to this state if not stretched by an external force. This may be contrasted with a
“compression spring,” which can be under tension or compression according to Hooke’s law.
Compression springs have an unstretched length ℓ0at which the turns are not in contact,
and the tension is zero. They may be compressed to a length ℓ1at which the turns are in
contact, and obey Hooke’s law during this compression. Fig. 1 shows tension versus length
diagrams for uniform extensions of the two types of springs.
The vertically falling slinky, mentioned above, exhibits interesting dynamics which de-
pend on the slinky being a tension spring.9A falling compression spring exhibits periodic
compressions and rarefactions, as longitudinal waves propagate along the spring length. A
falling tension spring collapses to the length ℓ1during a fall, assuming the spring is released
in an initially stretched state (with length > ℓ1).
If a slinky is hanging vertically under gravity from its top (at rest) and then released,
the bottom of the slinky does not start to move downwards until the collapsing top section
collides with the bottom. Figure 2 illustrates this peculiar effect for a plastic rainbow-
colored slinky; this figure shows a succession of frames extracted from a high-speed video of
the fall of the slinky.12 The continued suspension of the bottom of the slinky after release is
somewhat counter-intuitive and very intriguing—a recent YouTube video showing the effect
with a falling slinky has received more than 800,000 views.13 The physical explanation is
straightforward: the collapse of tension in the slinky occurs from the top down, and a finite
time is required for a wave front to propagate down the slinky communicating the release
of the top.
The basic wave physics behind this behavior follows from the equation of motion for a
2

falling (or suspended) compression spring14
m∂2x
∂t2=k∂2x
∂ξ2+mg, (1)
which applies to a tension spring when the turns are separated. In this equation x(ξ , t) is
the vertical location of a point along the spring at time t,mis the total spring mass, and kis
the spring constant. The (dimensionless) coordinate ξdescribes the mass distribution along
the spring, such that dm =m dξ is the increment in mass associated with an increment
in ξ, with 0 ≤ξ≤1. Thus, for a spring with Nturns, the end of turn icorresponds to
ξi=i/N and is located at position xi=x(ξi, t) = x(i/N, t) at time t. Equation (1) is an
inhomogenous wave equation. If the spring is falling under gravity, then in a coordinate
system falling with the center of mass of the spring (x′=x−1
2gt2), the equation of motion
is the usual wave equation
m∂2x′
∂t2=k∂2x′
∂ξ2.(2)
Equation (2) implies that waves in the mass distribution (turn spacing) propagate along the
length of the spring in a characteristic time tp=pm/k. This accounts for the periodic
rarefactions and compressions of a compression spring during a fall, and for the propagation
of the wave front ahead of the collapsing turns in a falling tension spring.
In this paper we present a new detailed model for the fall of a slinky, which improves
on past models by taking into account the finite time for collapse of the turns of the slinky
behind the wave front. In Sec. II we explain the need for this refinement in the modeling,
and we present the details of the new model in Sec. III. The new model is compared with
the behavior of two real falling slinkies in Sec. IV, and we discuss our conclusions in Sec. V.
II. THE COLLAPSE OF THE TURNS AT THE TOP OF THE SLINKY
A detailed description of the dynamical behavior of a falling tension spring requires so-
lution of the equations of motion for mass elements along the slinky that are subject to
gravity and local spring forces, taking into account the departure from Hooke’s law encoun-
tered when slinky turns come into contact. Because of the complexity of this modeling, past
efforts involve specific approximations.9–11
3

For a mass element m∆ξat a location ξion the slinky, the equation of motion is14
m∆ξ∂2x
∂t2


ξi,t
=f(ξi+ ∆ξ, t)−f(ξi, t) + m∆ξg
= ∆ξ∂f
∂ξ 


ξi,t
+m∆ξg,
(3)
where f(ξ, t) is the tension force at ξ. Equation (3) applies to all points except the top and
bottom of the slinky, where the tension is one sided. When slinky turns are separated at a
point along the slinky, the tension is given by Hooke’s law in the form9
f(ξ, t) = k∂x
∂ξ −ℓ0,(4)
where ∂x/∂ξ describes the local extension of the slinky, and ℓ0corresponds to a slinky length
at which the tension would be zero, assuming a Hooke’s law relation for all values of the local
extension. Substituting Eq. (4) into Eq. (3) leads to the inhomogenous wave equation (1).
For a tension spring the length ℓ0cannot be reached because there is a minimum length
∂x
∂ξ =ℓ1> ℓ0(5)
that corresponds to the spring coils being in contact with each other. At this point, the
minimum tension
f1=k(ℓ1−ℓ0) (6)
is achieved and the tension is replaced by a large (infinite) compression force as the collapsed
turns resist further contraction of the slinky (see Fig. 1). This non-Hooke’s law behavior is
met when turns collapse at the top of the falling slinky and the description of this process
complicates the modeling.
A simpler, approximate description of the dynamical collapse of the top of the slinky is
to assume a functional form for the position-mass distribution x(ξ, t) during the collapse,
and then impose conservation of momentum to ensure physical time evolution. Calkin9
introduced this semi-analytic approach and specifically assumed a distribution corresponding
to slinky turns collapsing instantly behind a downward propagating wave front located a
mass fraction ξc=ξc(t) along the slinky at time tafter the release. The turns of the slinky at
the front instantly assume a configuration with a minimum tension, so that Eq. (5) applies
for all points behind the front at a given time
f(ξ < ξc, t) = f1.(7)
4

For points ahead of the front (ξ > ξc) the tension is the same as in the hanging slinky.
The location of the front at time tis obtained by requiring that the total momentum of the
collapsing slinky matches the impulse imparted by gravity up to that time. (The modeling
is presented in detail in Sec. III B.) The Calkin model has also been derived in solving the
inhomogenous wave equation (1) subject to the boundary condition given by Eq. (7).10,11
With real falling slinkies the collapse of turns behind the front takes a finite time. Figure 3
illustrates the process of collapse of a real slinky using data extracted from a high-speed video
of a fall (this data is discussed in more detail in Sec. IV A.) The upper panel of Fig. 3 shows
the position of the top (blue circles), of turn eight (black + symbols), and of turn ten (red ×
symbols) versus time, for the first 0.2 s of the fall. The vertical position is shown as negative
in the downward direction measured from the initial position of the top [which corresponds
to −xin terms of the notation of Eq. (2)]. The upper panel shows that turns 8 and 10 of
the slinky remain at rest until the top has fallen some distance, and then turn eight begins
to fall before turn ten. The lower panel shows the spacing of turns eight and ten versus
time. The two turns change from the initial stretched configuration to the final collapsed
configuration in ∼0.1 s.
This paper presents a method of solution of Eq. (1) which adopts the approximate ap-
proach of Calkin, but includes a finite time for collapse of the turns. We assume a linear
profile for the decay in tension behind the wave front propagating down the slinky, which
provides a more realistic description of the slinky collapse.
III. MODELING THE FALL OF A SLINKY
In Secs. III A and III B we reiterate the Calkin9model for a hanging slinky as a tension
spring, and for the fall of the tension spring. In Sec. III C the new model for the fall of the
slinky is presented.
A. The hanging slinky
For hanging slinkies it is generally observed that the top section of the slinky has stretched
turns, and a small part at the bottom has collapsed turns.1Assuming mass fractions ξ1and
1−ξ1of the slinky with stretched and collapsed turns, respectively, the number of collapsed
5

turns Ncis related to the total number of turns Nby
Nc= (1 −ξ1)N. (8)
A hanging slinky such that the turns just touch at the bottom would have ξ1= 1.
The position X=X(ξ) of points along the stretched part of the stationary hanging slinky
is described by setting ∂2x/∂t2= 0 in Eq. (1) and integrating from ξ= 0 to ξ=ξ1with the
boundary conditions
X(ξ= 0) = 0 and ∂X
∂ξ 


ξ=ξ1
=ℓ1,(9)
corresponding to the fixed location of the top of the slinky, and the spacing of collapsed turns
at the bottom of the slinky, respectively. The position of points in the collapsed section at
the bottom is obtained by integrating Eq. (5) from ξ=ξ1to ξ= 1, with the boundary value
X(ξ1) matching the result obtained by the first integration. Carrying out these calculations
gives
X(ξ) = 




ℓ1ξ+mg
kξ1−1
2ξξ, for 0 ≤ξ≤ξ1
ℓ1ξ+mg
2kξ2
1,for ξ1≤ξ≤1.
(10)
The total length of the slinky in this configuration is
XB=X(1) = ℓ1+mg
2kξ2
1,(11)
where B refers to the bottom of the slinky, and the center of mass is at
Xcom =Z1
0
X(ξ)dξ =1
2ℓ1+mg
2k1−1
3ξ1ξ2
1.(12)
The left panel of Fig. 4 illustrates the model slinky in the hanging configuration. The
slinky is drawn as a helix with a turn spacing matching X(ξ), for model parameter values
typical of real slinkies (detailed modeling of real slinkies is presented in Sec. IV). The
chosen parameters are: 80 total turns (N= 80), slinky mass m= 200 g, hanging length
XB= 1 m, collapsed length ℓ1= 60 mm, slinky radius 30 mm, and 10% of the slinky mass
collapsed at the bottom when hanging (ξ1= 0.9). The light gray (green online) section of
the slinky at the bottom is the collapsed section, and the dot (red online) is the location of
the center-of-mass of the slinkey (given by Eq. (12) in the left panel).
6

B. The falling slinky with instant collapse of turns
We assume the slinky is released at t= 0 and the turns collapse from the top down behind
a propagating wave front. In the model with instant collapse9the process is completely
described by the location ξc=ξc(t) of the front at time t. The slinky is collapsed where
0≤ξ < ξcbut is still in the initial state where ξc< ξ ≤1. The position of points in
the collapsed section of the slinky, behind the front, is obtained by integrating Eq. (5) and
matching to the boundary condition
x(ξ=ξc, t) = X(ξc),(13)
to get
x(ξ, t) = ℓ1ξ+mg
kξcξ1−1
2ξc(14)
for 0 ≤ξ≤ξc. The lower part of the slinky (ξc≤ξ≤1) has positions described by Eq. (10).
The motion of the slinky after release follows from Newton’s second law. The collapsed
top section has a velocity given by the derivative of Eq. (14)
∂x(ξ, t)
∂t =mg
k(ξ1−ξc)dξc
dt ,(15)
and the mass of this section is mξc. The rest of the slinky is stationary so the total momentum
of the slinky is obtained by multiplying Eq. (15) by the mass mξc. Setting the momentum
equal to the net impulse mgt due to gravity on the slinky at time tgives
ξc(ξ1−ξc)dξc
dt =k
mt, (16)
which can be directly integrated to give
ξ2
cξ1−2
3ξc=k
mt2.(17)
At a given time Eq. (17) is a cubic in ξc; the first positive root to the cubic defines the
location of the collapse front at that time. The total collapse time tc—the time for the
collapse front to reach the bottom, collapsed section—is defined by ξc(tc) = ξ1, and from
Eq. (17) it follows that
tc=rm
3kξ3
1.(18)
7

Equation (17), together with Eqs. (10) and (14), defines the location x(ξ, t) of all points
on the slinky for t < tc. The center of mass falls from rest with acceleration gand so has
location
xcom(t) = Xcom +1
2gt2,(19)
where Xcom is given by Eq. (12).
Figure 5 shows solution of the instant-collapse model with the typical slinky parameters
used in Fig. 4.12 The upper panel shows the positions of the top (upper solid curve, blue
online), center-of-mass (middle solid curve, red online), and bottom (lower solid curve, red
online) of the slinky versus time. Position is negative in the downward direction so the upper
(blue) curve corresponds to the model expression −x(0, t). The position of the front versus
time is indicated by the dashed (black) curve. The lower panel shows the velocity of the top
of the slinky versus time (solid curve, blue online) and in both panels the total collapse time
tcis indicated by the vertical dashed (pink) line. For the typical slinky parameters used,
the spring constant is k= 0.84 N/m and the collapse time is tc≈0.24 s.
Figure 5 illustrates a number of unusual features of the model. For example, the initial
velocity of the top is non-zero—a consequence of the assumption of instant collapse at the
wave front. From Eqs. (15) and (17) the initial velocity of the top is
v0
T=−∂x
∂t 


ξ=0
=−grmξ1
k≈ −4.5 m/s.(20)
The acceleration of the top at time t= 0 must be infinite to produce a finite initial velocity.
The acceleration of the top of the slinky just after t= 0 is positive, i.e. in the upwards
direction, so the top of the slinky falls more slowly with time. From Eqs. (15) and (17) the
limiting value of the acceleration as t→0 is
a0
T=−∂2x
∂t2


ξ=0
=g
3ξ1≈3.6 m/s2.(21)
The acceleration of the top becomes negative (downwards) after the collision of the top and
bottom sections, when the whole slinky falls with acceleration −g. At the collapse time
when the top section impacts the bottom section there is an impulsive collision causing a
discontinuous jump in the velocity.
8

C. The falling slinky with a finite time for collapse of turns
The instant-collapse model requires an unphysical instant change in the angle of the
slinky turns behind the collapse front, as discussed in Secs. II and III B. This affects the
positions of all turns of the slinky as a function of time behind the front. To model the
positions of the turns of a real collapsing slinky it is necessary to modify the model.
The lower panel of Fig. 3 indicates that the spacing between the turns of the slinky
decreases approximately linearly with time during the collapse. Hence we modify the model
in Sec. III C to include a linear profile for the decay in tension behind the collapse front
propagating down the slinky, as a function of mass fraction ξ. The tension is assumed to be
given by Eq. (4) with
∂x
∂ξ = [X′(ξc)−ℓ1]1 + ξ−ξc
∆+ℓ1for max(0, ξc−∆) ≤ξ≤ξc,(22)
where X=X(ξ) is given by Eq. (10), and the prime denotes differentiation with respect to
the parameter ξ. In this equation, ∆ is a parameter that governs the distance over which the
tension decays back to its minimum value f1. Figure 6 illustrates the local slinky extension
at time tas described by Eq. (22). Behind the front at ξc(t) the extension decreases linearly
as a function of ξ, returning to the minimum value ℓ1over the fixed mass fraction ∆. Ahead
of the front the extension is the same as for the hanging slinky.
Equation (22) replaces Eq. (5) for the section of the slinky behind the collapse front and
provides a simple, approximate description of a finite collapse time for the turns behind the
front. The limit ∆ →0 in the new model recovers the instant-collapse model.
Using Eq. (10) to evaluate the gradient in Eq. (22) gives
∂x
∂ξ =mg
k(ξ1−ξc)1 + ξ−ξc
∆+ℓ1for max(0, ξc−∆) ≤ξ≤ξc.(23)
Integrating Eq. (23) and imposing the boundary condition x(ξc) = X(ξc) using Eq. (10)
gives
x=mg
k(ξ1−ξc)1−1
∆ξc+1
2∆ξξ+ℓ1ξ+mg
2kξ2
c1 + ξ1−ξc
∆(24)
for max(0, ξc−∆) ≤ξ≤ξc. If ξc>∆, there is a completely collapsed section at the top of
the slinky. The mass density in this section is obtained by integrating Eq. (5) and matching
to the value x(ξc−∆) given by Eq. (24), leading to
x=ℓ1ξ+mg
kξcξ1−1
2ξc−1
2∆ (ξ1−ξc)for 0 ≤ξ≤max(0, ξc−∆).(25)
9

Equations (24) and (25) are the counterparts to Eq. (14) in the instant-collapse model. In
the limit ∆ →0, Eq. (25) is the same as Eq. (14).
The motion of the slinky in the new model is determined in the same way as for the
instant-collapse model. The velocity of the top section of the slinky prior to the complete
collapse of the top is obtained by differentiating Eqs. (24) and (25) to get
∂x
∂t =mg
k1 + ξ1−3ξc/2
∆ξc−1 + ξ1−2ξc+ξ/2
∆ξdξc
dt ,
for max(0, ξc−∆) ≤ξ≤ξc,
(26)
and
∂x
∂t =mg
kξ1−ξc+1
2∆dξc
dt ,for 0 ≤ξ≤max(0, ξc−∆).(27)
Equation (27) is the counterpart to Eq. (15). The total momentum of the slinky is given by
p=mZξc
0
∂x
∂t dξ , (28)
and using Eqs. (26) and (27) to evaluate the integral gives
p=m2g
2kξ2
c1 + ξ1−4ξc/3
∆dξc
dt if ξc≤∆,(29)
and
p=m2g
kξc(ξ1−ξc) + ∆ ξc−1
2ξ1−1
6∆dξc
dt if ξc≥∆.(30)
Setting Eqs. (29) and (30) equal to the total impulse mgt on the slinky up to time tgives
equations defining the location ξc(t) of the front at time t:
1
2ξ2
c1 + ξ1−4
3ξc
∆dξc
dt =kt
mif ξc≤∆,(31)
and
ξc(ξ1−ξc) + ∆ ξc−1
2ξ1−1
6∆dξc
dt =kt
mif ξc≥∆,(32)
which are the counterparts to Eq. (16) in the instant-collapse model. Equations (31) and (32)
may be integrated with respect to ξc, leading to
ξ3
c
∆ + ξ1−ξc
3∆ =kt2
mif ξc≤∆,(33)
and
ξ2
cξ1−2
3ξc−∆ (ξ1−ξc)ξc−1
3∆=kt2
mif ξc≥∆,(34)
10

which are the counterparts to Eq. (17) that defines the location of the front in the instant-
collapse model.
The total collapse time tcfor the slinky (the time for the front to reach ξ1) is obtained by
setting ξc=ξ1in Eq. (34). Interestingly, the result is unchanged from the instant-collapse
case and is given by Eq. (18). A second time scale relevant for the model is the time tlin
for the top of the slinky to undergo the initial linear collapse (for times t > tlin there are
completely collapsed turns at the top of the slinky). This is obtained by setting ξc= ∆ in
Eq. (33) to get
tlin =rmξ1
3k∆.(35)
Figure 7 shows solution of the finite-collapse time model for the typical parameters used
in Figs. 4 and 5, and with a value of ∆ chosen to match 10 turns of the 80-turn slinky
(∆ = 10/80 = 0.125). The layout of the figure is the same as for Fig. 5. The position versus
time of the top of the slinky (upper solid curve in the upper panel) is very similar to that
in the instant-collapse model, but the top initially accelerates downwards from rest rather
than having an initial non-zero velocity. The location of the front versus time (dashed curve
in the upper panel) is significantly different to that shown in Fig. 5, and comparison of this
curve and the position of the top of the slinky shows the effect of the finite time for turns
to collapse behind the front. A specific feature of the motion of the front is that the initial
velocity of the front is infinite (the dashed curve has a vertical slope at t= 0). The lower
panel of Fig. 7 plots the velocity versus time of the top of the slinky and shows that the top
is initially at rest, then accelerates rapidly until time tlin = 0.03 s during the initial linear
collapse, which is marked by a sudden change in curvature of the velocity profile. The initial
dynamics of the top differ from the instant-collapse model; in particular the velocity of the
top of the slinky at time t= 0 is zero, rather than having a finite value. However, after
the initial acceleration of the top, the velocity variation of the top is similar to that in the
instant-collapse model.
The right panel in Fig. 4 also illustrates the solution of the finite-collapse-time model
with the typical parameters, showing a helix drawn to match x(ξ, 1
2tc)—the model slinky
at one half the total collapse time. The upper, dark-gray (blue online) section of the helix
is the portion of the slinky above the collapse front, described by Eq. (23). The location
of the collapse front is shown by a dashed horizontal line, while the dot (red online) shows
11

the center of mass and the light-gray (green online) section at the bottom is the collapsed
section in the hanging configuration.
IV. MODELING REAL SLINKIES
A. Data
The finite-collapse-time model from Section III C is compared with data obtained for two
real slinkies, labeled A and B. The masses, lengths, and numbers of turns of the slinkies
are listed in Table I. Slinky A is a typical metal slinky and slinky B is the light plastic
rainbow-colored slinky shown in Fig. 2. These two slinkies were chosen because they have
significantly different parameters.
TABLE I. Measured data for two real slinkies.
Slinky A Slinky B
Mass m(g) 215.5 48.7
Collapsed length l1(mm) 58 66
Stretched length XB(m) 1.26 1.14
Number of turns N86 39
The slinkies are suspended from a tripod and released, and the fall is captured with a
Casio EX-F1 camera at 300 frames/s. The positions of the top and bottom of each slinky
are determined from the movies at time steps of τ= 0.01 s in each case. Figure 2 shows
frames from the movie used to obtain the data for slinky B.
B. Fitting the data and model
The finite-collapse time model from Section III C is applied to the data for the two slinkies
as follows. The observed positions for the top of each slinky during its fall are fitted to the
model using least squares for all time steps. The free parameters in the model are taken
to be the collapse mass fraction ∆, the spring constant k, and an offset t0to time, which
describes the time of release of the slinky compared to the time of the first observation. The
parameter t0is needed because the precise time of release is difficult to determine accurately.
12

The additional slinky parameters used are the measured values of the collapsed length ℓ1,
the hanging length XB, and the mass m. (Given ℓ1,XB,m, and a chosen value of k, Eq. (11)
determines the value of ξ1, so equivalently, ξ1could be taken as a free parameter instead of
k.)
The method of fitting is to fit the data values xT(tn) for the positions of the top of a slinky
(Tdenotes top) at the observed times tn= (n−1)τ(with n= 1,2, ...) to the model function
for the positions evaluated at the offset time, i.e. the fit is made to x(ξ, t) evaluated at ξ= 0
and t=tn−t0. The model function x(ξ, t) is defined by Eqs. (24), (25), (33), and (34) (and
by the hanging configuration X(ξ) for t < t0). This procedure correctly identifies t=t0as
the time of release.
Table II lists the best-fit parameters for the slinkies. The value of ∆ is given both as
a mass fraction and in terms of the corresponding number of turns of the slinky. For the
plastic slinky ξ1= 1, implying that no turns are collapsed at the bottom of the slinky in
the hanging configuration. Inspection of the top left frame in Fig. 2 suggests that this is
correct.
TABLE II. Best-fit model parameters for the slinkies.
Slinky A Slinky B
Spring constant k(N/m) 0.69 0.22
ξ10.89 1
ξ1(collapsed turns) 9.5 0
∆ 0.045 0.45
∆ (turns) 3.9 18
t0(s) 0.022 0.01
Total collapse time tc(s) 0.27 0.27
Linear collapse time tlin (s) 0.014 0.12
Figures 8 and 9 show the fits between the model and the observed data for Slinkies A
and B, respectively. The upper panel in each figure shows positions versus time for the
slinky top (model: upper solid curve, data: circles, blue online), turn 10 (model: middle
solid curve, data: squares, black online), and the slinky bottom (model: lower solid curve,
data: ×, green online). The lower panel in each figure shows the velocity of the top of the
13

slinky versus time (model: solid curve, data: circles, blue online). The measured velocity of
the top of the slinky is determined by centered differencing of the observed position values,
i.e. the velocity at time tnis approximated by
vT(tn) = xT(tn+1)−xT(tn−1)
2τ.(36)
These values are estimated for illustrative comparison with the model, but they are not used
in the fitting, which uses only the position data for the top shown in the upper panel. Note
also that the lower panel shows downward values as negative, i.e. it shows −vT(tn) versus tn.
Both panels in Figs. 8 and 9 also show the time offset t0for the model by the left vertical
(red online) dashed line, and the total collapse time for the model by the right vertical (pink
online) dashed line.
The results in Figs. 8 and 9 demonstrate that the model achieves a good qualitative fit
to the observed positions of the top of each slinky. The quality of the fit is shown in the
approximate reproduction of the values of the velocity of the top of each slinky obtained by
differencing the position data for the top. In particular, the description of the finite time for
collapse of the slinky top given by Eq. (22), with the best-fit model values, approximately
reproduces the observed initial variation in the velocity of the top of each slinky shown in
the lower panels of the figures. Although we do not attempt a detailed error analysis, it
is useful to consider the expected size of uncertainties in the data values. If the observed
position values are accurate to σx≈0.5 cm, the uncertainty in velocity implied by the
centered differencing formula Eq. (36) is
σv=σx
√2τ≈0.4 m/s.(37)
The detailed differences between the observed and best-fit model velocity values are approx-
imately consistent with Eq. (37).
The fit is better for slinky A than slinky B, as shown by specific discrepancies between
the model and observed data for the position of turn 10 (upper panel of Fig. 9), and the
velocity of the top (lower panel of Fig. 9). This may be due to the technique used to hang
the slinky: the top turns are tied together to allow the slinky to be hung vertically (see
the first frame in Fig. 2). About a turn and a half of the slinky was joined at the top, and
as a result the top of the slinky is heavier than in the model, and there is approximately
one fewer turn. The same technique was used for both slinkies, but the effect may be more
14

important for slinky B, which is significantly lighter and has fewer turns, than slinky A. We
make no attempt to incorporate this in our model.
The best-fit values for the model parameter ξ1may be checked by comparison with
the observed number of collapsed turns Ncat the bottom of each slinky in the hanging
configuration, which is given by Eq. (8). Alternatively, the model values for the spring
constant kmay be checked by comparison with the period of the fundamental mode of
oscillation of the slinky when it is hanging5
T0= 4rm
k.(38)
Table III lists the predictions for Ncand T0based on the model values of ξ1and k, and the
observed values for each slinky.
TABLE III. Predictions (for best-fit model parameters) and observations for the number of col-
lapsed turns when hanging, and for the fundamental mode frequency.
Slinky A Slinky B
Model fundamental period T0(s) 2.23 1.88
Observed fundamental period (s) 2.18 1.77
Model number of collapsed turns Nc9.5 0
Observed number of collapsed turns 10 0
Table III shows that the slinky model with best-fit parameters approximately reproduces
the observed fundamental mode periods and numbers of collapsed turns for the two slinkies.
(Note that the two model values Ncand T0are not independent.) The discrepancies between
the model and observed values for the fundamental periods are ∼5%, with the model values
being too large in both cases. It is useful to consider the expected size of discrepancies in
the period produced by observational uncertainties. From Eqs. (11), (18), and (38) it follows
that
tc=4
32(XB−ℓ1)
g3/41
√T0
.(39)
Assuming the distances XBand ℓ1are well-determined, Eq. (39) implies
σT0
T0
= 2σtc
tc
,(40)
where σT0and σtcare the uncertainties in T0and tcrespectively. Taking the value of the
time step τ= 0.01 s as a representative value for σtcin Eq. (40) gives σT0/T0= 0.08, i.e.
15

an 8% error in the model value for the fundamental mode period. This suggests that the
model values for T0are as accurate as might be expected from observational uncertainties,
and indicates that it is difficult to determine the mode period for a real slinky based on
measuring the fall of the slinky.
The technique of suspension of the top of the slinky, involving tying about a turn and a
half of the slinky together to ensure that it hangs vertically, introduces some uncertainty into
the modeling. It is interesting to investigate the effect of this step on the initial dynamics of
the slinky during the fall. For this purpose the slinky is suspended in two additional ways,
with a string tied across both sides of just the top turn, and with a string tied across both
sides of the top two turns, linked together. These methods of suspension involve fewer, and
greater numbers of turns tied together at the top, respectively, compared with the original
method (which had about a turn and a half tied together at the top). Figure 10 illustrates
the two methods of suspension, showing images (in inverted grayscale for clarity) of the
top few turns of the slinky in the two cases. The left-hand image shows the case with one
turn tied together at the top while the right-hand image shows the case with two turns tied
together.
With these methods of suspension the slinky is filmed being dropped, and data are
extracted for the first 0.06 s of the fall in each case; the results are shown in Fig. 11. The
upper panel shows the positions versus time for the top of the slinky and for the first turn
below the turns tied at the top, for each case: circles and squares, respectively for suspension
by one turn, and + and ×symbols, respectively for suspension by two. The lower panel
shows the velocities of the top in each case, obtained by differencing the position data using
Eq. (36) (circles for suspension by one turn and + symbols for suspension by two). These
results show that the top of the slinky accelerates from rest more rapidly when fewer turns
are tied together at the top, which is expected because the inertia of the top is reduced.
However, in both cases the top achieves a very similar velocity after ∼0.05 s. It is expected
that the subsequent dynamics of the collapse of turns will be similar in the two cases. The
dependence of the initial dynamics of the top on the method of suspension will influence
the estimates of model parameters, in particular the collapse mass fraction ∆. However, it
is expected that the estimate of the spring constant kwill be less influenced because this
parameter is determined largely by the identification from the data of the total collapse time
tc. The dependence of the fitting on the method of suspension of the top of the slinky could
16

perhaps be reduced by fitting to the positions of turns other than the top turn during the
fall.
V. CONCLUSIONS
The fall of a slinky illustrates the physics of a tension spring, and more generally wave
propagation in a spring. This paper investigates the dynamics of an initially stretched slinky
that is dropped. During the fall the slinky turns collapse from the top down as a wave front
propagates along the slinky. The bottom of the slinky does not begin to fall until the top
collides with it. A modification to an existing model9for the fall is presented, providing an
improved description of the collapse of the slinky turns. The modification is the inclusion
of a finite time for collapse of turns behind the downward propagating wave front. The new
model is fitted to data obtained from videos of the falls of two real slinkies having different
properties.
The model is shown to account for the observed positions of the top of each slinky in
the experiments, and in particular reproduces the initial time-profile for the velocity of
the top after release. The spring constant of the slinky is assumed as a free parameter
in the model, and the best-fit model values are tested by comparison with independent
determinations of the fundamental mode periods for the two slinkies, which depend on the
spring constants. The model values appear consistent with the observations taking into
account the observational uncertainties.
The new model for the slinky dynamics during the fall developed here is semi-analytic,
and allows treatment of a tension spring including approximate description of the dynamics
of the collapse of the spring. During the collapse of the top of the slinky the turns collide,
but the model does not describe this process in detail. Instead, the collapse is approximately
described by the assumption of a linear decrease in tension as a function of mass density along
the spring behind the front initiating the collapse. The linear approximation is motivated by
the experimental data from the slinky videos, which shows that the spacing between slinky
turns during the collapse decreases approximately linearly with time.
The behavior of a falling slinky is likely to be counter-intuitive to students and provides
a useful (and very simple) undergraduate physics lecture demonstration. The explanation of
the behavior may be supplemented by showing high-speed videos of the fall. The modeling of
17

the process presented here is also relatively simple, and should be accessible to undergraduate
students.
∗cross@physics.usyd.edu.au
†m.wheatland@physics.usyd.edu.au
1S. Y. Mak, “The static effective mass of a slinkyTM ,” Am. J. Phys. 61, 261–264 (1993).
2M. Sawicki, “Static elongation of a suspended slinkyTM ,” Phys. Teacher 40, 276–278 (2002).
3A.-P. Hu, “A simple model of a Slinky walking down stairs,” Am. J. Phys. 78, 35–39 (2010).
4J. M. Bowen, “Slinky oscillations and the notion of effective mass,” Am. J. Phys. 50, 1145–1148
(1982).
5R. A. Young, “Longitudinal standing waves on a vertically suspended slinky,” Am. J. Phys.
61, 353–360 (1993). Young’s expressions for mode periods and frequencies refer to the mass per
turn µ, and spring constant per turn κ, of the slinky. These are related to the values for the
whole slinky, used in this paper, by m=Nµ and k=κ/N, respectively.
6J. Blake and L. N. Smith, “The Slinky R
as a model for transverse waves in a tenuous plasma,”
Am. J. Phys. 47, 807–808 (1979).
7F. S. Crawford, “Slinky whistlers,” Am. J. Phys. 55, 130–134 (1987).
8G. Vandegrift, T. Baker, J. DiGrazio, A. Dohne, A. Flori, R. Loomis, C. Steel, and D. Velat,
“Wave cutoff in a suspended slinky,” Am. J. Phys. 57, 949–951 (1989).
9M. G. Calkin, “Motion of a falling spring,” Am. J. Phys. 61, 261–264 (1993).
10 J. M. Aguirregabiria, A. Hernandez, and M. Rivas, “Falling elastic bars and springs,” Am. J.
Phys. 78, 583–587 (2007).
11 W. G. Unruh, “The Falling slinky,” arXiv:1110.4368v1 (19 Oct 2011): <http://arxiv.org/
abs/1110.4368>.
12 The high-speed videos used to obtain data in this paper are available (as reduced, com-
pressed versions) as supplementary material from [AIP to insert URL] and also at <http:
//www.physics.usyd.edu.au/~wheat/slinky/>. Also provided are animations of the solutions
to the instant-collapse and finite-collapse-time model, and an animation of the fundamental
mode of oscillation of the hanging slinky.
13 The YouTube video is at <http://www.youtube.com/watch?v=eCMmmEEyOO0>.
18

14 T. W. Edwards and R. A. Hultsch, “Mass distribution and frequencies of a vertical spring,”
Am. J. Phys. 40, 445–449 (1972).
19

Tension Tension
Compression spring Tension spring
Length Length
0 0
Fig. 1. Tension versus length diagrams for a compression spring (left) and a tension spring (right).
The tension in each spring is zero for spring length ℓ0assuming Hooke’s law applies (this length is
not achieved for the tension spring). The turns of the spring touch for length ℓ1.
20

Fig. 2. Frames extracted from a high-speed video of the fall of a rainbow-colored slinky, illustrating
the collapse of the top of the slinky, and the continued suspension of the bottom after release of
the top. The top end of the slinky takes ∼0.25 s to reach the bottom.
21

0.00 0.05 0.10 0.15 0.20
−0.80
−0.60
−0.40
−0.20
0.00
Vertical position (m)
Top
Turn 8
Turn 10
0.00 0.05 0.10 0.15 0.20
0.00
0.02
0.04
0.06
0.08
0.10
Time (s)
Spacing of turns 8 and 10 (m)
Fig. 3. Data extracted from the video shown in Fig. 2, illustrating the finite time for collapse of
the turns of the slinky. Upper panel: position versus time of the top of the slinky (circles, blue
online), turn eight of the slinky (+ symbols, black online) and turn ten of the slinky (×symbols,
red online). Position is negative downwards in this panel. Lower panel: The spacing of turns eight
and ten versus time.
22

−0.10 0.00 0.10
−1.00
−0.90
−0.80
−0.70
−0.60
−0.50
−0.40
−0.30
−0.20
−0.10
0.00
Horizontal position (m)
Vertical position (m)
−0.10 0.00 0.10
−1.00
−0.90
−0.80
−0.70
−0.60
−0.50
−0.40
−0.30
−0.20
−0.10
0.00
Horizontal position (m)
Vertical position (m)
Fig. 4. Left panel: the model for a hanging slinky, with the slinky represented as a helix with turn
spacing matching X(ξ). Typical slinky parameters are used. The dot in each panel indicates the
location of the center of mass, and the light gray (green online) part of the slinky at the bottom
is the collapsed section. Right panel: the finite-collapse-time model for the slinky during the fall
at time t=tc/2. The top dark gray (blue online) section of the slinky is the section undergoing
collapse, above the downward-propagating collapse front indicated by a dashed horizontal line.
23

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
−1.20
−1.00
−0.80
−0.60
−0.40
−0.20
0.00
Vertical position (m)
Top
Bottom
com
Wave front
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
−4.50
−4.00
−3.50
−3.00
−2.50
Time (s)
Velocity of top (m/s)
Fig. 5. The instant-collapse model9for a falling slinky using parameters typical of a real slinky.
Upper panel: position versus time of the slinky top (upper solid curve), center-of-mass (middle
solid curve), bottom (lower solid curve), and wave front initiating collapse (dashed curve). Position
is negative downwards in this figure. Lower panel: velocity of the slinky top versus time. The total
collapse time tcis shown as the vertical dashed line in both panels.
24

c
ξ −∆
c
(ξ )X
c
ξ ( )tξ1
0
0ξ1
Fig. 6. The gradient ∂x/∂ξ, which describes the local slinky extension, versus mass density ξin
the finite-collapse-time model. The tension defined by this profile declines linearly behind the wave
front [located at ξc(t)] from a value matching the tension in the hanging slinky at the front, to the
minimum tension value f1=k(ℓ1−ℓ0) at ξ=ξc−∆. Ahead of the front the tension is unchanged
from that in the hanging slinky.
25

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
−1.20
−1.00
−0.80
−0.60
−0.40
−0.20
0.00
Vertical position (m)
Top
Bottom
com
Wave front
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
−5.00
−4.00
−3.00
−2.00
−1.00
0.00
Time (s)
Velocity of top (m/s)
Fig. 7. The finite-collapse-time model for a falling slinky using the same slinky parameters as in
Fig. 5. The collapse of the model slinky is assumed to occur via a linear decay in tension over
ten turns of the slinky. Upper panel: position versus time of the slinky top (upper solid curve),
center-of-mass (middle solid curve), bottom (lower solid curve), and wave front initiating collapse
(dashed curve). Position is negative downwards in this figure. Lower panel: velocity of the slinky
top versus time. The total collapse time tcis shown as the vertical dashed line in both panels.
26

0.00 0.05 0.10 0.15 0.20 0.25 0.30
−1.20
−1.00
−0.80
−0.60
−0.40
−0.20
0.00
Vertical position (m)
0.00 0.05 0.10 0.15 0.20 0.25 0.30
−5.00
−4.00
−3.00
−2.00
−1.00
0.00
Time (s)
Velocity of top (m/s)
Fig. 8. The finite-collapse-time model applied to slinky A. The upper panel shows position versus
time for the slinky top (upper), turn 10 (middle), and slinky bottom (lower), with the observed
data represented by symbols and the best-fit model values by curves. The fitting is based on the
observed positions of the slinky top. The lower panel shows the velocity of the slinky top versus
time. The vertical dashed lines in both panels show the time of release of the slinky (left), which
is a model parameter, and the model collapse time (right).
27

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
−1.20
−1.00
−0.80
−0.60
−0.40
−0.20
0.00
Vertical position (m)
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
−4.00
−3.00
−2.00
−1.00
0.00
Time (s)
Velocity of top (m/s)
Fig. 9. The finite-collapse-time model applied to slinky B. The presentation is the same as in
Fig. 8.
28

Fig. 10. An experiment with different methods of suspension of the top of slinky B. In the left-hand
image the top is suspended from a string tied across a diameter of the first turn of the slinky. In
the right-hand image the string is tied around the first two turns.
29

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07
−0.25
−0.20
−0.15
−0.10
−0.05
0.00
Vertical position (m)
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07
−5.00
−4.00
−3.00
−2.00
−1.00
0.00
Time (s)
Velocity of top (m/s)
Fig. 11. Data extracted for the initial fall of slinky B following suspension using the two methods
shown in Fig. 10. The circles and squares show results for suspension by one turn and the + and
×symbols for suspension by two turns. The upper panel shows the position versus time of the top
and of the first turn below the initially tied top section. The lower panel shows the velocities of
the top in each case, obtained by differencing the position data (circles for suspension by one turn
and + symbols for suspension by two).
30