A trip to the end of the universe and the twin “paradox”

Thomas Müller,a兲 Andreas King, and Daria Adis

Institut für Astronomie und Astrophysik, Abteilung Theoretische Astrophysik, Auf der Morgenstelle 10,
72076 Tübingen, Germany
共Received 12 December 2006; accepted 8 December 2007兲
Special relativity offers the possibility of going on a trip to the center of our galaxy or even to the
end of our universe within a lifetime. On the basis of the well known twin paradox, we discuss
uniformly accelerated motion and emphasize the local perspective of each twin concerning the
interchange of light signals between both twins as well as their different views of the stellar sky. For
this purpose we developed two Java applets that students can use to explore interactively and
understand the topics presented here. © 2008 American Association of Physics Teachers.
关DOI: 10.1119/1.2830528兴

I. INTRODUCTION
The purpose of this article is to emphasize the local per-
spective of each twin in the well known twin paradox who
can only do measurements with respect to his/her individual
reference frame. This local perspective resembles Bondi’s
k-calculus1 or the concept of radar time as discussed by
Dolby and Gull,2 where hypersurfaces of simultaneity are
defined by local measurements of reflected light signals. The
interesting information that both twins can interchange in our
example are their individual proper times. They will not be
able to determine the time dilation from this proper time
because the information carrying the proper time of the other
twin needs some time to travel the distance between them.
This time interchange, as the motion itself, is not symmetric.
Besides the time signals, both twins also receive light from
distant stars or other astronomical sources. Depending on the
relative velocity with respect to the light source, each twin
will have a different view of the stellar sky because of aber-
ration and the Doppler shift.
It is well known that acceleration is not crucial to explain
the twin paradox.3,4 共For a detailed discussion of the twin
paradox we refer the reader to the standard literature.5,6,8兲
The crucial idea of the twin paradox is that there is an asym-
metry between both twins. Only the difference in length of
their paths makes the twins age at different rates. The trav-
eling twin Tina leaves Earth with a rocket starting with zero
velocity. Having reached her destination the twin might re-
turn to Earth. In order for the journey to be comfortable we
consider a uniformly accelerated motion which is separated
into four phases of equal duration. Even though acceleration
is usually considered in conjunction with general relativity, it
can also be treated within special relativity.7–9
We expect the reader to be familiar with basic calculations
such as the Lorentz transformations and the description of
acceleration within special relativity. In Sec. II we review the
the visual effects due to aberration and the frequency shift. In
the last section we show how far we can travel with the help
of time dilation and length contraction.
In the appendix we give a short introduction to the Java
applets: TwinApplet and RelSkyApplet.11 The
TwinApplet illustrates how far the traveling twin can go
by means of relativistic time dilation and what time signals
they can receive from each other. To use this applet we rec-
ommend reading Secs. II–V. The visualization of the stellar
sky discussed in Sec. VII is realized in RelSkyApplet.
II. UNIFORM ACCELERATION
The earth twin Eric stays in the inertial reference frame S,
while his traveling twin sister Tina leaves the Earth with
constant acceleration ␣ with respect to her own instanta-
neous system S⬘. All primed quantities refer to S⬘, and
unprimed quantities to S. From the Newtonian point of view,
Tina will reach a velocity v = ␣⌬t in the time ⌬t. She will
then have covered a distance ⌬x = 21 ␣⌬t2. Here, both twins
will agree on time, velocity, and distance. In the theory of
special relativity, both twins will still agree concerning time,
velocity, and distance as long as Tina’s velocity is much less
than the speed of light, v Ⰶ c. But in general, the earth twin
Eric would measure an acceleration5,8
␣
a = ␣共1 − ␤2兲3/2 = , 共1兲
␥共␤兲3
with ␤ = v / c and ␥ = 1 / 冑1 − ␤2, which is different to the New-
tonian case. Given the initial velocity v0, we obtain Tina’s
velocity v at time t by substituting a = dv / dt in Eq. 共1兲. The
result is
␣t + c␨
v= 共2兲
冑1 + 共␣t/c + ␨兲2 ,
with ␨ = ␥共␤0兲␤0. The distance traveled x = x共t兲 follows from
Eq. 共2兲 by another integration,

冋冑 冉 冊 册
description of uniformly accelerated motion in special rela-
tivity. Following the example of Ruder10 we present a special c2 ␣t 2
round trip consisting of four equal acceleration phases in x= 1+ +␨ − 冑1 + ␨2 + x0 , 共3兲
␣ c
Sec. III. From the local perspective taken in Sec. IV only
time or light signals can be measured by each twin. As an where x0 is the initial position at time t = 0.
example, we consider a flight to Vega and examine the time Time in both systems will go by at different rates depend-
signal exchange in Sec. V. Because we stress the local per- ing on the relative velocity ␤,
spective, we derive the frequency shift and aberration rela-
dt⬘ = ␥共␤兲−1dt. 共4兲
tions for the visualization of the stellar sky by means of a
local reference frame 共see Sec. VI兲. In Sec. VII we describe Synchronizing at t = t⬘ = 0 gives

360 Am. J. Phys. 76 共4&5兲, April/May 2008 http://aapt.org/ajp © 2008 American Association of Physics Teachers 360
 t⬘ =
c
␣
冋 冉 冊
arcsinh
␣t
c
+ ␨ − arcsinh ␨ , 册 共5兲

or

t=
c
␣
冋 冉
sinh
␣t⬘
c
+ arcsinh ␨ − ␨ . 冊 册 共6兲

For ␤0 = 0 Eq. 共6兲 simplifies to

c ␣t⬘
t= sinh . 共7兲
␣ c
If the traveling twin starts at x0 = 0, her current position x
= x共t⬘兲 is given by

x=
␣
冉
c2
cosh
␣t⬘
c
−1 . 冊 共8兲

Tina’s velocity with respect to her proper time t⬘ follows

from Eqs. 共7兲 and 共2兲, Fig. 1. Tina’s journey is separated into four phases. She starts from point ¬,
␣t⬘ accelerates up to maximum velocity at point −, and slows down until she
v = c tanh . 共9兲 reaches the turning point ®. Then she accelerates in the opposite direction
c and slows down again until she comes home. Signals that were emitted by
Tina in the accelerating phase i reach the Earth twin Eric in the interval
Hence, because 兩tanh共x兲兩 艋 1, the locally measured speed is 关ti , ti+1兴.
always less than the speed of light in accordance with special
relativity.

III. A SPECIFIC ROUND TRIP
The case we will investigate is discussed in several stan-
dard textbooks on special relativity, but not in as much
detail.1,5,10
While Eric stays at home, Tina goes on a journey that is
separated into four phases each of which last time T⬘. In the
first phase she starts at space-time point ¬ and accelerates
with ␣ until she reaches the maximum velocity
␣T⬘
vmax = c tanh
c
at point −. Next, she decelerates with −␣ until she stops at
her destination ® where she has covered the distance
tive. Hence, we can only use light signals exchanged by both
twins to determine the proper time in each case. To do so the
⬘
signals are coded with the proper emission time temit or temit
which will be received by the other twin at observation time
⬘ or tobs. This measured time is not the actual time of the
tobs
other twin. Because of the finite speed of light, the signals
need some time to travel the distance between both twins.
Eric’s perspective. Eric observes at his proper time tobs a
signal sent from Tina. Because of the finite speed of light, the
signal must have been emitted at Earth time temit,
共10兲
x共temit兲
tobs = temit + . 共13兲
c

xmax = 2
␣
冉
c2
cosh
␣T⬘
c
−1 . 冊 共11兲
The difference between tobs and temit is the time needed for
the signal to travel from Tina’s current position xemit
= x共temit兲 to the Earth. To determine the position xemit we have
The same procedure, but now in the opposite direction, to find out the interval of acceleration when Tina has emitted
brings Tina back home. Thus, the whole trip takes 4T⬘ with the signal 共compare Fig. 1兲.
respect to Tina’s proper time, whereas Eric has to wait the The border points ti, 共i = 2 , 3 , 4兲, of the intervals follow
time from conditions such as t2 = t− + x− / c, whereas t5 follows
c ␣T⬘ from Eq. 共12兲. Thus, we have
4T = 4 sinh 共12兲
␣ c
t1 = 0, 共14a兲
for his sister’s return. Hence, the bigger the acceleration and

冉 冊
the longer the journey lasts, the bigger is the effect of time
dilation and the different aging of both twins. Tina’s world- c ␣T⬘ ␣T⬘
t2 = sinh + cosh −1 , 共14b兲
line is shown in a spacetime diagram in Fig. 1, where Eric’s ␣ c c
proper time ct is plotted versus the distance x between Tina

冉 冊
and Eric.
c ␣T⬘ ␣T⬘
t3 = 2 sinh + 2 cosh −2 , 共14c兲
IV. OBSERVATION OF TIME SIGNALS ␣ c c

冉 冊
In contrast to usual discussions of special relativity con-
cerning length contraction or time dilation with a group of c ␣T⬘ ␣T⬘
t4 = 3 sinh + cosh −1 , 共14d兲
synchronized observers, we concentrate on a local perspec- ␣ c c

361 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008 Müller, King, and Adis 361
 c ␣T⬘ ⬘ of the signal follows from the intersection of the
time tobs
t5 = 4 sinh . 共14e兲 future light cone of Eric at time temit with Tina’s world line,
␣ c
Depending on the interval 关ti , ti+1兴 Eric receives signals from xobs
tobs = temit + . 共23兲
the accelerating phase i, where Tina’s current position was c
xemit = x共temit兲. In the first time interval t1 − t2 we get from Eq.
共3兲 with x0 = 0, In contrast to Eric’s perspective the calculations are a bit
more straightforward because we determine temit from tobs ⬘ .
x¬–−共temit兲 While Tina is in the first accelerating phase ¬-− her proper
tobs = temit + 共15a兲
c ⬘ transforms into tobs via Eq. 共7兲. Thus, her position is
time tobs

冉冑 冊
given by

冉 冊
c ␣2temit
2
=temit + 1+ −1 . 共15b兲 c2 ␣t⬘
␣ c2 xobs = cosh obs − 1 , 共24兲
␣ c
We solve Eq. 共15b兲 for temit and substitute it into Eq. 共5兲 with
⬘ ,
␤0 = 0 and find the observed signal time temit ⬘ has the
and the signal from Eric which arrives at time tobs
c ␣temit c ␣tobs
冉 冊 time signature

冉 冊
⬘ =
temit arcsinh = ln 1 + . 共16兲
␣ c ␣ c c ␣t⬘ ␣t⬘
temit = sinh obs − cosh obs + 1 . 共25兲
In the accelerating time interval t2 − t4, Eric’s observation ␣ c c
time tobs is related to the emission time temit via
In the accelerating phase −-¯ we obtain t̃obs = tobs − T from
x−–¯共t̃emit兲 Eq. 共6兲 with ␨ = sinh共␣T⬘ / c兲 for the observation time t̃obs ⬘
tobs = T + t̃emit + , 共17兲
c ⬘ − T⬘. Thus, we have
= tobs

where t̃emit = temit − T measures time from point −. Tina’s cur-

rent position x−–¯共t̃emit兲 follows from Eq. 共3兲, tobs = T −
c
␣
冉
sinh
⬘ + 2T⬘兲
␣共− tobs
c
− sinh
␣T⬘
c
. 冊 共26兲

x−–¯共t̃emit兲 = −
c2
␣
冉冑 冉 1+
− ␣t̃emit
c
+␨ 冊 2
− 冑1 + ␨2 冊 Tina’s position xobs when she receives the signal is given by
Eq. 共3兲 which simplifies to

+
c2
␣
冉冑 1+
␣ 2T 2
c2
−1 , 冊 共18兲 xobs = −
c2
␣
冉
cosh
⬘ + 2T⬘兲
␣共− tobs
c
− 2 cosh
␣T⬘
c
+1 . 冊
where ␨ = ␤0␥共␤0兲, and the velocity ␤0 is given by 共27兲
␣T/c The same procedure also applies to the last accelerating
␤0 = 共19兲
冑1 + ␣2T2/c2 . phase ¯-° for which we find

From Eq. 共17兲 we can determine t̃emit to be

1 2␰冑1 + ␨2 − ␰2␣/c
tobs = 3T +
c
␣
冋
sinh
⬘ − 4T⬘兲
␣共tobs
c
+ sinh
␣T⬘
c
册 共28兲
t̃emit = , 共20兲
2 冑1 + ␨2 + ␨ − ␰␣/c for the observation time, and
with

␰ = tobs − T −
c
␣
冉冑 1+
␣ 2T 2
c2
−1 . 冊 共21兲

Equation 共5兲 gives the corresponding time t̃emit ⬘ . Hence, Eric

receives at his observation time tobs the emission time temit ⬘
⬘ + T ⬘.
= t̃emit
In the last time interval t4 − t5 Eric’s observation time tobs
follows from
x¯–°共t̄emit兲
tobs = t̄emit + 3T + , 共22兲
c
where x¯–°共t̄emit兲 equals x−–¯共t̃emit兲 with the substitutions
␣ 哫 −␣, ␨ 哫 −␨, t̃emit 哫 t̄emit, and t̄emit = temit − 3T. To obtain
the emission time temit from tobs we follow the same proce-
dure as in the previous accelerating time interval. Fig. 2. Tina’s current velocity for her flight to Vega and return to Earth with
Tina’s perspective. Now we consider the opposite situation respect to her own proper time. The maximum speed ␤ ⬇ 0.9975 is reached
where Eric sends his proper time temit to Tina. The arrival at points − and ¯.

362 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008 Müller, King, and Adis 362

⬘ 共ordinate兲 at
Fig. 3. The earth twin Eric sees/receives Tina’s proper time temit
his proper time tobs 共abscissa兲.
冋 册
c2 ⬘ − 4T⬘兲
␣共tobs
xobs = cosh −1 共29兲
␣ c
for Tina’s position. In both cases, the emission time temit
follows from Eq. 共23兲 with the corresponding time tobs and
position xobs.
V. FLIGHT TO VEGA
As an example we consider a flight to our neighboring star
Vega, which is roughly 25.3 ly away.12 To be a most com-
fortable journey Tina accelerates with ␣ = g ⬇ 9.81 m / s2.
⬘ = 4T⬘ ⬇ 12.93 years with re-
While the journey takes Ttotal
spect to her proper time, Eric has to wait Ttotal
⬇ 54.48 years for his sister to return. The worldline of Tina
with respect to Eric’s rest frame is shown in the spacetime
diagrams, Figs. 4 and 6. Even though Tina follows an accel-
erated motion, her worldline is almost linear. Because of the
moderate acceleration Tina’s rocket needs roughly one year
Fig. 4. Spacetime diagram for Tina’s flight to Vega and return to Earth with
respect to Eric’s frame. At point ® Tina reaches Vega and immediately
returns. At − and ¯ she changes her acceleration direction. The small
circles represent the events when Tina sends a time signal to Eric. They are
separated by a single year with respect to her proper time t⬘. Note that the
time units are not equally spaced on Tina’s worldline.
363 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008
Fig. 5. The rocket twin Tina sees/receives Eric’s proper time temit 共ordinate兲
⬘ 共abscissa兲.
at her proper time tobs
to reach 80% of the speed of light. At position x− she reaches
maximum speed ␤max ⬇ 0.9975 共compare with Fig. 2兲.
Consider the exchange of time signals between both twins.
The proper time sent by the one twin and received by the
other is plotted in a time-time diagram 共compare Figs. 3 and
5兲. The proper time of the twin who receives a signal is
plotted on the abscissa and the proper time, which was sent
by the other twin, is shown on the ordinate. The shapes of the
curves are difficult to understand because two effects are
mixed, time dilation and the nonlinear change of distance. To
explain the differences between these curves, we also show
the space-time diagram with Tina’s worldline where the time
signals are represented by straight lines with 45° slope 共com-
pare Figs. 4 and 6兲. Time intervals along Tina’s worldline are
not equally spaced because of her accelerated motion.
Eric’s perspective 共Figs. 3 and 4兲. While Eric stays at
home, Tina leaves Earth with proper acceleration ␣. Because
the distance between both twins grows, it is obvious that the
signals emitted by Tina need more and more time to reach
Eric. Also, because of time dilation, Tina seems to wait a
longer time than expected before emitting the next signal.
Even though Tina is already on the way back to home when
Fig. 6. The earth twin Eric sends a time signal every year with respect to his
proper time t. The dashes on Tina’s worldline mark the years of her proper
time. Eric’s first signal does not reach Tina until she already decelerates to
her destination ®.
Müller, King, and Adis 363
she has left her destination ®, Eric will collect most of her
time signals only during the very end of her journey.
Eric observes at his proper time tobs ⬇ 20 years a time sig-
⬘ = 3 y when
nal which was sent by Tina at her proper time temit
she was 10 ly away from home. But her current position at
Eric’s observation time is x ⬇ 19 ly and her watch shows
roughly 4 y.
Tina’s perspective 共Fig. 5 and 6兲. It might be obvious that
Tina’s perspective is different. Eric’s “first year” signal
reaches Tina only when she already decelerates to her desti-
nation ®. On her way home, Tina seems to receive the sig- Fig. 7. Wave vector k of an incoming light ray in spherical coordinates ␽
nals regularly, but because of time dilation she collects most 苸 共0 , ␲兲 , ␸ 苸 共−␲ , ␲兲 with respect to Eric’s rest frame. The twin Tina is
of the signals around point ¯ when she has maximum speed currently moving with velocity ␤ along the ex direction.
and the time dilation effect is strongest.
To investigate this situation for different proper accelera-
tions ␣ and different durations T⬘, we have written the inter-
active Java applet TwinApplet.11 A short description of ␣t⬘ ␣t⬘
the applet is given in Appendix A; a more detailed discussion eS1⬘共t⬘兲 = sinh et + cosh e x, eS3⬘共t⬘兲 = ez . 共35b兲
c c
as well as two examples can be found in the help menu of the
applet. With the help of the local tetrad it is straightforward to
derive the aberration and Doppler effect formula. Consider a
wave vector k of an incoming light ray 共compare Fig. 7兲.
VI. THE ACCELERATED REFERENCE FRAME This wave vector can either be described with respect to
Tina’s frame 兵eSi ⬘其共i=0,1,2,3兲 or to Eric’s rest frame 兵e j其共j=t,x,z,y兲.
A. Aberration, Doppler shift and length contraction
Thus,
In four-dimensional spacetime all observers have their
own local reference system which is given by four base vec- k = ␻共et − sin ␽ cos ␸ex − sin ␽ sin ␸ey − cos ␽ez兲 共36a兲
tors. Each measurement is taken with respect to this local
tetrad. = ␻⬘共eS0⬘ − sin ␽⬘ cos ␸⬘eS1⬘ − sin ␽⬘ sin ␸⬘eS2⬘
Consider the flat Minkowski spacetime which is repre-
sented by the metric − cos ␽⬘eS3⬘兲. 共36b兲
2 2 2 2
ds = − c dt + dx + dy + dz . 2 2
共30兲 If we transform both representations into coordinates, we can
compare each component. Thus, the time component gives
The local tetrad 兵ei其共i=0,. . .,3兲 of an observer moving with ve- the Doppler shift
locity ␤ = v / c in the direction of the x axis reads
␻ = ␻⬘␥共1 − ␤ sin ␽⬘ cos ␸⬘兲, 共37兲
e0 = ␥共et + ␤ex兲, e2 = e y , 共31a兲
and from Eq. 共37兲 we obtain the redshift factor z,
e1 = ␥共␤et + ex兲, e3 = ez , 共31b兲
␻
where 共et , ex , ey , ez兲 are the four base vectors of an observer z= − 1 = ␥共1 − ␤ sin ␽⬘ cos ␸⬘兲 − 1. 共38兲
␻⬘
at rest with respect to the Minkowski coordinate system 共30兲.
Each base vector points in its positive coordinate direction, The spatial components give the aberration formulas
where e0 is adapted to the four-velocity u of the moving
cos ␽⬘
observer, cos ␽ = , 共39a兲
␥共1 − ␤ sin ␽⬘ cos ␸⬘兲
1
e0 = u. 共32兲
c sin ␽⬘ cos ␸⬘ − ␤
cos ␸ = , 共39b兲
The current four-velocity of the accelerated twin Tina at time sin ␽共1 − ␤ sin ␽⬘ cos ␸⬘兲
t⬘ is given by u = u je j , 共j = t , x , y , z兲 with
dt dx sin ␽⬘ sin ␸⬘
ut = c = c␥共␤兲, ux = = ␤␥共␤兲, 共33兲 sin ␸ = . 共39c兲
dt⬘ dt⬘ ␥ sin ␽共1 − ␤ sin ␽⬘ cos ␸⬘兲

and uy = uz = 0. Her four-acceleration a␮ is The inverse formulas for the Doppler shift and the aberration
follow from Eqs. 共37兲 and 共39兲 by letting 共␤ , ␻ , ␽ , ␸兲
d 2t ␣ d 2x 哫 共−␤ , ␻⬘ , ␽⬘ , ␸⬘兲. Because aberration and the Doppler shift
at = = ␤␥共␤兲, ax = = ␣␥共␤兲, 共34兲
dt⬘2 c dt⬘2 depend only on the relative direction, we can also use the
and ay = az = 0, where ␤ = ␤共t⬘兲 is the relative velocity at Ti- angle ␹ between the wave vector −k and the direction of
motion ex with
na’s proper time t⬘. Thus, her local tetrad at time t⬘ is given
by cos ␹ = sin ␽ cos ␸ . 共40兲
␣t⬘ ␣t⬘ From Eq. 共38兲 it follows that Tina will see objects to be
eS0⬘共t⬘兲 = cosh et + sinh e x, eS2⬘共t⬘兲 = ey , 共35a兲
c c blueshifted for z ⬍ 0 and redshifted for z ⬎ 0.

364 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008 Müller, King, and Adis 364
 Fig. 9. An object p is located at position 共r p , ␹ p兲 with respect to the initial
position x共␥ = 1兲 of Tina. After some time, Tina’s current position is x共␥兲 and
the point p would have the relative position 共r , ␹兲 with respect to an ob-
server at rest at Tina’s current position.

Fig. 8. An object a distance d from the origin x = 0 has an apex angle ␰0. An
observer at rest at the current position x共t⬘兲 would measure an apex angle
␰共t⬘兲.
dous magnification effect, the distance to the object grows
exponentially and dominates the magnification effect.
Example. If Tina leaves home with acceleration ␣ = −g
A similar consideration leads to length contraction. A fixed away from the Sun,13 the Sun has an apex angle ␰0
distance ᐉ⬘ to some point in the direction 共␽⬘ , ␸⬘兲 with re- ⬇ 0.267° at a distance d ⬇ 1.5⫻ 1011 m. After tn⬘ ⬇ 500 s Tina
spect to Tina’s current frame would have a length reaches ␤ ⬇ 1.6⫻ 10−5, and the maximum magnification is
ᐉ = ᐉ⬘冑sin2 ␽⬘共␥2 cos2 ␸⬘ + sin2 ␸⬘兲 + cos2 ␽⬘ 共41兲 ␰⬘共tn⬘兲 1 ␣d
⬇1− ⬇ 1 + 8.2 ⫻ 10−6 . 共46兲
as determined by Eric. ␰0 2 c2
The angular size as seen by an observer at rest at her current
B. Apparent size of an object distance d共t⬘兲 ⬇ d + 1.2⫻ 106 m would be ␰ / ␰0 ⬇ 1 − 8.2
⫻ 10−6.
For accelerated motion objects in the direction of motion
always seem to recede in the first moments even though one
is approaching. The reason is also based on the aberration
effect. Consider an object a distance d apart with apex angle C. Apparent position of an object
␰0 as seen from the origin x = 0, Fig. 8.
An observer at rest at the current position x given by Eq. What’s the apparent position of an object p with respect to
共8兲 of the accelerating twin Tina would measure a current the accelerating observer Tina? If we parameterize Tina’s
apex angle ␰共t⬘兲 with current position, Eq. 共8兲, by her velocity, Eq. 共9兲, we obtain
d tan ␰0 c2 c2
tan ␰共t⬘兲 = . 共42兲 x= 兵cosh关arctanh共␤兲兴 − 1其 = 共␥ − 1兲. 共47兲
d − x共t⬘兲 ␣ ␣
The acceleration ␣ in Eq. 共8兲 determines whether Tina ap- The position x p = r p cos ␹ p, y p = r p sin ␹ p of the object p,
proaches 共␣ ⬎ 0兲 the object or recedes 共␣ ⬍ 0兲. The apex where r p ⬎ 0 is the distance to the initial position of Tina and
angle ␰⬘共t⬘兲 of the object with respect to Tina follows from 0 艋 ␹ p 艋 ␲ is the initial angle with respect to Tina’s direction
the aberration formula 共39b兲, of motion 共compare Fig. 9兲, transforms into Tina’s frame
according to the aberration formula,
1 + ␤共t⬘兲冑1 + tan2 ␰共t⬘兲
cos ␰⬘共t⬘兲 = , 共43兲 共x p − x兲/r + ␤
冑1 + tan2 ␰共t⬘兲 + ␤共t⬘兲 cos ␹⬘ = , 共48兲
1 + ␤共x p − x兲/r
where ␤共t⬘兲 is given by Eq. 共9兲. Hence, from the derivative
where r2 = 共x p − x兲2 + y 2p.
of Eq. 共43兲 with respect to Tina’s proper time t⬘ it follows
that Equation 共48兲 can be written for ␤ 艌 0 as

冏 冏d␰⬘
dt⬘ t⬘=0
␣
= − sin ␰0 .
c
共44兲 cos ␹⬘ =
关cos ␹ p − A共␥ − 1兲兴/r̃ + 冑1 − 1/␥2
1 + 冑1 − 1/␥2关cos ␹ p − A共␥ − 1兲兴/r̃
, 共49兲

Thus, accelerating from zero velocity in the direction to the with A = c2 / 共␣r p兲 and
object 共␣ ⬎ 0兲 has the effect that the object initially appears
r̃ = 冑A2共␥ − 1兲2 − 2A共␥ − 1兲cos ␹ p + 1. 共50兲
to shrink, giving the impression of receding from it instead
of approaching. This effect holds until d␰⬘ / dt⬘ = 0, The observation angle ␹⬘ = ␹⬘共␤兲 is shown in Figs. 10 and 11
c ␺1 + sign共␣兲␦冑共1 + tan2 ␰0兲␺2 for two values of A.
tn⬘ = ln , 共45兲 From the first derivative d␹⬘ / d␥ = 0 of Eq. 共49兲 we find an
␣ 2共1 + ␦兲 extremum at
where ␦ = ␣d / c2, ␺1 = 共1 + ␦兲2 + 1 + ␦2 tan2 ␰0, and ␺2 = 共2 1
+ ␦兲2 + ␦2 tan2 ␰0. ␥e = + 1, 共51兲
2A共A + cos ␹ p兲
An acceleration in the opposite direction 共␣ ⬍ 0兲 results in
a magnification which has a maximum at time tn⬘. Even which is valid only if 共A + cos ␹ p兲 ⬎ 0. The associated veloc-
though a velocity close to the speed of light has a tremen- ity ␤e is

365 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008 Müller, King, and Adis 365

Fig. 10. The observation angle ␹⬘ is plotted versus the velocity ␤ for A
= c2 / 共␣r p兲 ⬇ 9.68. In the first instance, an object with fixed distance r p de-
pending on A and arbitrary angle ␹ p apparently approaches the center of
motion. For higher velocities, it recedes again. Because ␤ = 1 is reached only
⬘ = ␹⬘共␤ → 1兲.
approximately, an object at 共r p , ␹ p兲 seems to “freeze” at ␹lim
冑1 + 4A cos ␹p + 4A2
␤e = . 共52兲
1 + 2A cos ␹ p + 2A2
For 共A + cos ␹ p兲 ⬎ 0 the second derivative follows after a
lengthy calculation to be
Fig. 12. Stellar sky at ␤ = 0.5 in the 4␲-representation where ␸ = 共−␲ , ␲兲 is
the abscissa and ␽ = 共0 , ␲ / 2兲 is the ordinate. The center of the image 共␽
= ␲ / 2 , ␸ = 0兲 corresponds to the direction of motion. The circles of latitude
and the meridians are separated by 5°.
eration ␣ of the observer. This dependence might be coun-
terintuitive in some cases. From a nonrelativistic point of
view, any object is located nearly behind the observer if the
latter is infinitely apart from it. But in the relativistic case,
the object could also appear at a different position. In brief, it
is a contest between the relativistic aberration and the expo-
nential increase of distance between star and observer.

冏 冏
d 2␹ ⬘
d␥2 ␥e
=
sin ␹ p
A共␥2e − 1兲2
艌 0. 共53兲 VII. VISUALIZATION OF THE STELLAR SKY
A. Aberration and Doppler shift
Thus, if there is an extremum, it is always a minimum.
In the limit ␤ → 1, ␥ goes to infinity, and an object with What could Tina really see if she looked through a win-
coordinates 共r p , ␹ p兲 seems to “freeze” at an observation angle dow of her rocket? If there were a sphere fixed at infinity
⬘ with respect to Tina’s frame,14
␹lim with circles of latitude and meridians each separated by 5°,
Tina would see a warped lattice as in Figs. 12 and 13. In
sin2 ␹ p − A2 contrast to the visualization of Scott and van Driel15 we use
⬘ 兲 = lim cos ␹⬘ =
cos共␹lim . 共54兲
␥→⬁ sin2 ␹ p + A2 the 4␲ representation where the azimuth angle ␸ and the
zenith distance ␽ are plotted like Cartesian coordinates to
Note that two objects p1 and p2 with r p1 = r p2 and ␹ p2 = ␲ show the full sky. The disadvantage of this representation is
− ␹ p1 will “freeze” at the same observation angle ␹lim⬘ . the distortion at the nodes ␽ = 0 and ␽ = ␲. Here, the direction
As we have seen, the apparent position of an object de- of motion corresponds to the center of the representation
pends considerably on its position 共r p , ␹ p兲 and on the accel- 共␽ = ␲ / 2 , ␸ = 0兲.
Because of aberration, Eq. 共39兲 shows that the nodes of
the stellar sphere move together according to Tina’s velocity.
Their angular separation ⌬␽⬘ follows from Eq. 共39a兲,

Fig. 11. The observation angle ␹⬘ is plotted versus the velocity ␤ for A
= c2 / 共␣r p兲 ⬇ 0.0968. Note that even objects that are actually behind the ob- Fig. 13. Stellar sky at ␤ = 0.9 in the 4␲ representation. Because of aberra-
server 共␹ p ⬎ ␲ / 2兲 might apparently “freeze” in front of the observer. tion, the nodes of the stellar sphere move together.

366 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008 Müller, King, and Adis 366
Fig. 14. Lines of constant redshift z at velocity ␤ = 0.5 in the 4␲ represen-
tation. From inside to outside: z = −0.4 to z = 0.6, step 0.2; the bold line
marks z = 0.
⌬␽⬘ = ␲ − 2 arccos 冑1 − ␤2 . 共55兲
Because angular distance becomes smaller in the direction of
motion, an object seems to be farther away in comparison to
its real distance. In contrast, objects in the opposite direction
Fig. 16. The observation angle ␹⬘ is plotted versus the velocity ␤. The solid
lines are lines of constant Doppler shift z according to Eq. 共59兲; the dashed
lines represent the aberration of the angle ␹ 共see Eq. 共60兲兲.
Another interesting detail is shown in Fig. 16, where the
observed angle ␹⬘ is plotted versus the velocity ␤ according
to the Doppler shift,

冉 冊
seem to grow.
Besides the mere geometrical aspects, Figs. 14 and 15 1 − 共z + 1兲冑1 − ␤2
␹⬘ = arccos , 共59兲
show lines of constant Doppler shift at different velocities. ␤
As long as the velocity ␤ = 0, there is no Doppler shift, but
for ␤ ⬎ 0 light is Doppler shifted following Eq. 共38兲, and aberration

z = ␥共1 − ␤ cos ␹⬘兲 − 1, 共56兲 ␹⬘ = arccos 冉 cos ␹ + ␤

1 + ␤ cos ␹
.冊 共60兲
where ␹⬘ is the angle between the direction of motion and
the incoming light ray. For ␤ = 0 there is no Doppler shift and no aberration.
Zero Doppler shift occurs for ␤ ⬎ 0 at an angle ␹0⬘ with Objects in front of the observer, ␹ ⬍ ␲ / 2, will always be
blueshifted and seem to be in front of the observer, ␹⬘
1 − 冑1 − ␤2 ⬍ ␲ / 2. But, for objects in back of the observer, ␹ ⬎ ␲ / 2,
cos ␹0⬘ = . 共57兲
␤ whether they apparently are in front or not depends on the
velocity. An object at an angle ␹ ⬎ ␲ / 2 will turn from being
The difference ⌬z between the maximum blueshift and the redshifted to being blueshifted when ␤ is greater than
maximum redshift equals
2 cos ␹
⌬z = z共␹⬘ = ␲兲 − z共␹⬘ = 0兲 = 2␤␥ . 共58兲 ␤red-blue = − . 共61兲
1 + cos2 ␹
Thus, the faster Tina moves, the more of the sky is red-
shifted. Only a small portion of the sky in the direction of
motion is blueshifted. B. Temperature and brightness
For visualizing the stellar sky we use the Hipparcos star
catalogue.16 We extract the Johnson B-V color and assign a
temperature T = TB-V to each star by the empirical law17

B−V= 再 C1 log10共T兲 + C2 共log10T 艋 3.961兲

C3 log10共T兲2 + C4 log10共T兲 + C5 共log10T ⬎ 3.961兲
冎
共62兲
with constants from Table I. Equation 共62兲 is only a limited
approximation to the real temperatures of the stars, but it
simplifies the following calculations.
We also need the bolometric correction 共BC兲18 to trans-
form from the visual M V to the bolometric magnitude M bol,
M bol = M V + BC, 共63兲
with
Fig. 15. Lines of constant redshift z at velocity ␤ = 0.9 in the 4␲ represen-
tation. From inside to outside: z = −0.4 to z = 2.0, step 0.2; the bold line BC = C6t̃ 4 + C7t̃ 3 + C8t̃ 2 + C9t̃ + C10 , 共64兲
marks z = 0. Note that most of the sky is redshifted even for directions ␹⬘
⬍ ␲ / 2. where t̃ = log10共T兲 − 4.

367 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008 Müller, King, and Adis 367
 Table I. The coefficients Ci for Eqs. 共62兲 and 共64兲 are taken from Ref. 17.

Coefficient Value Coefficient Value

C1 −3.684 C6 −8.499
C2 14.551 C7 13.421
C3 0.344 C8 −8.131
C4 −3.402 C9 −3.901
C5 8.037 C10 −0.438

Instead of the actual spectrum of each star we use a Planck L

spectrum at temperature T = TB−V with spectral intensity M bol − M bol,䉺 = − 2.5 log10
L䉺
2h␯3 1
I␯ = , 共65兲 R T
c2 eh␯/共kBT兲 − 1 = − 5 log10 − 10 log10 , 共69兲
R䉺 T䉺
where h is Planck’s constant and kB is Boltzmann’s
constant.20 If the spectral intensity is expressed in terms of where we approximated the spectrum of the sun by a Planck
wavelength, we obtain from I␯d␯ = −I␭d␭ and c = ␭␯ the ex- spectrum at temperature T䉺.22 Because the absolute bolom-
pression etric magnitude M bol is defined as the brightness of a star at
a distance of 10 pc, the apparent bolometric magnitude mbol
2hc2 1 is given by
I␭ = . 共66兲
␭ e
5 hc/共␭kBT兲
−1
r
The typical shape of a Planck spectrum is shown in Fig. 17, mbol − M bol = 5 log10 , 共70兲
10pc
where the wavelength ␭max of the maximum of the intensity
follows from Wien’s displacement law where r is the distance between the star and the observer.19
From Tina’s point of view, we have to transform the
␭max T = b, 共67兲
Planck spectrum of a star into her current rest frame accord-
with Wien’s displacement constant b = 2.8978⫻ 10 km. −3 ing to Eq. 共68兲 resulting in a different absolute bolometric
We use a Planck spectrum here, because it simply trans- ⬘ ,
magnitude M bol
forms with the Doppler factor according to the relativistic
Liouville theorem.21 From I␯ / ␯3 = constant together with Eqs. R⬘ T⬘
⬘ − M bol,䉺 = − 5 log10
M bol − 10 log10 . 共71兲
共65兲 and 共38兲 the temperature transforms as R䉺 T䉺
Tstar ␯ If we subtract Eq. 共69兲 from Eq. 共71兲, the transformation
= = z + 1. 共68兲
⬘
Tstar ␯⬘ between the absolute bolometric magnitudes reads
Thus, a blueshifted star with −1 ⬍ z ⬍ 0 seems to be hotter R⬘
than it really is compared to its own rest frame. ⬘ − M bol = − 5 log10
M bol − 10 log10共z + 1兲, 共72兲
R
The luminosity L of an isotropically radiating black body
of radius R is given by L = 4␲␴R2T4 with the Stefan– where we have used Eq. 共68兲 for the transformation of the
Boltzmann constant ␴ ⬇ 5.67⫻ 10−8 Wm−2 K−4.19 Thus, the temperatures. Because we are interested in the apparent mag-
absolute bolometric magnitude M bol of an object with a nitudes, we substitute the absolute magnitudes by means of
Planck spectrum at temperature T is Eq. 共70兲,

Fig. 17. Planck spectrum at temperature T = 6000 K with a maximum at ␭max ⬇ 0.483 ␮m.

368 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008 Müller, King, and Adis 368
Fig. 18. The stellar sky marked by some constellations as seen at rest. In the
4␲ representation the right ascension ␣ is plotted on the abscissa and the
declination ␦ is plotted on the ordinate. Abbreviations: 共Aql兲 Aquila, 共Cas兲
Cassiopeia, 共Crt兲 Crater, 共Cru兲 Crux, 共Cyg兲 Cygnus, 共Her兲 Hercules, 共Leo兲
Leo, 共Ori兲 Orion, 共Peg兲 Pegasus, 共UMi兲 Ursa Minor.
Fig. 19. The stellar sky as seen by an observer passing the Earth with 50%
of the speed of light. The distortion of the constellation Southern Cross
共Cru兲 is due to the 4␲ projection and the aberration effect 共see Fig. 12兲.

bolometric magnitudes mbol for ␤ = 0.5 and ␤ = 0.9. The ap-

rR⬘ parent distance d⬘ follows from the inverse form of Eq. 共41兲,
⬘ − mbol = − 5 log10
mbol
r ⬘R
+ 10 log10共z + 1兲.

As long as we approximate the stars as nearly point-like

共73兲
冋
d⬘ = d ␥2 − ␤2
1 − cos2 ␦ cos2 ␣
共1 + ␤ cos ␦ cos ␣兲2
册 −1/2
, 共76兲

sources which are very far from Tina’s current position, we where the distance d, measured in parsec, is related to the
can write ⌬␹ = 2R / r, where ⌬␹ is the angular diameter of the trigonometric parallax ␲ via d = 1 / ␲. The redshift factor z is
star. With the apparent angular diameter ⌬␹⬘ as seen from given by the inverse of Eq. 共37兲,
Tina’s reference frame, we obtain
␻ 1
⌬␹⬘ z= −1= − 1, 共77兲
⬘ − mbol = − 5 log10
mbol + 10 log10共z + 1兲. 共74兲 ␻⬘ ␥共1 + ␤ cos ␦ cos ␣兲
⌬␹
and determines the temperature T⬘ = T / 共z + 1兲.
Here, the quotient ⌬␹⬘ / ⌬␹ can be replaced by the derivative Because Cassiopeia and Orion are in front of the observer,
of the aberration formula 共60兲, ␹Cas ⬇ 1.07 and ␹Ori ⬇ 1.47, they will always be blueshifted
d␹⬘ 1 共see Fig. 16兲. But the Southern Cross 共Cru兲, ␹Cru ⬇ 2.14, will
= . 共75兲 be redshifted until the observer reaches the velocity
d␹ ␥共1 + ␤ cos ␹兲 ␤red-blue ⬇ 0.83, which follows from Eq. 共61兲. At ␤ = ␤red-blue
As expected, the transformation of the apparent magnitude, Cru has apparently already changed from the back side to the
Eq. 共74兲, is composed of an aberration and a Doppler shift front side of the observer. This change follows from Eq. 共60兲
component. and occurs at
If we cannot approximate the stars to be nearly point-like, ␤␹⬘=␲/2 = − cos ␹Cru ⬇ 0.53. 共78兲
we have to take their expansion into account. For more in-
formation on what an expanded star would look like, we We have written the interactive Java applet
refer the reader to Ref. 23. Because we cannot extract the RelSkyApplet11 for visualizing the stellar sky at different
size of a star from the Hipparcos catalogue, we neglect its velocities as seen from the position of the Earth. The applet
expansion. clearly demonstrates the strong geometrical distortion as well
as the direction dependent Doppler shift for velocities closed
C. Constellations to the speed of light. A short introduction is given in

We have discussed how aberration and Doppler shift de-

termine the view of the stellar sky as seen by a relativistic
observer. Figures 18–20 show the stellar sky of an observer
from the position of the Earth but at different velocities in
the direction right ascension ␣ = 0 and declination ␦ = 0. Note
that the declination ␦ corresponds to ␲ / 2 − ␽ in comparison
to Fig. 7. The stars are connected by lines showing the
constellations.24
As explained in Sec. VI B, the aberration effect lets the
constellations apparently shrink in the direction of motion,
while the ones that are behind the observer, such as Leo,
Virgo 共Vir兲 and Crater 共Crt兲, seem to grow.
In Table II we list the stars of the constellations Orion
共Ori兲, Cassiopeia 共Cas兲, and Southern Cross 共Cru兲 with their
rest frame data. For these constellations we give in Tables III Fig. 20. The stellar sky as seen by an observer passing the Earth with 90%
and IV the apparent distance d⬘, the temperature T⬘, and the of the speed of light.

369 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008 Müller, King, and Adis 369
 Table II. Star data of some constellations from Figs. 18–20. ␣: right ascension, ␦: declination, ␲: trigonometric
parallax 共milliarcsec兲, B-V: Johnson B-V color, T: temperature 共Kelvin兲 from Eq. 共62兲, HIP: Hipparcos number.

Abbr. ␣ ␦ ␲ B-V T HIP

␣ Ori 1.5497 0.1293 7.63 1.50 3488 27989

␤ Ori 1.3724 −0.1431 4.22 −0.03 9077 24436
␥ Ori 1.4187 0.1108 13.42 −0.22 19245 25336
␦ Ori 1.4487 0.0052 3.56 −0.17 15279 25930
⑀ Ori 1.4670 −0.0210 2.43 −0.18 15903 26311
␨ Ori 1.4868 −0.0339 3.99 −0.20 17038 26727
␬ Ori 1.5174 −0.1688 4.52 −0.17 14819 27366

␣ Cas 0.1767 0.9868 14.27 1.17 4287 3179

␤ Cas 0.0400 1.0324 59.89 0.38 7025 746
␥ Cas 0.2474 1.0597 5.32 −0.05 9295 4427
␦ Cas 0.3744 1.0513 32.81 0.16 8060 6686
⑀ Cas 0.4991 1.1113 7.38 −0.15 13732 8886

␣ Cru −3.0255 −1.1013 10.17 −0.24 21259 60718

␤ Cru −2.9334 −1.0418 9.25 −0.24 20696 62434
␥ Cru −3.0056 −0.9968 37.09 1.60 3277 61084
␦ Cru −3.0755 −1.0254 8.96 −0.19 16569 59747

Appendix A while a more detailed discussion as well as sev-
eral examples can be found in the help menu of the applet.
VIII. A TRIP TO THE END OF THE UNIVERSE
As a first trip Tina goes on an expedition to the center of
our galaxy 共SgrA*兲 8 kpc away. From Table V we see that
her maximum speed is only 2.8 ppb 共parts per billion兲 below
the speed of light.
With this tremendous velocity even the extremely cold
microwave background radiation25 at Tcmb = 2.725 K comes
into the visual regime. But in contrast to one’s expectation,
the Doppler shifted background radiation will not fill the
whole sky. By Wien’s displacement law 共67兲 an object must
have a temperature of about T780 ⬇ 3700 K to emit its maxi-
mum radiation in the red light regime with ␭ = 780 nm. To
shift the background temperature to T780 the observer has to
move with a velocity very close to the speed of light, ␤780
⬇ 1 – 1.07⫻ 10−6. From Eq. 共57兲 it follows that the back-
ground radiation is blueshifted only in the small region of
␹0⬘ ⬍ 3°. The rest of the sky is redshifted with maximum z
⬇ 1767 at ␹⬘ = ␲.
In contrast, the redshift brings the x-ray and ␥-ray sky
down to the visual regime. For x rays with wavelengths of
␭ ⬇ 10−10 m Tina has to fly with ␤
1 – 6.6⫻ 10−7. The much
higher velocity ␤
1 – 6.6⫻ 10−12 is needed for ␥ rays with
␭ ⬇ 10−12 m.26 Just before the ␥-ray sky, the hydrogen 21 cm

Table IV. The apparent visual magnitude mV of the stars of Table II have
␤
Table III. The stars of Table II have distance d⬘ 共parsec兲 and temperature T⬘ bolometric magnitudes mbol at velocities ␤ = 0, ␤ = 0.5, and ␤ = 0.9 in the
共Kelvin兲 at velocities ␤ = 0.5 and ␤ = 0.9 in the direction ␣ = ␦ = 0. direction ␣ = ␦ = 0.

␤=0 ␤=0.5 ␤=0.9

Abbr. d␤=0 d␤⬘ =0.5 T␤⬘ =0.5 d␤⬘ =0.9 T␤⬘ =0.9 Abbr. mV BC mbol mbol mbol

␣ Ori 131.06 125.61 4070 61.90 8153 ␣ Ori 0.45 −2.01 −1.56 −2.32 −6.88
␤ Ori 236.97 222.56 11503 109.31 24479 ␤ Ori 0.18 −0.29 −0.11 −1.27 −6.10
␥ Ori 74.52 70.35 23895 34.56 50135 ␥ Ori 1.64 −1.95 −0.31 −1.38 −6.14
␦ Ori 280.90 266.08 18717 130.76 38894 ␦ Ori 2.25 −1.36 0.89 −0.11 −4.83
⑀ Ori 411.52 390.64 19315 192.02 39886 ⑀ Ori 1.69 −1.46 0.23 −0.73 −5.42
␨ Ori 250.63 238.46 20499 117.27 42038 ␨ Ori 1.74 −1.63 0.11 −0.81 −5.47
␬ Ori 221.24 211.27 17562 103.98 35608 ␬ Ori 2.07 −1.28 0.79 −0.05 −4.66

␣ Cas 70.08 63.34 6294 31.32 14641 ␣ Cas 2.24 −0.93 1.31 −0.57 −5.77
␤ Cas 16.70 15.13 10190 7.48 23547 ␤ Cas 2.28 −0.08 2.20 0.37 −4.80
␥ Cas 187.97 171.11 13277 84.45 30424 ␥ Cas 2.15 −0.32 1.83 0.07 −5.06
␦ Cas 30.48 27.78 11457 13.71 26181 ␦ Cas 2.66 −0.16 2.50 0.78 −4.35
⑀ Cas 135.50 14.49 18943 61.30 42544 ⑀ Cas 3.35 −1.10 2.25 0.67 −4.38

␣ Cru 98.33 98.26 19032 53.01 29046 ␣ Cru 0.77 −2.21 −1.44 −0.97 −4.14
␤ Cru 108.11 108.11 17998 59.70 26380 ␤ Cru 1.25 −2.14 −0.89 −0.29 −3.24
␥ Cru 26.96 26.95 2766 15.31 3878 ␥ Cru 1.59 −2.45 −0.86 −0.13 −2.82
␦ Cru 111.61 111.60 14181 62.54 20304 ␦ Cru 2.79 −1.56 1.23 1.90 −0.91

370 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008 Müller, King, and Adis 370
Table V. Distance from Earth, maximum speed and proper time of both twins for several stellar destinations. In the solar system we will reach only a few
percent of the speed of light. Thus, time dilation can be neglected. However, in the neighborhood of the solar system time dilation is crucial. The “END” of
the universe represents the maximum distance of about 13.7 billion light years that astronomers are able to observe.

Object Distance ␤max = vmax / c Tina’s time 2T⬘ Eric’s time 2T

Mars 0.524 AU⬇ 4.4 lm 0.003 49h45min13.4sec 49h45min13.7sec

Saturn 8.53 AU⬇ 1.2 lh 0.012 8d8h44min21sec 8d8h44min38sec
Pluto 38.81 AU⬇ 5.4 lh 0.025 17d20h10min 17d20h13min
␣ Cen C 共HIP 70890兲 1.29 pc 0.949 3.54a 5.84a
Vega 共HIP 91262兲 7.76 pc 0.9975 6.46a 27.17a
SgrA* 8 kpc 1 − 2.8⫻ 10−9 19.74a 26ka
LMC 50 kpc 1 − 7.1⫻ 10−11 23.30a 163ka
M81 2 Mpc 1 − 4.4⫻ 10−14 30.45a 6.5Ma
END 13.7⫻ 109 ly 1 − 1.0⫻ 10−20 45.27a 13.7Ga

line of interstellar gas, resulting from a transition between
two hyperfine structure energy levels in the hydrogen atom,
will become visible at a velocity of ␤ ⬇ 1 – 2.7⫻ 10−11.
A wavelength ␭ will be seen at wavelength ␭⬘ for a veloc-
ity
quite strong, Tina would see only a very small, bright dot in
the direction of motion. In contrast, most of the sky would be
very cold and very dark. Navigation by stars would be com-
pletely impossible.
One might ask if she would be able to see the evolution of

␤= 冏 冏
␭ ⬘2 − ␭ 2
␭ ⬘2 + ␭ 2
. 共79兲
the universe.37

ACKNOWLEDGMENTS
Thus, the minimum and maximum wavelengths at velocity ␤ The authors thank Professor Hanns Ruder for the idea of
that are transformed into the visual regime follow from this work and Professor Jörg Frauendiener for many discus-

␭min = 380 nm 冑 1−␤

1+␤
and ␭max = 780 nm 冑 1+␤
1−␤
.
What happens with the Milky Way in the meantime?
While Tina’s journey to the center of our galaxy lasts only
20 years with respect to her proper time, the Milky Way ages
about 26,000 years. In that time, our sun with rotational
velocity27 v ⬇ 220 km/ s ⬇ 7.34⫻ 10−4 ly/ y 共light years per
year兲 will cover a distance of about 19 ly. To fully describe
the galactic evolution, we need the position as well as the
proper motion of each star in the galaxy. The Hipparcos cata-
logue, which consists of about 118,000 stars, with most of
them at a distance of about 100 pc, delivers the best posi-
tional star reference 共1 milli-arc-sec兲. From the GAIA29 mis-
sion we expect an accuracy in positional astrometry of about
20 ␮as. Thus, all stars in our galaxy up to the 20th magni-
tude should be included.
As in the title of this article, Tina could also go on a trip to
the “end of the universe” within her lifetime. Here, the “end”
means the maximum distance of about 13.7 billion light
years that astronomers are able to observe. For these specu-
lations we neglect the expansion of the universe, but do the
calculations in the flat Minkowskian spacetime instead of a
Robertson–Walker5 spacetime.
When Tina reaches her maximum velocity ␤ = 1 − ⑀ with
⑀ = 1.0⫻ 10−20 in the middle of her journey, the relativistic
effects are extremely dramatic. A ␥ factor of about 7.07
⫻ 109 results in a different time rate: ⌬t⬘ = 1 s corresponds to
⌬t = 224a! Thus, in roughly 12 days with respect to Tina’s
proper time our Sun would have finished one revolution
around the center of our galaxy. Zero Doppler shift occurs at
an angle ␹0⬘ ⬇ 冑4 8⑀ ⬇ 3.5 as. The maximum blueshift in the
direction of motion is z共␹⬘ = 0兲 ⬇ 冑⑀ / 2 ⬇ −1 + 7.07⫻ 10−11,
whereas the maximum redshift is z共␹⬘ = ␲兲 ⬇ 冑2 / ⑀ ⬇ 1.4
⫻ 1010. Because the transition from redshift to blueshift is
sions and for carefully reading the manuscript. Thanks also
to Professor Jeff Rabin for the suggestion of the book Tau
Zero 共see Ref. 30兲. This work was supported by the Deutsche
Forschungsgesellschaft 共DFG兲, SFB 382, Teilprojekt D4.
APPENDIX A: JAVA APPLETS
The Java applets used to generate the diagrams in this
article are available from the Electronic Physics Auxiliary
Publication Service 共EPAPS兲.11 The Java applet TwinAp-
plet generates the diagrams of Sec. V. There are three input
parameters, two of them are obligatory: the acceleration of
the traveling twin Tina in terms of Earth’s gravity g and
either her travel time or the maximum reachable distance.
With these parameters four plots are generated. They are
grouped in two panels: one showing the distance and the
velocity of Tina and the other showing the time signals
which the twins receive of each other.
The acceleration of Tina can be modified between 10−6 g
and 103 g. The travel time of one acceleration phase can be
entered either into a text field or with a slider. While moving
the time slider, we can observe the immediate change of the
plots and the maximum reachable distance. The maximum
time is limited to the time that the twin needs to reach the
end of the universe. Alternatively we can choose the travel
distance and hence calculate the time needed. It is possible to
save the plots as PNG images or the calculated data points in
a text file.
Because we have to deal with velocities very close to the
speed of light, the usual double-precision floating-point num-
bers will not suffice, compare Appendix B. Thus, we use a
series expansion for the equations depending on the velocity
␤.
The Java applet RelSkyApplet can visualize either the
stellar sky, the star constellations, or the cosmic microwave
background, each with different velocities as seen by an ob-

371 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008 Müller, King, and Adis 371
server passing the Earth. The main input parameters for all
the views are the velocity and the line of sight 共right ascen-
sion and declination兲. For the stellar sky there are some ad-
ditional parameters such as the maximum displayed magni-
tude and the magnitude scale. For the stellar sky and the star
constellations it is possible to increase the velocity stepwise
共without changing position兲 to observe the increase of aber-
ration and Doppler shift. By varying the velocity, the user
obtains a better understanding of how relativistic aberration
and the Doppler shift affects the visualization.
It is possible to choose between two views for the visual-
izations: a 4␲-representation or an hemispherical representa-
tion. In both cases the line of sight corresponds to the center
of the representation and can be set manually. The stellar sky
can also be seen from any desired point in our galaxy 共either
in equatorial coordinates or in galactic coordinates兲. By mov- Fig. 21. CIE 1931 2-deg color matching functions x̄␭ , ȳ ␭ , z̄␭.
ing the mouse over the sky view, the sky coordinates of the
actual mouse position are displayed in the lower left corner
of the applet. The longitude and latitude of the sky can also impreciseness propagates. Hence, a straightforward numeri-
be displayed as a grid, which deforms according to the in- cal calculation of the ␥-factor results in
creasing velocity.
The stars are realized as small disks whose colors are cal-
␥ˆ = 1.101111101110100100000112 ⫻ 27
culated according to Appendix C and the star temperatures. It ⬇ 223.45512310 . 共B2兲
is possible to change the color mode in order to examine a
particular temperature range. In manual mode the colors that The series expansion of the ␥ factor is
are displayed correspond to an arbitrary color scale. The size 1 1
␥=
冑1 − 共1 − ␧兲2 ⬇ 冑2␧ + O共␧ 兲, 共B3兲
1/2
of the disk represents the star’s apparent bolometric magni-
tude mbol.
By clicking on a star its Hipparcos identifier and its com- and the evaluation of ␧ = 10−5 gives
mon name 共if available兲 are displayed. A click on Get Info
provides further information about the star: the complete ␥ˆ = 1.101111110011011010101112 ⫻ 27
Hipparcos data set, and some of its real values compared to ⬇ 223.60679610 . 共B4兲
its apparent values.
rel
The star constellations view is suitable for observing the From the relative errors E for the straightforward calcula-
distortion due to the aberration effect. Remember that the tion and the series expansion,
distortion of some constellations in the border area of the rel
Estraight ⬇ 6.8 ⫻ 10−4 rel
and Eseries ⬇ 2.5 ⫻ 10−6 , 共B5兲
4␲-representation might be caused by the representation it-
self. we conclude that it is important to replace an equation by its
The basic data of the cosmic microwave background was series expansion for high velocities.
obtained from the Wilkinson Microwave Anisotropy Probe
共WMAP兲.31 To speed up the applets’ execution, the original
APPENDIX C: FROM SPECTRUM TO COLOR
data was downgraded to a 4␲ representation with an image
resolution of 720⫻ 360 pixels. By selecting the microwave The human visual perception of wavelengths lies in the
mode the color scale is automatically set to a temperature range between 380 and 780 nm. There are three types of
range of 2.5– 2.9 K 共which corresponds to microwave wave- cones in the retina which are sensitive to red, green and blue
lengths兲. Due to the Doppler effect, it is also possible to light. Hence, any visible color can be composed of three
observe the microwave background even in the visual regime primary colors.33
共visual color mode兲 starting at a velocity of about 0.999c. In 1931 the Commission Internationale de l’Éclairage
共CIE兲 defined three primary colors X, Y, and Z with the
APPENDIX B: NUMERICAL CALCULATIONS corresponding color matching functions34 x̄共␭兲, ȳ共␭兲 and z̄共␭兲
共see Fig. 21兲.
As we can see from Table V, we have to consider veloci-
Any color C = XX + YY + ZZ can be composed of these pri-
ties ␤ that are very close to the speed of light. But computers
mary colors,35 where the components X, Y and Z follow from
can handle only a limited number of digits. In general,
the spectral intensity distribution I共␭兲 by convolution with
double-precision floating-point numbers with a 52 bit man-
tissa are used for precise calculations.32 For clarity we will the color matching functions, for example,
describe here the velocity ␤ = 1 − 10−5 as a single-precision
floating-point number with a 23 bit mantissa X=k 冕 I共␭兲x̄共␭兲d␭. 共C1兲

␤ˆ = 1.111111111111111010110002 ⫻ 2−1 Because we are interested only in the chromaticity values x,

⬇ 0.999989986410 , 共B1兲 y, and z with x = X / 共X + Y + Z兲 and y = Y / 共X + Y + Z兲 and z = 1
− x − y, the constant k cancels out. To find the RGB values for
where ␤ˆ is expressed either in normalized binary form 共sub- a specific device, we need its primary chromaticity values
script 2兲 or in decimal form 共subscript 10兲. This machine x共r,g,b,w兲 and y 共r,g,b,w兲 for red, green, blue and the white

372 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008 Müller, King, and Adis 372
point.36 After the transformation from xyz to rgb, we normal-
ize the rgb values according to their maximum,
共r,g,b兲
共r,g,b兲 哫 , 共C2兲
max共r,g,b兲
which has the effect that all colors have their largest possible
brightness.
a兲
Author to whom correspondence should be addressed. Electronic mail:
tmueller@tat.physik.uni-tuebingen.de
1
R. D’Inverno, Introducing Einstein’s Relativity 共Clarendon Press, Oxford,
1992兲.
2
C. E. Dolby and S. F. Gull, “On radar time and the twin ‘paradox,’” Am.
J. Phys. 69, 1257–1261 共2001兲.
3
S. P. Boughn, “The case of the identically accelerated twins,” Am. J.
Phys. 57, 791–793 共1988兲.
4
R. P. Gruber and Richard H. Price, “Zero time dilation in an accelerating
rocket,” Am. J. Phys. 65, 979–980 共1997兲.
5
W. Rindler, Relativity: Special, General, and Cosmology 共Oxford Univer-
sity Press, New York, 2001兲.
6
R. U. Sexl, and H. K. Urbantke, Relativity, Groups, Particles: Special
Relativity and Relativistic Symmetry in Field and Particle Physics
共Springer, Wien/New York, 2000兲.
7
R. Perrin, “Twin paradox: A complete treatment from the point of view of
each twin,” Am. J. Phys. 47, 317–319 共1979兲.
8
A. P. French, Special Relativity 共W. W. Nelson & Co, London, 1968兲.
9
R. H. Good, “Uniformly accelerated reference frame and twin paradox,”
Am. J. Phys. 50, 232–238 共1982兲.
10
H. und M. Ruder, Die Spezielle Relativitätstheorie 共Vieweg Studium,
Braunschweig/Wiesbaden, 1993兲.
11
See EPAPS Document No. E-AJPIAS-76-002803 for the Java applets,
RelSkyApplet and TwinApplet, and accompanying help manuals.
This document can be reached through a direct link in the online article’s
HTML reference section or via the EPAPS homepage 共http://
www.aip.org/pubservs/epaps.html兲.
12
One light year 共1 ly兲 equals the distance which is covered by light with
velocity c = 299, 792, 458 m / s in one year.
13
The negative acceleration of Tina away from the Sun equals the situation
where Tina has positive acceleration but looks contrary to her direction of
motion.
14
The limit can be calculated using l’Hospital’s rule.
15
G. D. Scott and H. J. van Driel, “Geometrical appearances at relativistic
speeds,” Am. J. Phys. 38, 971–977 共1970兲.
16
The Hipparcos star catalogue consists of about 118,000 stars most of
which are at a distance of roughly 100 pc. The data we are interested in
are right ascension 共H3兲, declination 共H4兲, magnitude in Johnson V 共H5兲,
trigonometric parallax 共H11兲, Johnson B-V color 共H37兲, and Henry
Draper 共HD兲 catalogue number 共H71兲. The digital catalogue I/239 can be
found at 具cdsarc.u-strasbg.fr/viz-bin/Cat?I/239典 1997HIP…C……0E-
European Space Agency SP-1200.
17
B. C. Reed, “The composite observational-theoretical HR diagram,” J. R.
373 Am. J. Phys., Vol. 76, Nos. 4 & 5, April/May 2008
Astron. Soc. Can. 92, 36–37 共1998兲.
18
The bolometric magnitude of a star is defined by the flux density inte-
grated over all wavelengths. However, a real telescope can only measure
the flux density within a small region of wavelengths. The bolometric
correction compensates the difference between the bolometric and the
visual magnitude 共Ref. 19兲.
19
Fundamental Astronomy, edited by Hannu Karttunen, Pekka Kröger,
Heikki Oja, Markku Poutanen, and Karl J. Donner 共Springer-Verlag,
Heidelberg, 2003兲.
20
We take the physical constants from the National Institute of Standards
and Technology, 具physics.nist.gov/cuu/Constants典.
21
R. W. Lindquist, “Relativistic transport theory,” Ann. Phys. 37, 487–518
共1966兲.
22
The absolute bolometric magnitude of the sun is M bol,䉺 = 4.83 and the
effective temperature is T䉺 ⬇ 5778 K, see also 具nssdc.gsfc.nasa.gov/
planetary/factsheet/sunfact.html典.
23
U. Kraus, “Brightness and color of rapidly moving objects: The visual
appearance of a large sphere revisited,” Am. J. Phys. 68, 56–60 共2000兲.
24
The constellations were obtained from the free planetarium software Stel-
larium, 具www.stellarium.org典.
25
J. C. Mather et al., “Calibrator design for the COBE far-infrared absolute
spectrophotometer 共FIRAS兲,” Astrophys. J. 512, 511–520 共1999兲.
26
For more information on x rays and ␥ rays we refer the reader to the
following missions. X-ray: ROSAT, 具wave.xray.mpe.mpg.de/rosat典, Chan-
dra, 具chandra.harvard.edu典, XMM-Newton, 具sci.esa.int典; ␥-ray: INTE-
GRAL 具www.esa.int/esaMI/Integral/典.
27
The rotational velocity of a star in a galaxy does not follow Newton’s
law, but can be read from the rotation curve of the galaxy. See Ref. 28 for
the outer rotation curve of the Milky Way.
28
J. Brand and L. Blitz, “The velocity field of the outer galaxy,” Astron.
Astrophys. 275, 67–90 共1993兲.
29
The goal of the GAIA mission is to collect high-precision astrometric
data for the brightest one billion objects. A detailed description of the
ESA Science missions is available at 具www.esa.int/esaSC/典.
30
P. Anderson, Tau Zero 共Orion Publishing Group, London, 2006兲.
31
Data are taken from 具map.gsfc.nasa.gov典.
32
Single-precision floating-point numbers are stored in a 32 bit word,
whereas double-precision ones are stored in a 64 bit word. The word
itself is composed of a 23共52兲 bit mantissa, a 8共11兲 bit exponent, and one
bit for the sign. More information can be found in the IEEE standard for
binary floating-point arithmetic for microprocessor systems 共ANSI/IEEE
Std 754–1985兲.
33
This fact is known as the tristimulus theory. For more information see, for
example, Foley 共Ref. 35兲.
34
Color matching functions can be found at the Institute of Ophthalmology,
具cvrl.ioo.ucl.ac.uk/basicindex.htm典.
35
J. D. Foley, A. Van Dam, S. K. Feiner, and J. F. Hughes, Computer
Graphics: Principles and Practice 共Addison–Wesley, 1991兲.
36
We use the color rendering of spectra by John Walker, 具www.fourmila-
b.ch典. The white point is the set of chromaticity values 共xw , y w兲 that serve
to define the color white.
37
An excellent science fiction book on this topic is given in Ref. 30 which
might be a useful basis for discussion.
Müller, King, and Adis 373