Tau Manifesto

Download as pdf or txt
Download as pdf or txt
You are on page 1of 29

The Tau Manifesto

Michael Hartl
Tau Day, 2010
updated Pi Day, 2012
1 The circle constant
The Tau Manifesto is dedicated to one of the most important numbers in mathematics, perhaps the most
important: the circle constant relating the circumference of a circle to its linear dimension. For millennia, the
circle has been considered the most perfect of shapes, and the circle constant captures the geometry of the
circle in a single number. Of course, the traditional choice for the circle constant is but, as mathematician
Bob Palais notes in his delightful article Is Wrong!
1
, is wrong. Its time to set things right.
1.1 An immodest proposal
We begin repairing the damage wrought by by rst understanding the notorious number itself. The tradi-
tional denition for the circle constant sets (pi) equal to the ratio of a circles circumference to its diameter:
2

C
D
= 3.14159265 . . .
The number has many remarkable propertiesamong other things, it is irrational and indeed transcenden-
taland its presence in mathematical formulas is widespread.
It should be obvious that is not wrong in the sense of being factually incorrect; the number is
perfectly well-dened, and it has all the properties normally ascribed to it by mathematicians. When we say
that is wrong, we mean that is a confusing and unnatural choice for the circle constant. In particular,
since a circle is dened as the set of points a xed distancethe radiusfrom a given point, a more natural
denition for the circle constant uses r in place of D:
circle constant
C
r
.
Because the diameter of a circle is twice its radius, this number is numerically equal to 2. Like , it
is transcendental and hence irrational, and (as well see in Section 2) its use in mathematics is similarly
widespread.
In Is Wrong!, Bob Palais argues persuasively in favor of the second of these two denitions for the
circle constant, and in my view he deserves principal credit for identifying this issue and bringing it to a broad
audience. He calls the true circle constant one turn, and he also introduces a new symbol to represent it
(Figure 1). As well see, the description is prescient, but unfortunately the symbol is rather strange, and (as
discussed in Section 4) it seems unlikely to gain wide adoption.
1
Palais, Robert. Is Wrong!, The Mathematical Intelligencer, Volume 23, Number 3, 2001, pp. 78. Many of the arguments in
The Tau Manifesto are based on or are inspired by Is Wrong!. It is available online at http://bit.ly/pi-is-wrong.
2
The symbol means is dened as.
1
Figure 1: The strange symbol for the circle constant from Is Wrong!.
Figure 2: The Google logo on March 14 (3/14), 2010 (Pi Day).
The Tau Manifesto is dedicated to the proposition that the proper response to is wrong is No, really.
And the true circle constant deserves a proper name. As you may have guessed by now, The Tau Manifesto
proposes that this name should be the Greek letter (tau):

C
r
= 6.283185307179586 . . .
Throughout the rest of this manifesto, we will see that the number is the correct choice, and we will show
through usage (Section 2 and Section 3) and by direct argumentation (Section 4) that the letter is a natural
choice as well.
1.2 A powerful enemy
Before proceeding with the demonstration that is the natural choice for the circle constant, let us rst ac-
knowledge what we are up againstfor there is a powerful conspiracy, centuries old, determined to propagate
pro- propaganda. Entire books are written extolling the virtues of . (I mean, books!) And irrational devo-
tion to has spread even to the highest levels of geekdom; for example, on Pi Day 2010 Google changed
its logo to honor (Figure 2).
Meanwhile, some people memorize dozens, hundreds, even thousands of digits of this mystical number.
What kind of sad sack memorizes even 40 digits of (Figure 3)?
3
Truly, proponents of face a mighty opponent. And yet, we have a powerful allyfor the truth is on our
side.
3
The video in Figure 3 (available at http://vimeo.com/12914981) is an excerpt from a lecture given by Dr. Sarah Greenwald, a
professor of mathematics at Appalachian State University. Dr. Greenwald uses math references from The Simpsons and Futurama to
engage her students interest and to help them get over their math anxiety. She is also the maintainer of the Futurama Math Page.
2
Figure 3: Michael Hartl proves Matt Groening wrong by reciting to 40 decimal places.
2 The number tau
We sawin Section 1.1 that the number can also be written as 2. As noted in Is Wrong!, it is therefore of
great interest to discover that the combination 2 occurs with astonishing frequency throughout mathematics.
For example, consider integrals over all space in polar coordinates:
_
2
0
_

0
f(r, ) r dr d.
The upper limit of the integration is always 2. The same factor appears in the denition of the Gaussian
(normal) distribution,
1

2
e

(x)
2
2
2
,
and again in the Fourier transform,
f(x) =
_

F(k) e
2ikx
dk
F(k) =
_

f(x) e
2ikx
dx.
It recurs in Cauchys integral formula,
f(a) =
1
2i
_

f(z)
z a
dz,
in the nth roots of unity,
z
n
= 1 z = e
2i/n
,
3
s
1
s
2
r
1
r
2

Figure 4: An angle with two concentric circles.


and in the values of the Riemann zeta function for positive even integers:
4
(2n) =

k=1
1
k
2n
=
B
n
2(2n)!
(2)
2n
. n = 1, 2, 3, . . .
These formulas are not cherry-pickedcrack open your favorite physics or mathematics text and try it your-
self. Indeed, there are many more examples, and the conclusion is clear: there is something special about 2.
To get to the bottom of this mystery, we must return to rst principles by considering the nature of circles,
and especially the nature of angles. Although its likely that much of this material will be familiar, it pays to
revisit it, for this is where the true understanding of begins.
2.1 Circles and angles
There is an intimate relationship between circles and angles, as shown in Figure 4. Since the concentric
circles in Figure 4 have different radii, the lines in the gure cut off different lengths of arc (or arc lengths),
but the angle (theta) is the same in each case. In other words, the size of the angle does not depend on the
radius of the circle used to dene the arc. The principal task of angle measurement is to create a system that
captures this radius-invariance.
4
Here Bn is the nth Bernoulli number.
4
Figure 5: Some special angles, in degrees.
Perhaps the most elementary angle system is degrees, which breaks a circle into 360 equal parts. One
result of this system is the set of special angles (familiar to students of trigonometry) shown in Figure 5.
A more fundamental system of angle measure involves a direct comparison of the arc length s with the
radius r. Although the lengths in Figure 4 differ, the arc length grows in proportion to the radius, so the ratio
of the arc length to the radius is the same in each case:
s r
s
1
r
1
=
s
2
r
2
.
This suggests the following denition of radian angle measure:

s
r
.
This denition has the required property of radius-invariance, and since both s and r have units of length,
radians are dimensionless by construction. The use of radian angle measure leads to succinct and elegant
formulas throughout mathematics; for example, the usual formula for the derivative of sin is true only when
is expressed in radians:
d
d
sin = cos . (true only when is in radians)
Naturally, the special angles in Figure 5 can be expressed in radians, and when you took high-school trigonom-
etry you probably memorized the special values shown in Figure 6. (I call this system of measure -radians
to emphasize that they are written in terms of .)
Now, a moments reection shows that the so-called special angles are just particularly simple rational
fractions of a full circle, as shown in Figure 7. This suggests revisiting the denition of radian angle measure,
rewriting the arc length s in terms of the fraction f of the full circumference C, i.e., s = fC:
=
s
r
=
fC
r
= f
_
C
r
_
f.
5
Figure 6: Some special angles, in -radians.
6
Figure 7: The special angles are fractions of a full circle.
7
0,
Figure 8: Some special angles, in radians.
Notice how naturally falls out of this analysis. If you are a believer in , I fear that the resulting diagram of
special anglesshown in Figure 8will shake your faith to its very core.
Although there are many other arguments in s favor, Figure 8 may be the most striking. Indeed, upon
comparing Figure 8 with Figure 7, I consider it decisive. We also see from Figure 8 the genius of Bob Palais
identication of the circle constant as one turn: is the radian angle measure for one turn of a circle.
Moreover, note that with there is nothing to memorize: a twelfth of a turn is /12, an eighth of a turn is
/8, and so on. Using gives us the best of both worlds by combining conceptual clarity with all the concrete
benets of radians; the abstract meaning of, say, /12 is obvious, but it is also just a number:
a twelfth of a turn =

12

6.283185
12
= 0.5235988.
Finally, by comparing Figure 6 with Figure 8, we see where those pesky factors of 2 come from: one turn
of a circle is 1, but 2. Numerically they are equal, but conceptually they are quite distinct.
2.1.1 The ramications
The unnecessary factors of 2 arising from the use of are annoying enough by themselves, but far more
serious is their tendency to cancel when divided by any even number. The absurd results, such as a half for
a quarter turn, obscure the underlying relationship between angle measure and the circle constant. To those
who maintain that it doesnt matter whether we use or when teaching trigonometry, I simply ask you
8

(cos , sin )
Figure 9: The circle functions are coordinates on the unit circle.
to view Figure 6, Figure 7, and Figure 8 through the eyes of a child. You will see that, from the perspective
of a beginner, using instead of is a pedagogical disaster.
2.2 The circle functions
Although radian angle measure provides some of the most compelling arguments for the true circle constant,
its worth comparing the virtues of and in some other contexts as well. We begin by considering the
important elementary functions sin and cos . Known as the circle functions because they give the coor-
dinates of a point on the unit circle (i.e., a circle with radius 1), sine and cosine are the fundamental functions
of trigonometry (Figure 9).
Lets examine the graphs of the circle functions to better understand their behavior.
5
Youll notice from
Figure 10 and Figure 11 that both functions are periodic with period T. As shown in Figure 10, the sine
function sin starts at zero, reaches a maximum at a quarter period, passes through zero at a half period,
reaches a minimum at three-quarters of a period, and returns to zero after one full period. Meanwhile, the
cosine function cos starts at a maximum, has a minimum at a half period, and passes through zero at one-
quarter and three-quarters of a period (Figure 11). For reference, both gures show the value of (in radians)
at each special point.
Of course, since sine and cosine both go through one full cycle during one turn of the circle, we have
T = ; i.e., the circle functions have periods equal to the circle constant. As a result, the special values of
are utterly natural: a quarter-period is /4, a half-period is /2, etc. In fact, when making Figure 10, at one
5
These graphs were produced with the help of Wolfram|Alpha.
9
sin

3T
4
T
4
T
2
T
Figure 10: Important points for sin in terms of the period T.

3T
4
T
4
T
2
T
cos
Figure 11: Important points for cos in terms of the period T.
10
point I found myself wondering about the numerical value of for the zero of the sine function. Since the
zero occurs after half a period, and since 6.28, a quick mental calculation led to the following result:

zero
=

2
3.14.
Thats right: I was astonished to discover that I had already forgotten that /2 is sometimes called .
Perhaps this even happened to you just now. Welcome to my world.
2.3 Eulers identity
I would be remiss in this manifesto not to address Eulers identity, sometimes called the most beautiful
equation in mathematics. This identity involves complex exponentiation, which is deeply connected both to
the circle functions and to the geometry of the circle itself.
Depending on the route chosen, the following equation can either be proved as a theorem or taken as a
denition; either way, it is quite remarkable:
e
i
= cos + i sin .
Known as Eulers formula (after Leonhard Euler), this equation relates an exponential with imaginary argu-
ment to the circle functions sine and cosine and to the imaginary unit i. Although justifying Eulers formula is
beyond the scope of this manifesto, its provenance is above suspicion, and its importance is beyond dispute.
Evaluating Eulers formula at = yields Eulers identity:
6
e
i
= 1.
In words, this equation makes the following fundamental observation:
The complex exponential of the circle constant is unity.
Geometrically, multiplying by e
i
corresponds to rotating a complex number by an angle in the complex
plane, which suggests a second interpretation of Eulers identity:
A rotation by one turn is 1.
Since the number 1 is the multiplicative identity, the geometric meaning of e
i
= 1 is that rotating a point in
the complex plane by one turn simply returns it to its original position.
As in the case of radian angle measure, we see how natural the association is between and one turn of a
circle. Indeed, the identication of with one turn makes Eulers identity sound almost like a tautology.
7
2.3.1 Not the most beautiful equation
Of course, the traditional form of Eulers identity is written in terms of instead of . To derive it, we start
by evaluating Eulers formula at = , which yields
e
i
= 1.
6
Here Im implicitly dening Eulers identity to be the complex exponential of the circle constant, rather than dening it to be
the complex exponential of any particular number. If we choose as the circle constant, we obtain the identity shown. As well see
momentarily, this is not the traditional form of the identity, which of course involves , but the version with is the most mathematically
meaningful statement of the identity, so I believe it deserves the name.
7
Technically, all mathematical theorems are tautologies, but lets not be so pedantic.
11
Rotation angle Eulerian identity
0 e
i0
= 1
/4 e
i/4
= i
/2 e
i/2
= 1
3/4 e
i(3/4)
= i
e
i
= 1
Table 1: Eulerian identities for half, quarter, and full rotations.
But that minus sign is so ugly that the formula is almost always rearranged immediately, giving the following
beautiful equation:
e
i
+ 1 = 0.
At this point, the expositor usually makes some grandiose statement about how Eulers identity relates 0, 1,
e, i, and sometimes called the ve most important numbers in mathematics. Its remarkable how many
people complain that Eulers identity with relates only four of those ve. Fine:
e
i
= 1 + 0.
This formula, without rearrangement, actually does relate the ve most important numbers in mathematics:
0, 1, e, i, and .
2.3.2 Eulerian identities
Since you can add zero anywhere in any equation, the introduction of 0 into the formula e
i
= 1 + 0 is a
somewhat tongue-in-cheek counterpoint to e
i
+ 1 = 0, but the identity e
i
= 1 does have a more serious
point to make. Lets see what happens when we rewrite it in terms of :
e
i/2
= 1.
Geometrically, this says that a rotation by half a turn is the same as multiplying by 1. And indeed this is the
case: under a rotation of /2 radians, the complex number z = a + ib gets mapped to a ib, which is in
fact just 1 z.
Written in terms of , we see that the original form of Eulers identity has a transparent geometric
meaning that it lacks when written in terms of . (Of course, e
i
= 1 can be interpreted as a rotation
by radians, but the near-universal rearrangement to form e
i
+ 1 = 0 shows how using distracts from
the identitys natural geometric meaning.) The quarter-angle identities have similar geometric interpretations:
e
i/4
= i says that a quarter turn in the complex plane is the same as multiplication by i, while e
i(3/4)
= i
says that three-quarters of a turn is the same as multiplication by i. A summary of these results, which we
might reasonably call Eulerian identities, appears in Table 1.
We can take this analysis a step further by noting that, for any angle , e
i
can be interpreted as a point
lying on the unit circle in the complex plane. Since the complex plane identies the horizontal axis with
the real part of the number and the vertical axis with the imaginary part, Eulers formula tells us that e
i
corresponds to the coordinates (cos , sin ). Plugging in the values of the special angles from Figure 8
then gives the points shown in Table 2, and plotting these points in the complex plane yields Figure 12. A
comparison of Figure 12 with Figure 8 quickly dispels any doubts about which choice of circle constant better
reveals the relationship between Eulers formula and the geometry of the circle.
12
Polar form Rectangular form Coordinates
e
i
cos + i sin (cos , sin )
e
i0
1 (1, 0)
e
i/12

3
2
+
1
2
i (

3
2
,
1
2
)
e
i/8 1

2
+
1

2
i (
1

2
,
1

2
)
e
i/6 1
2
+

3
2
i (
1
2
,

3
2
)
e
i/4
i (0, 1)
e
i/3

1
2
+

3
2
i (
1
2
,

3
2
)
e
i/2
1 (1, 0)
e
i(3/4)
i (0, 1)
e
i
1 (1, 0)
Table 2: Complex exponentials of the special angles from Figure 8.
e
i0
, e
i
e
i/2
e
i(3/4)
e
i/4
e
i/3
e
i/6
e
i/8
e
i/12
Figure 12: Complex exponentials of some special angles, plotted in the complex plane.
13
3 Circular area: the coup de grce
If you arrived here as a believer, you must by now be questioning your faith. is so natural, its meaning so
transparentis there no example where shines through in all its radiant glory? A memory stirsyes, there
is such a formulait is the formula for circular area! Behold:
A = r
2
.
We see here , unadorned, in one of the most important equations in mathematicsa formula rst proved by
Archimedes himself. Order is restored! And yet, the name of this section sounds ominous. . . If this equation
is s crowning glory, how can it also be the coup de grce?
3.1 Quadratic forms
Let us examine this paragon of , A = r
2
. We notice that it involves the diameterno, wait, the radius
raised to the second power. This makes it a simple quadratic form. Such forms arise in many contexts; as a
physicist, my favorite examples come from the elementary physics curriculum. We will now consider several
in turn.
3.1.1 Falling in a uniform gravitational eld
Galileo Galilei found that the velocity of an object falling in a uniform gravitational eld is proportional to
the time fallen:
v t.
The constant of proportionality is the gravitational acceleration g:
v = gt.
Since velocity is the derivative of position, we can calculate the distance fallen by integration:
y =
_
v dt =
_
t
0
gt dt =
1
2
gt
2
.
3.1.2 Potential energy in a linear spring
Robert Hooke found that the external force required to stretch a spring is proportional to the distance stretched:
F x.
The constant of proportionality is the spring constant k:
8
F = kx.
The potential energy in the spring is then equal to the work done by the external force:
U =
_
F dx =
_
x
0
kxdx =
1
2
kx
2
.
8
You may have seen this written as F = kx. In this case, F refers to the force exerted by the spring. By Newtons third law, the
external force discussed above is the negative of the spring force.
14
3.1.3 Energy of motion
Isaac Newton found that the force on an object is proportional to its acceleration:
F a.
The constant of proportionality is the mass m:
F = ma.
The energy of motion, or kinetic energy, is equal to the total work done in accelerating the mass to velocity v:
K =
_
F dx =
_
ma dx =
_
m
dv
dt
dx =
_
m
dx
dt
dv =
_
v
0
mv dv =
1
2
mv
2
.
3.2 A sense of foreboding
Having seen several examples of simple quadratic forms in physics, you may by now have a sense of fore-
boding as we return to the geometry of the circle. This feeling is justied.
As seen in Figure 13, the area of a circle can be calculated by breaking it down into circular rings of
length C and width dr, where the area of each ring is C dr:
dA = C dr.
Now, the circumference of a circle is proportional to its radius:
C r.
The constant of proportionality is :
C = r.
The area of the circle is then the integral over all rings:
A =
_
dA =
_
r
0
C dr =
_
r
0
r dr =
1
2
r
2
.
If you were still a partisan at the beginning of this section, your head has now exploded. For we see that
even in this case, where supposedly shines, in fact there is a missing factor of 2. Indeed, the original proof
by Archimedes shows not that the area of a circle is r
2
, but that it is equal to the area of a right triangle with
base C and height r. Applying the formula for triangular area then gives
A =
1
2
bh =
1
2
Cr =
1
2
r
2
.
There is simply no avoiding that factor of a half (Table 3).
3.2.1 Quod erat demonstrandum
We set out in this manifesto to show that is the true circle constant. Since the formula for circular area was
just about the last, best argument that had going for it, Im going to go out on a limb here and say: Q.E.D.
15
r
dr
dA = C dr
Figure 13: Breaking down a circle into rings.
Quantity Symbol Expression
Distance fallen y
1
2
gt
2
Spring energy U
1
2
kx
2
Kinetic energy K
1
2
mv
2
Circular area A
1
2
r
2
Table 3: Some common quadratic forms.
16
4 Conict and resistance
Despite the denitive demonstration of the superiority of , there are nevertheless many who oppose it, both
as notation and as number. In this section, we address the concerns of those who accept the value but not the
letter. We then rebut some of the many arguments marshaled against C/r itself, including the so-called Pi
Manifesto that defends the primacy of . In this context, well discuss the rather advanced subject of the
volume of a hypersphere (Section 4.3), which augments and amplies the arguments in Section 3 on circular
area.
4.1 One turn
The true test of any notation is usage; having seen used throughout this manifesto, you may already be
convinced that it serves its role well. But for a constant as fundamental as it would be nice to have some
deeper reasons for our choice. Why not , for example, or ? Whats so great about ?
There are two main reasons to use for the circle constant. The rst is that visually resembles :
after centuries of use, the association of with the circle constant is unavoidable, and using feeds on this
association instead of ghting it. (Indeed, the horizontal line in each letter suggests that we interpret the
legs as denominators, so that has two legs in its denominator, while has only one. Seen this way, the
relationship = 2 is perfectly natural.)
9
The second reason is that corresponds to one turn of a circle, and
you may have noticed that and turn both start with a t sound. This was the original motivation for
the choice of , and it is not a coincidence: the root of the English word turn is the Greek word for lathe,
tornosor, as the Greeks would put it,
oo.
Since the original launch of The Tau Manifesto, I have learned that physicist Peter Harremos indepen-
dently proposed using to Is Wrong! author Bob Palais, for essentially the same reasons, and in fact
Joseph Lindenberg anticipated both the argument and the symbol more than twenty years before! (Linden-
berg has included both his original typewritten manuscript and a large number of other arguments at his site
Tau Before It Was Cool.) Dr. Harremos has emphasized the importance of a point rst made in Section 1.1:
using gives the circle constant a name. Since is an ordinary Greek letter, people encountering it for the
rst time can pronounce it immediately. Moreover, unlike calling the circle constant a turn, works well
in both written and spoken contexts. For example, saying that a quarter circle has radian angle measure one
quarter turn sounds great, but turn over four radians sounds awkward, and the area of a circle is one-half
turn r squared sounds downright odd. Using , we can say tau over four radians and the area of a circle
is one-half tau r squared.
4.1.1 Ambiguous notation
Of course, with any new notation there is the potential for conict with present usage. As noted in Section 1.1,
Is Wrong! avoids this problem by introducing a new symbol (Figure 1). There is precedent for this; for
example, in the early days of quantum mechanics Max Planck introduced the constant h, which relates a
light particles energy to its frequency (through E = h), but physicists soon realized that it is often more
convenient to use (read h-bar)where is just h divided by. . . um. . . 2and this usage is now standard.
But getting a new symbol accepted is difcult: it has to be given a name, that name has to be popularized,
and the symbol itself has to be added to word processing and typesetting systems. Moreover, promulgating
a new symbol for 2 would require the cooperation of the academic mathematical community, which on the
subject of vs. has so far been apathetic at best and hostile at worst. Using an existing symbol allows us
9
Thanks to Tau Manifesto reader Jim Porter for pointing out this interpretation.
17
to route around the mathematical establishment. (Perhaps someday academic mathematicians will come to a
consensus on a different symbol for the number 2; if that ever happens, I reserve the right to support their
proposed notation. But they have had over 300 years to x this problem, so I wouldnt hold my breath.)
Rather than advocating a new symbol, The Tau Manifesto opts for the use of an existing Greek letter.
As a result, since is already used in some current contexts, we must address the conicts with existing
practice. Fortunately, there are surprisingly few common uses. Moreover, while is used for certain specic
variablese.g., shear stress in mechanical engineering, torque in rotational mechanics, and proper time in
special and general relativitythere is no universal conicting usage.
10
In those cases, we can either tolerate
ambiguity or route around the few present conicts by selectively changing notation, such as using N for
torque
11
or
p
for proper time.
Despite these arguments, potential usage conicts have proven to be the greatest source of resistance to .
Some correspondents have even atly denied that (or, presumably, any other currently used symbol) could
possibly overcome these issues. But scientists and engineers have a high tolerance for notational ambiguity,
and claiming that -the-circle-constant cant coexist with other uses ignores considerable evidence to the
contrary.
One example of easily tolerated ambiguity occurs in quantum mechanics, where we encounter the follow-
ing formula the Bohr radius, which (roughly speaking) is the size of a hydrogen atom in its lowest energy
state (the ground state):
a
0
=

2
me
2
,
where m is the mass of an electron and e is its charge. Meanwhile, the ground state itself is described by
a quantity known as the wavefunction, which falls off exponentially with radius on a length scale set by the
Bohr radius:
(r) = N e
r/a0
,
where N is a normalization constant.
Have you noticed the problem yet? Probably not, which is just the point. The problem is that the e in
the Bohr radius and the e in the wavefunction are not the same ethe rst is the charge on an electron, while
the second is the exponential number (the base of natural logarithms). In fact, if we expand the factor of a
0
in the argument of the exponent, we get
(r) = N e
me
2
r/
2
,
which has an e raised the power of something with e in it. This may seem crazy, but in practice no one
ever has any problem with using e in both contexts. There are many other examples, including situations
where even is used for two different things,
12
and its hard to see how using for multiple quantities is any
different.
By the way, the -pedants out there (and there have proven to be many) might note that hydrogens
ground-state wavefunction is typically written with a factor of :
(r) =

1
a
3
0
e
r/a0
.
10
The only possible exception to this is the golden ratio, which is often denoted by in Europe. But not only is there an existing
common alternative to this notationnamely, the Greek letter this usage shows that there is precedent for using to denote a
fundamental mathematical constant.
11
This alternative for torque is already in use; see, for example, Introduction to Electrodynamics by David Grifths, p. 162.
12
See, for instance, An Introduction to Quantum Field Theory by Peskin and Schroeder, where is used to denote both the circle
constant and a conjugate momentum on the very same page (p. 282).
18
At rst glance, this appears to be more natural than the version with :
(r) =

2
a
3
0
e
r/a0
.
As usual, appearances are deceiving: the value of N comes from the product
1

2
1

2
2
a
3/2
0
,
which shows that the circle constant enters the calculation through 1/

2, i.e., 1/

. As with the formula


for circular area, the cancellation to leave a bare is a coincidence. Well have more to say about this game
of whack-a-mole in Section 4.4.
4.2 The Pi Manifesto
Although most objections to come from scattered email correspondence and miscellaneous comments on
the Web, there is also an organized resistance. In particular, since the publication of The Tau Manifesto in
June 2010, a Pi Manifesto has appeared to make the case for the traditional circle constant. This section and
the two after it contain a rebuttal of its arguments. Of necessity, this treatment is terser and more advanced
than the rest of the manifesto, but even a cursory reading of what follows will give an impression of the
weakness of the Pi Manifestos case.
While we can certainly consider the appearance of the Pi Manifesto a good sign of continuing interest in
this subject, it makes several false claims. For example, it says that the factor of 2 in the Gaussian (normal)
distribution is a coincidence, and that it can more naturally be written as
1

2)
e
x
2
(

2)
2
.
This is wrong: the factor of

2 does go with the

. The factor of 2 comes from squaring the unnormalized


Gaussian distribution and switching to polar coordinates, which leads to a factor of 1 from the radial integral
and a 2 from the angular integral. As in the case of circular area, the factor of comes from 1/2 2, not
from alone.
A related claim is that the Gamma function evaluated at 1/2 is more natural in terms of :
(
1
2
) =

,
where
(p) =
_

0
x
p1
e
x
dx.
But (
1
2
) reduces to the same Gaussian integral as in the normal distribution (upon setting u = x
1/2
), so the
in this case is really 1/2 2 as well.Indeed, in many of the cases cited in the Pi Manifesto, the circle
constant enters through an integral over all angles, i.e., as ranges from 0 to .
The Pi Manifesto also examines some formulas for regular n-sided polygons (or n-gons). For instance,
it notes that the sum of the internal angles of an n-gon is given by
n

i=1

i
= (n 2).
19
This issue was dealt with in Pi Is Wrong!, which notes the following: The sum of the interior angles [of a
triangle] is , granted. But the sum of the exterior angles of any polygon, from which the sum of the interior
angles can easily be derived, and which generalizes to the integral of the curvature of a simple closed curve,
is 2. In addition, the Pi Manifesto offers the formula for the area of an n-gon with unit radius (the distance
from center to vertex),
A = nsin

n
cos

n
,
calling it clearly. . . another win for . But using the double-angle identity sin cos =
1
2
sin 2 shows that
this can be written as
A = n/2 sin
2
n
,
which is just
A =
1
2
n sin

n
.
In other words, the area of an n-gon has a natural factor of 1/2. In fact, taking the limit n (and applying
LHpitals rule) gives the area of a regular polygon with innitely many sides, i.e., a circle:
A = lim
n
1
2
n sin

n
=
1
2
lim
n
sin

n
1/n
=
1
2
.
In this context, we should note that the Pi Manifesto makes much ado about being the area of a unit
disk, so that (for example) the area of a quarter (unit) circle is /4. This, it is claimed, makes just as good
a case for as radian angle measure does for . Unfortunately for this argument, as noted in Section 3 and
seen again immediately above, the factor of 1/2 arises naturally in the context of circular area. Indeed, the
formula for the area of a circular sector subtended by angle is
1
2
r
2
,
so theres no way to avoid the factor of 1/2 in general. (Indeed, we see that A =
1
2
r
2
is simply the special
case = .) Because of this natural factor of 1/2, its perfectly normal for the area of, say, the sector
subtended by a quarter turn to have area
1
8
r
2
.
In short, the difference between angle measure and area isnt arbitrary. There is no natural factor of 1/2
in the case of angle measure. In contrast, in the case of area the factor of 1/2 arises through the integral
of a linear function in association with a simple quadratic form. In fact, the case for is even worse than
it looks: the general formula for the volume of an n-dimensional sphere, which contains the volume of a
2-sphere (i.e., the area of a circle) as a special case, makes clear that the factor of 1 in front of in r
2
is
mere coincidence. This nal example, also mentioned by the Pi Manifesto as an argument in favor of , is so
instructive that it deserves a separate section (Section 4.3).
4.3 Volume of a hypersphere
The formula volume of a hypersphere, also called an n-sphere, is a case study in broken pattern recognition.
The traditional form of the formula, written in terms of , is ugly and unnatural. As well see in this section,
when written in terms of the formula is elegant and beautiful, thus providing yet another argument for .
(Note: This section is substantially more advanced that the rest of the manifesto and can be skipped without
loss of continuity.)
An n-sphere is the set of all points a given xed (the radius) away from a given point (the center) in
n-dimensional space (R
n
). We can represent such an object conveniently by centering it on the origin and
20
writing an equation for the distance squared. For example, a 2-sphere (that is, a circle) centered on the origin
satises the equation
x
2
+ y
2
= r
2
.
Similarly, a 3-sphere (that is, an ordinary sphere) satises
x
2
+ y
2
+ z
2
= r
2
.
The generalization to arbitrary n, although difcult to visualize when n > 3, is straightforward:
n

i=1
x
2
i
= r
2
.
(There is some ambiguity about whether, e.g., a 2-sphere is a circle or a disk, but in general this should be
clear from context. For example, the volume of a 2-sphere obviously refers to the area of a disk.)
The Pi Manifesto (discussed in Section 4.2) includes the volume of a unit n-sphere as an argument in
favor of :

n
(1 +
n
2
)
,
where (as noted in Section 4.2) the Gamma function is (p) =
_

0
x
p1
e
x
dx. This is a special case of the
general formula, which is also typically written in terms of :
V
n
(r) =

n/2
r
n
(1 +
n
2
)
.
Before proceeding, we should note that this formula, when written in terms of , has a pleasing recurrence of
factors of 1/2:
V
n
(r) =
(

2
)
n
2
r
n
(1 +
n
2
)
.
By itself this isnt a strong argument, but it shows that the case for isnt completely clear-cut. And unfortu-
nately for , the case gets far worse from here.
Well plant an initial seed of doubt by noting that, in the case of a 2-sphere (i.e., a circular disk), we know
from Section 3 that there is a hidden factor of 2:
A = V
2
(r) = r
2
=
1
2
(2)r
2
=
1
2
r
2
.
In addition, the volume of a 3-sphere has an even number in front of , leading to an other factor of 2:
V = V
3
(r) =
4
3
r
3
=
2(2)
3
r
3
=
2
3
r
3
.
These examples suggest that perhaps the volume of an n-sphere might have natural factors of 2 for n > 3
as well.
We begin our analysis by noting that the apparent simplicity of the volume formula is an illusion: although
the Gamma function is notationally simple, in fact it is an integral over a semi-innite domain, which is not a
simple idea at all. Fortunately, the Gamma function can be simplied in certain special cases. For example,
when n is an integer, it is easy to show (using integration by parts) that
(n) = (n 1)(n 2) . . . 2 1 = (n 1)!
21
Seen this way, can be interpreted as a generalization of the factorial function to real-valued arguments.
13
In the n-dimensional volume formula, the argument of is not necessarily an integer, but rather is (1+
n
2
),
which is an integer when n is even and is a half-integer when n is odd. The simplication in this case, applied
to the formula for the volume of an n-sphere, appears as follows:
V
n
(r) =
_

n/2
r
n
(
n
2
)!
if n is even;
2
(n+1)/2

(n1)/2
r
n
n!!
if n is odd.
This equation is taken from the standard reference Wolfram MathWorld, and as usual it is written in terms
of .
Lets examine this formula in more detail to see where it goes wrong. Notice rst that MathWorld uses
the double factorial function n!!but, strangely, it uses it only in the odd case. (This is a hint of things to
come.) The double factorial function, though rarely encountered in mathematics, is elementary: its like the
normal factorial function, but involves subtracting 2 at a time instead of 1, so that, e.g., 5!! = 5 3 1 and
6!! = 6 4 2. In general, we have
n!! =
_

_
n(n 2)(n 4) . . . 5 3 1 if n is even;
n(n 2)(n 4) . . . 6 4 2 if n is odd.
Notice that this denition naturally divides into even and odd cases, making MathWorlds decision to use it
only in the odd case still more mysterious.
To solve this mystery, well start by taking a closer look at the formula for odd n:
2
(n+1)/2

(n1)/2
r
n
n!!
Upon examining the expression
2
(n+1)/2

(n1)/2
,
we notice that it can be rewritten as
2(2)
(n1)/2
,
and here we recognize our old friend 2.
Now lets look at the even case. We noted above how strange it is to use the ordinary factorial in the
even case but the double factorial in the odd case. Indeed, because the double factorial is already dened
piecewise, if we unied the formulas by using n!! in both cases we could pull it out as a common factor:
V
n
(r) =
1
n!!

_

_
. . . if n is even;
. . . if n is odd.
So, is there any connection between the factorial and the double factorial? Yeswhen n is even, the two are
related by the following identity:
_
n
2
_
! =
n!!
2
n/2
.
13
Indeed, the generalization to complex-valued arguments is straightforward.
22
(This is easy to verify using mathematical induction.) Substituting this into the volume formula for even n
then yields
2
n/2

n/2
r
n
n!!
,
which bears a striking resemblance to
(2)
n/2
r
n
n!!
,
and again we nd a factor of 2.
Putting these results together, we see that the volume of an n-sphere can be rewritten as
V
n
(r) =
_

_
(2)
n/2
r
n
n!!
if n is even;
2(2)
(n1)/2
r
n
n!!
if n is odd.
Making the substitution = 2 then yields
V
n
(r) =
_

n/2
r
n
n!!
if n is even;
2
(n1)/2
r
n
n!!
if n is odd.
To unify the formulas further, we can use the oor function x, which is simply the largest integer less than
or equal to x (so that, e.g., 3.7 = 3.2 = 3). This gives
V
n
(r) =
_

n
2

r
n
n!!
if n is even;
2

n
2

r
n
n!!
if n is odd.
which allows us to factor out most of the formula as follows:
V
n
(r) =

n
2

r
n
n!!

_

_
1 if n is even;
2 if n is odd.
As a nal simplication, we can dene
c
n
=

n
2

n!!

_

_
1 if n is even;
2 if n is odd.
so that the n-dimensional volume can be written as
V
n
(r) = c
n
r
n
.
The improvement on the original -encrusted monstrosity is unmistakable.
23
We are now in a position to fully understand the formula for circular area and its higher-dimensional
generalizations. The number is just c
2
:
= c
2
=

2
2

2!!
=

2
.
We see here that the factor of 1 in front of in r
2
is a coincidence: its only because the denominator is
2!! = 2 that it cancels with the 2 in 2. In higher dimensions, there is no such cancellation; for example,
when n = 3, we have
c
3
= 2

3
2

3!!
=
2
3
,
when n = 4, we have
c
4
=

4
2

4!!
=

2
8
,
and so on. In other words, is part of a family of constants c
n
, each of which depends explicitly on the
dimension of the space. The only part of the family that is independent of the dimensionality is .
4.4 What is really going on here?
Even after reading this manifesto, you might be tempted by the thought that the difference between and
isnt really that important after all. You may nd yourself saying, Both numbers are important: sometimes
is better, sometimes is more natural, and sometimes they are equally good. For example, consider the
famous Leibniz series for :

4
= 1 1/3 + 1/5 1/7 +
Is it really any better to write

8
= 1 1/3 + 1/5 1/7 +
instead? As in previous examples, we could make a specic argument for , in this case by noting that the
Leibniz series is simply the innite series for arctan x evaluated at x = 1, which is the angle where sine and
cosine are equal, which is an eighth of a turn, which is /8. This same issue aficts many arguments in the Pi
Manifesto, which includes formulas such as elliptical area (ab, which comes from integrating an elliptical
function from 0 to 2) and integrals such as
_
1
1
1

1 x
2
dx,
which evaluates to sin
1
(1) sin
1
(1), i.e., the difference between where sin is 1 (a quarter turn) and
1 (a negative quarter turn): /4 (/4) = /2. Und so weiter. Its tempting to try to counter every one
of the many formulas involving on a case-by-case basisindeed, my life as a tauist has involved becoming
an expert on these Tau-mudic minutiae, defending one formula at a timebut unfortunately such examples
are inexhaustible because the number /2 is, in fact, important.
Resolving this tension requires wading into some deep philosophical waters about the nature of mathe-
matical notation. Computer hacker and mathematical philosopher Eric Raymond raises these issues in his
post Tau versus Pi, suggesting a systematic audit of mathematics: Representatives of the Tauists and the pi
partisans should be locked in a mathematics library until they choose a list of books and papers that covers
trigonometry, calculus, and analysis. Then, the burden should be on the Tauists to translate the entire pile
into notation. Then, both sides should count symbols, checking each others work. Most compact notation
24
wins! I am condent that would win such a contest in a fair ght, but we immediately face the issue
of dening fairi.e., we would have to develop an unambiguous way to evaluate the complexity of any
given representation. Its not always obvious how to do this. For instance, in an important sense
1
2
r
2
is
simpler than r
2
, despite using more symbols, because it follows the pattern of other quadratic forms (and
of other n-dimensional spherical volumes). Moreover, even if we could agree on the metric, we have seen
that teasing out the missing factors of 2 often requires arduous effort, so performing such an audit would be
a Brobdingnagian taskone we tauists would rather avoid tackling directly.
Instead of making a frontal assault on this problem, lets make an oblique approach. Consider an analogy:
imagine we lived in a world where we used the letter h to represent one half and had no separate notation
for 2h. We would then observe that h is ubiquitous in mathematics. In fact, 2h is the multiplicative identity,
so how can one doubt the importance of h? All mathematicians and geeks agree, h is where its at. But this
is madness: 2h is the fundamental number, not h. Let us therefore introduce a separate symbol for 2h; call it
1. We then see that h = 1/2, and there is no longer any reason to use h at all.
Now, there is no doubt that the set {1, 1/2} would defeat {h, 2h} in a Raymond-style complexity audit,
even though h uses fewer symbols than 1/2. I admit I feel the tug here of mathematical PlatonismIs the
presumed result of such an audit the real reason people prefer 1 over 2h?but we can certainly all agree
on the result, regardless of method. Since weve seen in this manifesto that, with respect to circles, is a
unit (one turn), while is twice that unit, we can conclude that {, /2} is better than {, 2} for the
same reason that {1, 1/2} is better than {h, 2h}thereby saving Eric the trouble of chaining us inside a
mathematics library for the foreseeable future.
We are now in a position to see why comparing /4 with /8 misses the point: saying that /4 is just
as good as /8 is like saying that h/4 is just as good as 1/8. They are mathematically equivalent, but , like
1, is more fundamental, and this is reason enough to use the version with . If you ever hear yourself saying
things like, Sometimes is the best choice, and sometimes its 2, stop and remember the words of Vi Hart
in her wonderful video about tau: No! Youre making excuses for . Its time to stop making excuses.
5 Conclusion
Over the years, I have heard many arguments against the wrongness of and against the rightness of , so
before concluding our discussion allow me to answer some of the most frequently asked questions.
5.1 Frequently Asked Questions
Are you serious?
Of course. I mean, Im having fun with this, and the tone is occasionally lighthearted, but there is a
serious purpose. Setting the circle constant equal to the circumference over the diameter is an awkward
and confusing convention. Although I would love to see mathematicians change their ways, Im not
particularly worried about them; they can take care of themselves. It is the neophytes I am most worried
about, for they take the brunt of the damage: as noted in Section 2.1, is a pedagogical disaster.
Try explaining to a twelve-year-old (or to a thirty-year-old) why the angle measure for an eighth of
a circleone slice of pizzais /8. Wait, I meant /4. See what I mean? Its madnesssheer,
unadulterated madness.
How can we switch from to ?
The next time you write something that uses the circle constant, simply say For convenience, we set
= 2, and then proceed as usual. (Of course, this might just prompt the question, Why would
you want to do that?, and I admit it would be nice to have a place to point them to. If only someone
25
would write, say, a manifesto on the subject. . . ) The way to get people to start using is to start using
it yourself.
Isnt it too late to switch? Wouldnt all the textbooks and math papers need to be rewritten?
No on both counts. It is true that some conventions, though unfortunate, are effectively irreversible.
For example, Benjamin Franklins choice for the signs of electric charges leads to electric current being
positive, even though the charge carriers themselves are negativethereby cursing electrical engineers
with confusing minus signs ever since.
14
To change this convention would require rewriting all the
textbooks (and burning the old ones) since it is impossible to tell at a glance which convention is being
used. In contrast, while redening is effectively impossible, we can switch from to on the y by
using the conversion

1
2
.
Its purely a matter of mechanical substitution, completely robust and indeed fully reversible. The
switch from to can therefore happen incrementally; unlike a redenition, it need not happen all at
once.
Wont using confuse people, especially students?
If you are smart enough to understand radian angle measure, you are smart enough to understand
and why is actually less confusing than . Also, there is nothing intrinsically confusing about
saying Let = 2; understood narrowly, its just a simple substitution. Finally, we can embrace
the situation as a teaching opportunity: the idea that might be wrong is interesting, and students
can engage with the material by converting the equations in their textbooks from to to see for
themselves which choice is better.
Does any of this really matter?
Of course it matters. The circle constant is important. People care enough about it to write entire
books on the subject, to celebrate it on a particular day each year, and to memorize tens of thousands
of its digits. I care enough to write a whole manifesto, and you care enough to read it. Its precisely
because it does matter that its hard to admit that the present convention is wrong. (I mean, how do you
break it to Lu Chao, the current world-record holder, that he just recited 67,890 digits of one half of
the true circle constant?)
15
Since the circle constant is important, its important to get it right, and we
have seen in this manifesto that the right number is . Although is of great historical importance, the
mathematical signicance of is that it is one-half .
Why did anyone ever use in the rst place?
As notation, was popularized around 300 years ago by Leonhard Euler (based on the work of William
Jones), but the origins of -the-number are lost in the mists of time. I suspect that the convention of
using C/D instead of C/r arose simply because it is easier to measure the diameter of a circular
object than it is to measure its radius. But that doesnt make it good mathematics, and Im surprised
that Archimedes, who famously approximated the circle constant, didnt realize that C/r is the more
fundamental number. Im even more surprised that Euler didnt correct the problem when he had the
chance; unlike Archimedes, Euler had the benet of modern algebraic notation, which (as we saw
starting in Section 2.1) makes the underlying relationships between circles and the circle constant
abundantly clear.
Why does this subject interest you?
First, as a truth-seeker I care about correctness of explanation. Second, as a teacher I care about clarity
14
The sign of the charge carriers couldnt be determined with the technology of Franklins time, so this isnt his fault. Its just bad
luck.
15
On the other hand, this could be an opportunity: the eld for recitation records is wide open.
26
of exposition. Third, as a hacker I love a nice hack. Fourth, as a student of history and of human
nature I nd it fascinating that the absurdity of was lying in plain sight for centuries before anyone
seemed to notice. Moreover, many of the people who missed the true circle constant are among the
most rational and intelligent people ever to live. What else might be staring us in the face, just waiting
for us to discover it?
Are you, like, a crazy person?
Thats really none of your business, but no. Apart from my unusual shoes, I am to all external appear-
ances normal in every way. You would never guess that, far from being an ordinary citizen, I am in fact
a notorious mathematical propagandist.
But what about puns?
We come now to the nal objection. I know, I know, in the sky is so very clever. And yet, itself
is pregnant with possibilities. ism tells us: it is not that is a piece of , but that is a piece of
one-half , to be exact. The identity e
i
= 1 says: Be 1 with the . And though the observation
that A rotation by one turn is 1 may sound like a -tology, it is the true nature of the . As we
contemplate this nature to seek the way of the , we must remember that ism is based on reason, not
on faith: ists are never ous.
5.2 Embrace the tau
We have seen in The Tau Manifesto that the natural choice for the circle constant is the ratio of a circles
circumference not to its diameter, but to its radius. This number needs a name, and I hope you will join me
in calling it :
circle constant =
C
r
= 6.283185307179586 . . .
The usage is natural, the motivation is clear, and the implications are profound. Plus, it comes with a really
cool diagram (Figure 14). We see in Figure 14 a movement through yang (light, white, moving up) to /2
and a return through yin (dark, black, moving down) back to .
16
Using instead of is like having yang
without yin.
5.3 Tau Day
The Tau Manifesto rst launched on Tau Day: June 28 (6/28), 2010. Tau Day is a time to celebrate and rejoice
in all things mathematical and true.
17
If you would like to receive updates about , including notications
about possible future Tau Day events, please join the Tau Manifesto mailing list below. And if you think that
the circular baked goods on Pi Day are tasty, just waitTau Day has twice as much pi(e)!
The signup form is available online at http://tauday.com/signup.
5.3.1 Acknowledgments
Id rst like to thank Bob Palais for writing Is Wrong!. I dont remember how deep my suspicions about
ran before I encountered that article, but Is Wrong! denitely opened my eyes, and every section of
16
The interpretations of yin and yang quoted here are from Zen Yoga: A Path to Enlightenment though Breathing, Movement and
Meditation by Aaron Hoopes.
17
Since 6 and 28 are the rst two perfect numbers, 6/28 is actually a perfect day.
27
0,
Figure 14: Followers of ism seek the way of the .
28
The Tau Manifesto owes it a debt of gratitude. Id also like to thank Bob for his helpful comments on this
manifesto, and especially for being such a good sport about it.
Ive been thinking about The Tau Manifesto for a while now, and many of the ideas presented here were
developed through conversations with my friend Sumit Daftuar. Sumit served as a sounding board and occa-
sional Devils advocate, and his insight as a teacher and as a mathematician inuenced my thinking in many
ways.
I have also received encouragement and helpful feedback from several readers. Id like to thank Vi Hart
and Michael Blake for their amazing -inspired videos, as well as Don Blue McConnell and Skona Brittain
for helping make part of geek culture (through the time-in- iPhone app and the tau clock, respectively).
The pleasing interpretation of the yin-yang symbol used in The Tau Manifesto is due to a suggestion by
Peter Harremos, who (as noted above) has the rare distinction of having independently proposed using
for the circle constant. I also got several good suggestions from Christopher Olah, particularly regarding
the geometric interpretation of Eulers identity, and Section 2.3.2 on Eulerian identities was inspired by an
excellent suggestion from Timothy Patashu Stiles. The site for Half Tau Day beneted from suggestions
by Evan Dorn, Wyatt Greene, Lynn Noel, Christopher Olah, and Bob Palais. Finally, Id like to thank Wyatt
Greene for his extraordinarily helpful feedback on a pre-launch draft of the manifesto; among other things,
if you ever need someone to tell you that pretty much all of [now deleted] section 5 is total crap, Wyatt is
your man.
5.3.2 About the author
The Tau Manifesto author Michael Hartl is a physicist, educator, and entrepreneur. He is the author of
the Ruby on Rails Tutorial, the leading introduction to web development with Ruby on Rails. Previously,
he taught theoretical and computational physics at Caltech, where he received the Lifetime Achievement
Award for Excellence in Teaching and served as Caltechs editor for The Feynman Lectures on Physics: The
Denitive and Extended Edition. He is a graduate of Harvard College, has a Ph.D. in Physics from the
California Institute of Technology, and is an alumnus of the Y Combinator entrepreneur program.
Michael is ashamed to admit that he knows to 50 decimal placesapproximately 48 more than Matt
Groening. To atone for this, he has now memorized 52 decimal places of .
5.3.3 Copyright and license
The Tau Manifesto. Copyright c 2011 by Michael Hartl. Please feel free to share The Tau Manifesto, which
is available under the Creative Commons Attribution 3.0 Unported License. This means that you can adapt
it, translate it, or even include it in commercial works, as long as you attribute it to me (Michael Hartl) and
link back to tauday.com. You also have permission to distribute copies of The Tau Manifesto PDF, print it
out, use it in classrooms, and so on. Go forth and spread the good news about !
29

You might also like