Copernicus’s epicycles from Newton’s gravitational force law

via linear perturbation theory in geometric algebra

Quirino M. Sugon Jr.,* Sarah Bragais, and Daniel J. McNamara
arXiv:0807.2708v1 [physics.space-ph] 17 Jul 2008

Ateneo de Manila University, Department of Physics, Loyola Heights, Quezon City, Philippines 1108
*Also at Manila Observatory, Upper Atmosphere Division, Ateneo de Manila University Campus
e-mail: qsugon@observatory.ph

17 July 2008

move closest to the sun (perihelion) and sometimes far-

thest (aphelion), Copernicus displaced the center of the
Abstract. We derive Copernicus’s epicycles from New- planet’s circular orbit a little away from the sun to a new
ton’s gravitational force law by assuming that a planet’s point called the eccentric, so that the new position r̂ of
orbit is a perturbed circular orbit, with the perturbation the planet is
defined to be co-rotating with the said orbit. We substi- r̂ = r−1 + r0 eı̂ω0 t , (2)
tute this orbit expression into Newton’s gravitation law
where r−1 is the distance of the eccentric from the sun.
and showed that the perturbation satisfies the linear part
Actually, the eccentric hypothesis in Eq. (2) is not
of Hill’s oscillator equation for lunar motion. We solve
Copernicus’ original idea but was already known more
this oscillator equation using an exponential Fourier se-
than a thousand years prior by Ptolemy (though he as-
ries and impose the boundary conditions at the aphelion
sumed that the earth is at rest and not the sun as in
and perihelion to derive the Copernicus’s formulas for the
Copernicus). In fact, if we factor out the exponential
eccentric, deferent, and epicycle. We show that for small
eı̂ω0 t in Eq. (2), we would arrive at Ptolemy’s theorem
eccetricity, the Copernican orbit expression also leads to
applied by Copernicus in his heliocentric theory:
Kepler’s law of areas for planetary motion. The formal-
ism we use is the Clifford (geometric) algebra Cl2,0 . r̂ = r−1 + r0 eı̂ω0 t = (r0 + r−1 e−ı̂ω0 t )eı̂ω0 t . (3)
In Ptolemaic terms, r−1 e−ı̂ω0 t is called an epicycle and
1 Introduction Eq. (3) is called the eccentric-epicycle equivalence theo-
rem. (The actual theorem is stated geometrically.[2, 3])
In many introductory physics courses, especially those Notice that the theorem essentially states the equivalence
dealing with the history and philosophy of science, of the description of the planet’s position in the inertial
the Copernican model is taught conceptually but not frame (left hand side) and in the rotating frame (quantity
mathematically.[1] And in undergraduate and graduate in parenthesis on the right hand side).
physics courses, it is not even mentioned at all. One Yet Eq. (2) is still not consistent with the numerical
possible reason is that, unlike in the case of Kepler’s el- data. Ptolemy resolved this problem by assuming that
lipse, Copernicus’ epicycles has not been rigorously de- the planet’s circular orbit is uniform not with respect to
rived before from first principles, i.e., from Newton’s laws the orbit’s geometric center but on another point called
of motion and gravitation. So our aim in this paper is to the equant[4, 5]. Though this construction saves the
present this derivation. But before we do so, let us first appearances, Copernicus claimed that the equant goes
review the Copernican model. against the idea of uniform circular motion and for him
Copernicus believed that planets orbit around the sun. this is ”not sufficiently pleasing to the mind”[6]. To rem-
If the orbit of a planet is circular with the sun at the edy this aesthetic difficulty, Copernicus added on top of
center, then the planet’s position in complex form is his original circle in the inertial frame another circle with
twice the frequency[7]. In complex notation, we write
r̂ = r0 eı̂ω0 t , (1)
r̂ = r−1 + r0 eı̂ω0 t + r1 e2ı̂ω0 t (4)
where r0 and ω0 are the planet’s orbital radius and fre-
quency, respectively. But because planets sometimes where r−1 = −3r1 .

1
 In terms of the orbit’s semimajor axis and eccentric- for µ, ν = 1, 2. That is,
ity ǫ, the Copernican expression in Eq. (4) becomes
e21 = e22 = 1, (11)
3 1
r̂ = A( ǫ + eı̂ω0 t − ǫ e2ı̂ω0 t ), (5) e1 e2 = e2 e1 . (12)
2 2
as given by Gallavotti[5]. Notice that Eq. (5) is different The first equation algebraically defines e1 and e2 as unit
from that derived from Kepler’s elliptical orbit for small vectors by setting their squares to unity; the second equa-
eccentricity:[8] tion defines the vectors as mutually orthogonal by mak-
ing their product anticommute.
r̂ = A(ǫ(1 − ı̂) + eiω0 t + ǫ(1 + ı̂)e2iω0 t ). (6) Let us define the unit bivector

In this paper, our aim is to show that the Copernican ı̂ = e1 e2 . (13)

expression in Eq. (5) is a consequence of Newton’s grav-
itational force law. From the orthonormality axiom in Eq. (10), it is easy to
We shall divide the paper into four sections. The first see that ı̂ is an imaginary number,
section is Introduction. In the second section, we shall
present a brief tutorial on the Clifford (geometric) al- ı̂2 = e1 e2 e1 e2 = −e1 (e2 e2 )e1 = −e1 e1 = −1, (14)
gebra Cl2,0 for the plane[9, 10, 11, 12, 13], which com-
bines scalars, vectors and imaginary numbers. We shall that anticommutes with vectors e1 and e2 :
show how the exponential Fourier series are related to
eccentrics, deferents, and epicycles.[14, 15] In the third e1 ı̂ = e2 = −ı̂e1 , (15)
section, we shall introduce a pertubation in the planet’s
position in the frame co-rotating with the planet’s un- e2 ı̂ = −e1 = −ı̂e2 . (16)
perturbed circular orbit and substitute the result to the
vector form of Newton’s law of gravitation. We shall Notice that right-multiplying ı̂ to a vector rotates it coun-
show that the perturbation in complex form satisfies the terclockwise by π/2.
linear harmonic oscillator equation In general, a vector a in the two-dimensional space
spanned by e1 and e2 is given by
3
0 = ¨ŝ + 2ı̂ω0 ŝ˙ − ω02 (ŝ + ŝ∗ ), (7)
2 a = ax e1 + ay e2 = e1 â = â∗ e1 , (17)
whose scalar and imaginary parts are
where
0 = ẍs − 2ω0 ẏs − 3ω02 xs , (8)
â = ax + ay ı̂, (18)
0 = y¨s + 2ω0 ẋs . (9)
â∗ = ax − ay ı̂. (19)
These equations are the linear part of Hill’s equations
for lunar motion[16, 17]. We shall solve Eq. (7) using ex- Equations (17) to (19) relates the vector a to the complex
ponential Fourier series and impose the boundary condi- number â and its complex conjugate â∗ .
tions at the aphelion and perihelion to derive the Fourier If vector b = bx e1 + by e2 , then the product of vectors
coefficients of the Copernican orbit. And the fourth sec- a and b is
tion is Conclusions. ab = a · b + a ∧ b = â∗ b̂, (20)

where
2 Geometric Algebra
a·b = ax b x + ay b y , (21)
2.1 Vectors and Complex Numbers a∧b = (ax by − ay bx )ı̂ (22)
The Clifford (geometric) algebra Cl2,0 is an associative
algebra generated by two vectors e1 and e2 that corre- are the scalar (dot) and imaginary (bivector or planar)
spond to the basis vectors along the x− and y−axis in parts of the product ab = â∗ b̂. Notice that the magni-
the Cartesian coordinate system. The vectors satisfy the tude of the wedge product is that of the cross product
orthonormality relation a × b. (Geometrically, we say that a × b is the vector
perpendicular to the oriented plane a ∧ b, though tech-
eµ eν + eν eµ = 2δµν , (10) nically, a × b is not defined in Cl2,0 —only in Cl3,0 ).

2
2.2 Circles, Epicycles, and Fourier Series we arrive at
Because ı̂ is an imaginary number, then Euler’s theorem r = e1 (r̂0 + r̂1 ψ̂1 + r̂2 ψ̂2 ). (32)
holds:
eı̂θ = cos θ + ı̂ sin θ, (23) One way to simplify Eq. (32) is to set ω1 = ω and ω2 =
where θ is a real number. If we left-multiply Eq. (23) by 2ω. So using the definition of the rotor ψ in Eq. (28), we
e1 , we get get
e1 eı̂θ = e1 cos θ + e2 sin θ, (24) r = e1 (r̂0 + r̂1 ψ̂ + r̂2 ψ 2 ). (33)

where we used Eq. (15). Equation (24) states that e1 eı̂θ The zeroth harmonic is the eccentric; the first, the def-
is the vector e1 rotated counterclockwise by an angle θ erent; and the second, the epicycle. In general, we may
(assuming that e1 points to the right and e2 points up). express the position r in time t as an infinite Fourier
series:
∞ ∞
e2 r = e1
X
r̂k ψ̂ k =
X
e1 reı̂(kωt+φk ) . (34)
k =−∞ k =−∞
r
Equation (34) represents the Copernican ideal of decom-
posing an orbit as a sum of epicycles with harmonic fre-
r quencies.
ωt
φ e1 3 Copernican Dynamics
3.1 Uniform Circular Orbit
In Newton’s law of gravitation, the equation of motion
of a planet of mass m revolving around the sun of mass
M is
r
r̈ = −GM 3 , (35)
|r|
Fig. 1. The vector r = e1 rei(ωt+φ) . where r is the position of the planet with respect to the
the sun at the origin.
The theorem in Eq. (24) enables us to express the po- One to solution to Eq. (35) is a circular orbit:
sition r of a point in uniform circular motion as
r = r0 = e1 r̂0 ψ̂0 = e1 r0 eı̂(ω0 t+φ0 ) , (36)
r = e1 reı̂(ωt+φ) = e1 r cos(ωt+φ)+e2 r sin(ωt+φ), (25)

where r is radius, ω is the angular frequency, and φ is where r0 is the orbital radius, ω0 is the orbital frequency,
the rotational phase angle. Another way to express r is and φ0 is the orbital phase angle.
To verify this claim, we first take the derivatives in
r = e1 r̂ψ̂, (26) time of the position vector r:

where ṙ = ω0 e1 ı̂r̂0 ψ̂0 = −ı̂ω0 r0 , (37)

r̈ = −ω02 r0 . (38)
r̂ = reı̂φ , (27)
ı̂ωt
ψ̂ = e (28) Next, we take the square of the position r by using the
conjugation theorem in Eq. (17):
are the complex radius and rotor (rotation operator), re-
spectively. (See Fig. (1)) r2 = e1 r̂0 ψ̂0 e1 r̂0 ψ̂0 = r̂0∗ ψ̂0−1 r̂0 ψ̂0 = r̂0∗ r̂0 = r02 , (39)
Let r1 and r2 be two rotating vectors:
so that |r| = r0 as we expect. And finally, we substitute
r1 = e1 r̂1 ψ̂1 = e1 r1 ei(ω1 t+φ1 ) , (29) Eqs. (37) to (39) back to Eq. (35) to arrive at
r2 = e1 r̂2 ψ̂2 = e1 r2 ei(ω2 t+φ2 ) . (30)
GM
ω02 = , (40)
If we displace their sum by a vector r0 , r03

r0 = e1 r̂0 = e1 r0 eı̂φ0 , (31) which is the circular orbit condition.

3
3.2 Linear Perturbation Theory 4 Copernican Analysis
Let us assume that the solution to Eq. (35) may be ex-
pressed as a sum of a circular orbital position r0 and its
4.1 Epicyclical Fourier Series
small correction r1 : We assume that the solution to the orbital harmonic os-
cillator equation in Eq. (50) is an exponential Fourier
r = r0 + λr1 , (41) series with ω0 as the fundamental angular frequency:
where λ is a perturbation parameter that will be set to ∞
X
unity later. If we also assume that the perturbation r1 ŝ = âk ψ̂0k . (51)
lies in the same orbital plane as the original circular orbit k =−∞
r0 in Eq. (36) and co-rotating with it, then we may write
r1 as The time derivatives of ŝ are
r1 = e1 ŝψ̂0 , (42) ∞
ŝ˙
X
= ı̂ω0 kâk ψ̂0k , (52)
where ŝ is a complex function. Hence,
k =−∞
∞
r = e1 (r̂0 + λŝ)ψ̂0 . (43) ¨ŝ =
X
−ω02 k 2 âk ψ̂0k , (53)
Taking the first and second time derivatives of the po- k =−∞

sition r in Eq. (43), we get

while the conjugate of ŝ is
ṙ = e1 (λŝ˙ + ı̂ω0 (r̂0 + λŝ))ψ̂0 , (44) ∞
X ∞
X
r̈ = e1 (λ¨ŝ + 2λı̂ω0 ŝ˙ − ω 2 (r̂0 + λŝ))ψ̂0 . (45) ŝ∗ = â∗k ψ̂0−k = â∗−k ψ̂0k . (54)
0
k =−∞ k =−∞
Equation (45) provides the expansion of the left side of
Newton’s gravitation law in Eq. (35). Substituting Eqs. (51) to (54) back to Eq. (50), we
On the other hand, to rewrite the right side of the get
gravitation law, we need first to take the square of the ∞
position vector r in Eq. (43) and retain only the terms
X 3 3
0= ((k 2 + 2k + )âk + η̂02 â∗k )ψ̂0k , (55)
up to first order in λ: 2 2
k =−∞

r2 = (r̂0 + λŝ)∗ (r̂0 + λŝ) ≈ r02 + λ(r̂0∗ ŝ + r̂0 ŝ∗ ). (46) after factoring out −ω02 and rearranging the terms. Be-
cause the rotors ψ̂0k are orthonormal in the Fourier sense,
Raising both sides of Eq. (46) to −3/2 power and em-
then Eq. (55) holds only if the coefficient of ψ̂0k is zero
ploying the binomial theorem, we get
for all k:
 
1 1 3 ∗ 3 3
≈ 3 1−λ (η̂ ŝ + η̂0 ŝ ) ,
∗
(47) 0 = (k 2 + 2k + )âk + η̂02 â∗k . (56)
|r|3 r0 2r0 0 2 2
where Solving for the coefficient âk in Eq. (56), we get
η̂0 = eı̂φ0 . (48)
2 3
Multiplying Eq. (47) by the position r in Eq. (43) yields â−k = − η̂02 (k 2 + 2k + )â∗k . (57)
3 2
 
r 1 λ 3 2 ∗
≈ r0 + e 1 3 ŝ − (ŝ + η̂0 ŝ ) ψ̂0 , (49) Replacing the index k by −k,
|r|3 r03 r0 2r0
2 3
where we retained only the terms up to first order in λ. ak = − η̂02 (k 2 − 2k + )â∗−k , (58)
3 2
Now, substituting Eqs. (43) and (47) back to the grav-
itation law in Eq. (35), we arrive at and substituting the result back in Eq. (57), we arrive at
3 3 3 9
0 = ¨ŝ + 2ı̂ω0 ŝ˙ − ω02 (ŝ + η̂02 ŝ∗ ). (50) 0 = (k 2 + 2k + )(k 2 − 2k + ) − = k 2 (k 2 − 1), (59)
2 2 2 4
If we set φ0 = 0 (this means that orbit is not tilted, as after factoring out âk and rearranging the terms. Hence,
we shall show later), so that η̂0 = eiφ0 = 1, we get Hill’s
oscillator equation in Eq. (7). k = {−1, 0, 1}. (60)

4
 Because the values of the index k are limited by the position vector r in Eq. (66) yields two simultaneous
Eq. (60), then the Fourier series for the perturbation ŝ equations:
in Eq. (51) simplifies to
ra = e1 ra eı̂γ = e1 (â−1 + r̂0 + â1 ), (67)
ŝ = â−1 ψ̂0−1 + â0 + â1 ψ̂0 . (61) rp = −e1 rp eı̂γ = e1 (â−1 − r̂0 + â1 ). (68)
Factoring out e1 from Eqs. (67) and (68) and using the
The relationship between the coefficients â−1 and â1 in
expression for â−1 in Eq. (62), we get
Eq. (61) may be obtained by setting k = 1 in Eq. (57):
ra eı̂γ = −3η̂02 â∗1 + r̂0 + â1 , (69)
â−1 = −3η̂02 â∗1 . (62)
−rp e ı̂γ
= −3η̂02 â∗1 − r̂0 + â1 , (70)
Similarly, the condition for â0 is which are two simultaneous equations for r̂0 and â1 .
Deferent. To solve for r̂0 , we take the difference of
â0 = −η̂02 â∗0 . (63) Eqs. (69) and (70) to get
This is satisfied in three possible ways: 1
r̂0 = r0 η̂0 = r0 eı̂φ0 = (ra + rp )eı̂γ , (71)
2
â0 = {±ı̂η̂0 , 0}. (64)
so that
Substituting the expression for ŝ in Eq. (61) back to 1
r0 = (ra + rp ), (72)
the position vector expression in Eq. (43), we get 2
φ0 = γ. (73)
r = e1 (â−1 + (r̂0 + â0 )ψ̂0 + â1 ψ̂02 ). (65)
Thus, the radius r0 of the deferent circle is the length
Let us count the number of unknowns in this equation. of the semimajor axis of the orbit; the phase angle φ0 is
The coefficient â−1 is related to â1 by Eq. (62). The the angle of inclination of the semimajor axis from the
x−axis along e1 .
angular frequency ω0 in ψ̂0 = eiω0 t is related to the radius
Epicycle. To solve for the coefficient â1 , we add the
r0 of r̂0 = r0 η̂0 = r0 eiφ0 by Eq. (40). The phase angle φ0
Eqs. (67) and (68) to obtain
of r̂0 is related to that of â0 by Eq. (64). Thus, there are
five unknowns in Eq. (65): a1x , a1y , r0 , φ0 , and a0 . 1
(ra − rp )eı̂γ = −3η̂02 â∗1 + â1 . (74)
However, the orbit of a planet in the plane is com- 2
pletely specified in two ways: (a) given the position r1 Because â1 = ax + ı̂ay cannot be readily isolated, we
and the velocity v1 at a particular time t1 or (b) given separate the real and imaginary parts of Eq. (74) to get
the positions r1 and r2 at their respective times t1 and t2 .
1
In other words, there are two constraint vector equations (ra − rp ) cos γ = a1x (−3 cos 2γ + 1)
that are equivalent to four scalar equations for the com- 2
ponents. These four equations can only determine four + a1y (−3 sin 2γ), (75)
unknowns and not five, so one of our unknowns is super- 1
(ra − rp ) sin γ = a1x (−3 sin 2γ)
fluous and this must be a0 because â0 = 0 is a possibility 2
in Eq. (64). Thus, Eq. (65) reduces to + a1y (3 cos 2γ + 1), (76)
where we used the relation φ0 = γ in Eq. (73). Solving for
r = e1 (â−1 + r̂0 ψ̂0 + â1 ψ̂02 ). (66)
the components a1x and a1y and combining the results,
Because of the similarity of Eq. (66) to Eq. (33), we rec- we arrive at
1
ognize e1 â−1 as the eccentric, e1 r̂0 ψ̂0 as the deferent, and â1 = − (ra − rp )eı̂γ . (77)
4
e1 â1 ψ̂02 as the epicycle in the Copernican model. Equation (77) states that the radius a1 of the epicycle â1
is one-fourth the difference between the aphelion distance
4.2 Boundary Conditions: Aphelion and ra and the perihelion distance rp . Note the negative sign.
Perihelion Eccentric. After knowing â1 , we use the coefficient
relation in Eq. (62) to solve for â−1 :
Suppose that at t = 0, the planet is at its aphelion
position ra at a distance ra from the sun at a coun- 3
(ra − rp )eı̂γ .
â−1 = (78)
terclockwise angle γ from the positive x−axis; while at 4
t = τ /2 = π/ω0 the planet is at its perihelion position rp Equation (78) states that the length a−1 of the eccen-
at a distance rp from the sun at a similar angle from the tric is three-fourth the difference between the aphelion
negative x−axis. Imposing these boundary conditions on distance ra and the perihelion distance rp .

5
4.3 Copernican Orbit The time derivative of the planet’s position r in
Eq. (83) is
We now substitute the expressions â−coefficients in
v = e1 ı̂ωA(ψ̂0 − ǫψ̂02 ). (86)
Eqs. (77) and (78) and that of r̂0 in Eq. (71) back to
the expression for the position r in Eq. (66) to get Left-multiplying this by r,
3 1
3 1 1 rv = ı̂ω0 A2 ( ǫ + ψ̂0−1 − ǫψ̂0−2 )(ψ̂0 − ǫψ̂02 )
r = e1 eı̂γ ( (ra −rp )+ (ra +rp )ψ̂0 − (ra −rp )ψ̂02 ). (79) 2 2
4 2 4 1 1
= ı̂ω0 A2 [ − ǫψ̂0−1 + (1 + ǫ2 )
If the semimajor axis’ inclination angle γ = φ0 = 0, then 2 2
Eq. (79) reduces to 1 3 2 2
+ ǫψ̂0 − ǫ ψ̂0 ], (87)
2 2
3 1 1
r = e1 ( (ra − rp ) + (ra + rp )ψ̂0 − (ra − rp )ψ̂02 ). (80) and separating the scalar and imaginary parts of the re-
4 2 4
sult, we arrive at
Equation (80) is our desired approximation of a planet’s
orbit around the sun using eccentric, deferent, and epicy- 3
cle in terms of the planet’s aphelion ra and perihelion rp . r·v = ω0 A2 (−ǫ sin(ω0 t) + ǫ2 sin(2ω0 t)), (88)
2
(See Fig. (2)) 1 3
If we employ the definitions of the Keplerian semima- r∧v = ı̂ω0 A2 ((1 + ǫ2 ) − ǫ2 cos(2ω0 t)). (89)
2 2
jor axis A and eccentricity ǫ,
1
A = (ra + rp ), (81)
2
ra − rp
ǫ = (82)
ra + rp

then we may rewrite Eq. (80) as

3 1
r = e1 A( ǫ + ψ̂0 − ǫψ̂02 ), (83)
2 2 rp ra
which is Eq. (5). Or in Cartesian coordinates,
3 1
x = A( ǫ + cos(ω0 t) − ǫ cos(2ω0 t)), (84)
2 2
1
y = A(sin(ω0 t) − ǫ sin(2ω0 t)). (85)
2
Fig. 3. The position of a planet in Copernican
orbit from t = 0 to t = τ /2 at a time interval
of τ /12. The orbit’s eccentricity is ǫ = 1/3.
For nearly circular orbits, the eccentricity ǫ ≈ 0 (ǫ is
r0 a1
0.0167 for earth and 0.0068 for Venus). So dropping the
ǫ2 terms in Eqs. (88) and (89), we arrive at

a−1 r·v = −ω0 A2 ǫ sin(ω0 t), (90)

r∧v = ı̂ω0 A2 . (91)
The first equation means that the position and velocity
of a planet are perpendicular at the aphelion (t = 0) and
perihelion (t = τ /2 = π/ω0 ); the second equation implies
that the oriented area
A = r ∧ (vδt) = ı̂ω0 A2 δt (92)
Fig. 2. Compass-and-straightedge plotting of swept by the radius vector r for a small interval of time δt
a Copernican orbit with eccentric distance a1 , is constant, which is Kepler’s second law (we can always
deferent radius r0 , and epicycle radius a1 . The perform an integral to show the validity of the law for
orbit’s eccentricity is ǫ = 1/3. large time intervals). (See Fig. (3))

6
5 Summary and Conclusions [4] Claudius Ptolemaeus, Ptolemy’s Almagest, trans.
Gerald J. Toomer (Princeton U.P., Princeton, New
In this paper, we derived the Copernican system of epicy- Jersey, 1998), p. 443.
cles from Newton’s gravitational force law in vector form
via linear perturbation theory in Clifford (geometric) al- [5] Giovanni Gallavotti, “Quasi-periodic mo-
gebra Cl2,0 of the plane. We assumed that the planet’s or- tions from Hipparchus to Kolmogorov,”
bit is a perturbed circular orbit, where the perturbation arXiv:chao-dyn/9907004 v1, p. 10 of 14 pages.
is defined as a vector co-rotating with the original orbit. [6] Edward Rosen, “The Commentariolus of Coperni-
We substituted this expression into Newton’s gravitation cus,” Osiris, 3, 123–141 (1937).
law. Using binomial expansion, we showed that this per-
turbation may be represented by a complex function ŝ [7] Ref. [3], pp. 742–743.
that satisfies the linearized form of Hill’s equation for lu- [8] Ref. [5], p. 6.
nar motion. This equation is a linear harmonic oscillator
with imaginary damping term and an extra forcing term [9] David Hestenes, New Foundations for Classical Me-
that is proportional to the conjugate ŝ∗ . chanics (Kluwer Academic, Dordrecht, 1990), 2nd
We solved this oscillator equation using exponential ed., p. 48–53.
Fourier series with the frequency ω0 of the unperturbed [10] Bernard Jancewicz, Multivectors and Clifford Al-
circular orbit as the fundamental frequency. We showed gebra in Electrodynamics (World Scientific, Singa-
that only three harmonics are allowed: -1, 0, and 1. pore, 1988), pp. 1–17.
This result makes the planet’s position as an expo-
nential Fourier series with three harmonics: 0, 1, 2— [11] Chris Doran and Anthony Lasenby, Geometric Al-
corresponding to the planet’s eccentric, deferent, and gebra for Physicists (Cambridge U.P., Cambridge,
epicycle. We determined the values of the Fourier coef- U.K., 2003), pp. 11–15.
ficients by imposing that the planet is at its aphelion at [12] Pertti Lounesto, “Clifford algebras and spinor op-
t = 0 and at its perihelion at t = τ /2 = π/ω0 . And from erators”, in Clifford (Geometric) Algebras with Ap-
this we derived Gallavotti’s expression for the Copernican plications to Physics, Mathematics, and Engineer-
orbit in terms of its semimajor axis A and eccentricity ǫ. ing, ed. by W. E. Baylis (Birkhäuser, Boston, 1996),
We also computed the dot and wedge products of pp. 5–32. See pp. 6–13 for a discussion on bivector
the planet’s position and velocity. We showed that for algebra.
small eccentricity ǫ, the dot product is proportional to
[13] Ramon González Calvet, Treatise of Plane Geom-
− sin(ω0 t); the wedge product is constant, ı̂ω0 A2 , which
etry through Geometric Algebra (R. G. Calvet,
implies that the planet’s position vector sweeps out equal
2007).
areas in equal times, as given by Kepler’s second law.
[14] Norwood Russel Hanson, “The mathematical power
Acknowledgments of epicyclical astronomy,” Isis, 52(2), 150–158
(1960).
This research was supported by the Manila Observatory [15] Donald G. Saari, “A visit to the Newtonian n-body
and by the Physics Department of Ateneo de Manila Uni- problem via elementary complex variables,” Am.
versity. Math. Monthly 97(2), 105–119 (1990).
[16] George W. Hill, “Researches in the lunar theory,”
References Am. J. Math. 1(2), 5–26 (1878). See p. 14. The non-
linear terms are µx/r3 and µy/r3 . Hill’s notation for
[1] Hugh Thurston, Early Astronomy (Springer, New ω0 is n.
York, 1994), pp. 208–209.
[17] Victor Szebehely, Theory of Orbits: The Restricted
[2] Claudius Ptolemaeus, The Almagest, trans. Catesby Problem of Three Bodies (Academic, New York,
Taliaferro, The Great Books of the Western World, 1967), p. 608.
vol. 16 (Encyclopaedia Britannica, Chicago, 1952),
pp. 86–88.
[3] Nicolaus Copernicus, On the Revolutions of the
Heavenly Spheres, trans. Charles Glenn Wallis, The
Great Books of the Western World, vol. 16 (Ency-
clopaedia Britannica, Chicago, 1952), pp. 653–657.