# Orbital averages and the secular variation of the orbits

Maurizio M. D’Eliseo∗
Osservatorio S.Elmo - Via A.Caccavello 22, 80129 Napoli Italy Orbital averages are employed to compute the secular variation of the elliptical planetary elements in the orbital plane in presence of perturbing forces of various kinds. They are also useful as an aid in the computation of certain complex integrals. An extensive list of computed integrals is given. Keywords: Unperturbed motion, perturbation equations, third-body perturbations, other force laws, orbital averages, loop integrals.

Introduction

tegral operations on Z
∫ C(Z ) ≡ ∫ H(Z ) ≡ ∫ E(Z ) ≡ ˙ ) i Z = 0. ¯ dt I(r r (5) ˙ Z ) = 0, ¯ dt R(r (4) dt I(¯ r Z ) = 0, (3)

We employ orbital averages for the analytical and numerical determination of the secular part of the variation of the elements a, e, ω (semi-major axis, eccentricity, argument of perihelion) of an elliptic orbit due to perturbing forces. The derivations are developed thorough an averaging process of the ﬁrst-order equations arising from the method of variation of the arbitrary constants. We shall use the formalism of complex variables, as we consider only the perturbations acting in the orbital plane. As a byproduct of our work we deduce some useful methods for the computation of certain awkward integrals related to the geometry of the ellipse.

We so easily obtain three constant functions, two real and one complex. They are ˙r − ˙r ¯) = i(r ¯ 1. Area integral: C(Z ) = 0 → I(r 2 ˙ ˙r ¯)/2 = r θ = c = real const. It follows: dt = r r2 dθ/c . ˙ |2 /2 − µ/r = h, 2. Energy integral: H(Z ) = 0 → |r ˙ ˙r ¯ = µ (2/r − 1/a). h = −µ/2a, r ˙ = 3. Eccentricity vector: E(Z ) = 0 → r (iµ/c)(r /r + e), e ≡ e exp(iω ), the eccentricity vector, is a complex constant, e is the eccentricity and ω is the argument of perihelion. 4. Elliptic orbit in terms of the true longitude: ˙r ¯) = c → r = r eiθ = (c2 /µ) eiθ [1 + e cos(θ − I (r −1 ω )] , a is the semi-major axis, c2 = µa(1 − e2 ), f ≡ θ − ω is the true anomaly, eiθ = eiω eif , df = dθ. Elliptic orbit in terms of the eccentric anomaly √ η : r = r (η ) = (a cos η + ib sin η − ¯ = a(1 − e cos η ) (origin of coordiae) eiω r = r r nates at the center of the ellipse), rdη = an dt. ) (∫ ∫T T ˙r ¯ dt = 5. Third law: cT = 0 c dt = I 0 r (∫ ) ) ( 2π ¯ dr = I 0 r ¯(η ) dr (η ) I r = 2πab =⇒ T 2 /a3 = 4π 2 /µ. √ 6. Other deﬁnitions: b = a (1 − e2 ), n2 ≡ µ/a3 (n is the mean motion ), T = 2π/n is the period of motion. The mean longitude is deﬁned as ℓ ≡ nt.
II. ORBITAL AVERAGES

I.

UNPERTURBED MOTION

We begin with a short review, in the complex notation, of the principal formulas and results of the two-body problem employed in this work. The position of a planet on the orbital plane we suppose lying on the complex plane, is given by the variable r = r (t), where r = r (t) = x(t) + iy (t). Then ¯, r ¯ being the complex conjugate of r . The real r2 = r r and the imaginary parts of a complex number r are de¯)/2 and by I(r ) = i(¯ noted by R(r ) = (r + r r − r )/2. Then R(r ) = R(¯ r ) = I(ir ), I(r ) = −I(¯ r ) = −R(ir ). In polar coordinates r = r eiθ = r(cos θ + i sin θ), θ being the true longitude, measured from the arbitrary ﬁxed axis x. The function r (t) will be known as soon as we found the time dependence of θ, so that r (t) = r[θ(t)] exp iθ(t); we have also ( ) 1 dr ˙ ˙ = r + iθ r . (1) r dt If we write µ = k 2 (M + m) ≈ k 2 M , where M is Sun’s mass which we take as unity, k is Gauss’ gravitational constant, the initial value problem ¨+µ Z≡r r = 0, r3 r (0) = r 0 , ˙ (0) = r ˙ 0, r (2)

is solved if are known four independent integrals of the motion, that can be found introducing the following in-

In presence of perturbations, each orbital element Ei ≡ a, c, e, becomes a function of time, and the perturbation

2 equations are ﬁrst-order equations for the elements of the form ( ) ˙ i = g r, r ˙,t . ˙,r ¯, r ¯ E (6) To obtain the secular part of the perturbations of the elements we average with respect to time the right-hand side of the perturbation equations, and obtain so the sec⟨ ⟩ ˙ ular values Ei . For this, we need to know the mean value of some orbital variables and functions over the unperturbed motion, where with the word orbital we mean periodicity sharing the same period of the elliptical motion. By deﬁnition, the temporal average of a periodic function g (t) over the periodicity interval (0, T ) is g (t)av 1 ≡ ⟨g (t)⟩ = T ∫
0 T

the orbital elements are supposed slowly changing with the time. Mathematically this concept can be treated with the method of variation of arbitrary constants that can be reduced to action, on the generic integral of the ˙ ), of the diﬀerential operator ¯, r ˙,r ¯ motion Ei = Ei (r , r d/dt ˙ ˙ ¯ ∂Ei dr ∂Ei ¯ ∂Ei , ˙ i = dEi ≡ ∂Ei dr + =F +F E ˙ dt ˙ ˙ dt ˙ dt ∂r ∂r ¯ ¯ ∂r ∂r (10) which means to consider each element as variable and to perform the ordinary time derivatives of the integrals of the motion with the convention that1
dr = 0, dt ˙ dr = F, dt (11)

g (t) dt.

(7)

If g (t) is a total derivative of a periodic function h(t), g (t) = dh(t) , dt then g (t)av = h(T ) − h(0) = 0, T

for the perturbed motion, evidencing so only the accelerations produced by the perturbing forces. Thus we ﬁnd, from the integrals ˙r ¯ ), c = I(r 1 1 2 ˙+ , ˙r ¯ − = r a µ r ic ( r) ˙− e=− r , µ r (12) (13) (14)

because of the periodicity of motion. If g is a constant, then ⟨g ⟩ = g . The temporal averages can be also calculated by means of angular variables in the range [0, 2π ] employing the following relations: dt ndt rdη r2 df dℓ = = = = , T 2π 2πa 2πab 2π (8)

the following expressions of the planetary equations in the plane ¯), c ˙ = I(F r a ˙ = 2a ˙ ), ¯ R(F r µ ) i ( ˙ =− ˙ I(F r ¯) , e cF + r µ (
2

applied respectively to the functions g (t), g (η ), g (f ), g (ℓ). In the following we shall give properties, methods of calculation and averages of the orbital functions we need to know. We shall use the notation Fav to indicate the average force exerted by the perturbing planet, while the notation ⟨Fav ⟩ is reserved to the successive average with respect to the orbit of the perturbed planet.

(15) (16) (17)

were we can put ˙ = r ) i n a2 ( r +e , b r ˙ =− ¯ r i n a2 b ) ¯ r ¯ . +e r (18)

III.

THE PLANETARY EQUATIONS

When the motion is perturbed, the equation of motion becomes r ¨ = −µ 3 + F, r (9) r ˙ , t) is the perturbing force in the ¯, r ˙,r ¯ where F = F (r , r plane xy . The force F is central if F = g (r) r , where g (r) is a real function of r. It is assumed that F is of small magnitude as compared to the Keplerian term. Therefore, the planet moves on a weakly perturbed elliptic orbit. The time scales of variation of its elements are a few orders of magnitude longer than the orbital period. Hence, one might perform the averaging of the quantities of interest over fast evolution, the mean anomaly, or any other angular variable according to the relations (8). The planet will move along a variable orbit which at every instant t can be described as an osculating ellipse, in which

In the right-hand side of these equations, in the ﬁrstorder of approximation, all the elements are considered as constants. The equation for c ˙ is useful in the treatment ¯) = 0, of central perturbing forces, because then c ˙ = I(F r ˙ is simpliﬁed to whereby the ﬁrst equation for e
( ˙ = e ) e ˙ ic e ˙ c + iω ˙ e = − F =⇒ + iω ˙ = −i F. (19) e µ e µe

Equating separately the real and the imaginary part we get ( ) ( ) iF ce F ce R = I , (20) e ˙=− µ e µ e ( ) ( ) c iF c F ω ˙ =− I =− R . (21) µ e µ e

A

Third-body perturbations and it follows Gauss’ theorem, because with
dµ′ = µ′ r′2 df ′ dt = µ′ , 2 π a ′ b′ T r′ − r dµ′ , |r ′ − r |3

3

If the force F is central, then ⟨F ⟩ = K e, with K real constant. From this we deduce that ( ) ce ⟨F ⟩ ce ⟨e ˙⟩ = − I =− I(K ) = 0, (22) µ e µ so that we can write ˙ ⟩ = i ⟨ω ⟨e ˙⟩, ⟨ω ˙⟩ = − c c ⟨F ⟩ = − K. µe µ (23)

(28)

we can write
Fav =

In this circumstance the eccentricity vector e rotates uniformly about the origin in the r -plane, with a constant length. In general we can write, by averaging,
⟨a ˙⟩ = ( ) ( 2 a2 n a2 ¯ = R F dr R µT πµ i ic ˙ ⟩ = − ⟨F ⟩ − ¯) dr ⟨e I (F r µ µT ic in ¯) dr , I (F r = − ⟨F ⟩− µ 2πµ ) ¯ , F dr (24)

where the integral is taken along the ellipse of the disturbing body in the direction of motion. So the problem is reduced to that of determining the secular orbital effects of a massive ring whose elementary distribution of mass has the measure given by Eq. (28).6–8 For the analytical determination of this force, we must approximate the irrational factor
( )−3/2 |r ′ − r |−3 = r2 + r′2 − 2rr′ cos(θ′ − θ) )−3/2 ( r r2 = r′−3 1 + ′2 − 2 ′ cos(θ′ − θ) r r ≡ r′−3 ∆1 ≡r
−3/2
−3 −3/2

(25)

,

r′ > r, r>r,

(29) (30)

−3/2 ∆2 ,

in which there appear closed contour or loop integrals2 over the unperturbed ellipse. From Eq. (24) we see that if F is central then ⟨a ˙ ⟩ doesn’t have secular terms, because the integral is a pure imaginary number.

IV.

SECULAR PERTURBATIONS A. Third-body perturbations

Suppose that another planet P ′ is moving on a coplanar orbit around the Sun in the hypothesis that the system P, P ′ be non-resonant, so that the respective mean motions are non-commensurable. If we denote with r , r ′ the position vectors of P and P ′ , and with µ′ the quantity k 2 m′ , being m′ the mass of P ′ , the perturbing force on P is given by3–5
F =µ

where ∆2 is as ∆1 , but with r and r′ interchanged. We develop these expressions in powers of the ratios r/r′ or r′ /r. This requires, when applied to planetary perturbations, a great number of terms for an acceptable convergence, but the hard work is done by computer alge−3/2 1 −1/2 bra. We can also write ∆i = ∆− and expand i ∆i ′ ′ in powers of r/r or r /r only the second factor. These two choices, when applied to the same problem, are equivalent, but they codify diﬀerently the information on the perturbing force in their succession of terms. For our purposes it is more advantageous to use the ﬁrst option in the later treatment of an elliptical perturbing orbit, while for the circular orbit we shall use the second one, that has the advantage to give more compact expressions.
B. Average force of a planet on a circular orbit

−3/2

(

r′ − r r′ − ′3 ′ 3 |r − r | r

) . (26)

The ﬁrst term is the direct force of P ′ on P , while the second term is the inertial indirect force due to the choice of the Sun as origin of the reference frame.5 The determination of the secular perturbations requires a double averaging process: ﬁrst, we average the perturbing force with respect to P ′ and obtain thus Fav , after we average ˙ after the respect to P the perturbation equations for a, ˙ e ˙ ⟩. This substitution of F with Fav , obtaining thus ⟨a ˙ ⟩ , ⟨e procedure is allowed for the ﬁrst-order perturbations, because r does not contain terms depending from P ′ . Notice that the indirect term of the force gives ⟨ a null ⟩ contribute to any secular perturbation, since r ′ r′−3 = 0. We have, by deﬁnition
Fav µ′ = 2πa′ b′ ∫
0 2π

Let be a planet P ′ of mass m′ in a circular orbit, so ′ ′ that r ′ = a′ eiθ = a′ ein t , where n′ is the constant mean motion. Consider the force F e on the point r = reiθ of the orbit of an internal planet P lying in the same plane. With the notation
α = r/a′ < 1, ϕ = θ − θ,

∆α = 1 + α2 − 2α cos ϕ, dϕ = dθ , ( )

(31) (32)

we have
F e = µ′

( ) ′ a′ eiθ − r |a′ eiθ′ − r |3

= µ′

eiϕ eiθ r − ′3 a′2 a ∫ )
0

1 ∆α
3/2

. (33)

Therefore, averaging with respect to P ′
e Fav =

r ′ − r ′2 ′ µ′ r df = |r ′ − r |3 T

0

T

r′ − r dt, |r ′ − r |3 (27)

∫ 2π 1 F ex r′2 dθ′ = 2πa′ b′ 0 ( ∫ r eiϕ µ′ 2π dϕ − = 3 / 2 2π 0 ∆α r a′2

1 2π r a′3

F ex dϕ . (34)

B

Average force of a planet on a circular orbit

4 (0.9833... , 1.0167...), where the formula is meaningless. We have then to order α5 ( ) 75 11 3 µ′ e C − A − B e, (46) ⟨Fav ⟩= 128 a′ a′3 a′5 and so
(35)

In this integral we write
e Fav =

µ′ 2π

0

µ′ ≈ 2πa′2 ( ) α2 3 × 1 + α cos ϕ + + α2 cos 2ϕ + . . . . 4 4

( ) dϕ r eiϕ r 1 − /2 ∆α r a′2 a′3 ∆1 α ∫ 2π ( ) r dϕ eiϕ − α × r 0 ∆α

⟨ω ˙ ⟩ = −κ

′′

c ⟨F e ⟩ , µ e av

(47)

From the formula

0 2π

e±inϕ = ∆α

0

cos nθ 2παn = , ∆α 1 − α2

in which κ is a numerical factor for the conversion from radian/day to arcsec/century given by
(36) κ= 365.25 × 100 × 180 × 3600 ≈ 7.533822048 × 109 . π (48)

with α < 0, n = 0, 1, 2, . . . we see that to order α5 in the numerator only the following terms contribute to the total mean force
e Fav =

) [( ∫ µ′ r 2π dϕ 5 4 3 2 1 − α − α cos ϕ 2πa′2 r 0 ∆α 8 64 ( ) α α3 3 + − cos 2ϕ + α2 cos 3ϕ 2 4 8 ] α α3 3 5 − − − α (37) 2 16 128

Then, recalling that µ′ /µ = m′ , the centennial precession rate is ( ) m′ c 11 3 75 ′′ ⟨ω ˙ ⟩ = −κ A + ′2 B + ′4 C . (49) 128 a′ a′ a a
e If we expand Fav in powers of α, we have ( ) 1 9 r2 75 r4 e Fav = µ′ r + + + . . . , 2 a′3 16 a′5 128 a′7

(50)

so that, after the integration we have
e Fav

µ′ = ′2 a

( ( )) α 64 + 8α2 + 3α4 r 128 (1 − α2 ) r ( ) ′ µ 75 α 3 r = − 11 α − 3 α 128 a′2 (1 − α2 ) r ( ) ′ 3 µ 75 r 11r 3r r = − ′2 − ′4 128 a′ (a′2 − r2 ) a a r

(38) (39) (40)

e It easily seen that also to higher approximations Fav retains the same general structure with the same ﬁrst term and more terms with higher odd powers of the ratio r/a′ . This averaged perturbing force is central, so that we can employ the simpliﬁed perturbation equation for ω ˙. As value of the radius a′ of the circular orbit we take the semi-major axis of the true elliptical orbit of P ′ because the average value of the modulus of the radius vector r′ is a′ to order e′ . We ﬁnd

the ﬁrst term it is proportional only to the position vector r . Thus an external planet exerts an approximately linear repulsive force, directed away from the center, on a particle located somewhere near the center of the orbit. We have to push the approximation as far as the term r r10 because of the relative greatness of the ratio r/a′ for the internal planets Mars, Earth, Venus and Mercury. In the literature9 was obtained with another method the approximation (see Eq. (38))
e = Fav

µ′ r , 2a′ (a′2 − r2 )

(51)

for a numerical evaluation of the classical part of the motion of Mercury’s perihelion. To this regard we note that the computation can be done analytically since we know that in this case by Eqs. (43), (47) we get ⟨ω ˙ ⟩ = −κ
′′

3 ⟨r ⟩ = − a e ≡ A e, (41) 2 ( ) ⟨ 2⟩ 5 (42) r r = − a3 4 + 3 e2 e ≡ B e, 8 ⟨ ⟩ 2 ′ r a(1 − e )(A − B ) − a (A + B )+ 2AB = e (a′2 − r2 ) 2a e2 AB (43) ≡ C e, (44)

c κc ∑ µ′ i e ⟨Fav ⟩=− Ci , µe 2µ i a′

(52)

where
A≡ √ (a + a′ )2 − a2 e2 , B≡ √ (a − a′ )2 − a2 e2 . (45)

A is real for every positive value of a′ , while B is real for a′ outside a deﬁnite interval, that, for the Earth, is

with the obvious meaning of the symbolism. We ﬁnd so the value of 532.53”, very near to the exact value obtained by treating the problem in its full generality. Obviously this is only a coincidence, due to an almost exact compensation among the various planets, but it leaves the false impression that the omitted terms do not destroy this excellent agreement. With the more complete e expression of Fav we obtain the value of 553.97”. The diﬀerence is due to the fact that we have neglected the ellipticity of the orbits of the disturbing planets, other that the mutual inclination of orbits, which in this case is rather signiﬁcant.

C

Average force of a planet in an elliptical orbit

5 where the coeﬃcients Hj are power series in γ , given, with ν = 3/2, by
1 r′3 2 H1 = ′3 r 3 H2 = ′3 r H3 = . . . H0 = ( ) 1 + ν 2 γ 2 + ν ′2 γ 4 + . . . , ( ) ν γ + ν ν ′ γ 3 + ν ′ ν ′′ γ 5 + . . . , ) ( ′ 2 ν γ + ν ν ′′ γ 4 + ν ′ ν ′′′ γ 4 + . . . , (60) (61) (62) (63)

We consider now the situation in which the orbit of P ′ is internal to that of P . With
∆β = 1 + β 2 − 2β cos ϕ, β = α−1 = a′ /r < 1, (53)

we have, at the order β 3
i Fav

µ r = 2π r3 = µ′ r 2π r3

∫ ∫ ∫ ∫
0

2π 0 2π 0 2π 0 ′2

µ r ≈ 2π r3 µ′ r ≈ 2π r3 = −µ′

) dϕ ( iϕ β e − 1 3 /2 ∆β ) 1 dϕ ( iϕ βe −1 1/2 ∆β ∆β ( ) ) β2 dϕ ( iϕ βe −1 1 + β cos ϕ + ∆β 4 β 2 /4 − 1 r β2 − 4 dϕ = µ′ 3 ∆β r 4(1 − β 2 )
2 ′2 2 2

with
ν′ = ν (ν + 1) , 2! ν ′′ = ν (ν + 1)(ν + 2) ,... 3! (64)

r a − 4r r (a − r ) − 3r = −µ′ 3 r3 4(a′2 − r2 ) r 4(a′2 − r2 ) ′ µ r 3 r 1 =− + µ′ . 4 r3 4 r (a′2 − r2 )

By developing the above expression of F e after the substitution
(54)

cos jψ =

¯j + r ¯′j r j r ′j r , ′ j 2r rj

(65)

By developing in powers of β we ﬁnd
i = −µ′ Fav

we can write the force as the sum (55) where
(56)
F0 = F2 = ∑
j 2j h0 j rr

r 3 r + µ′ a′2 5 + . . . . r3 4 r

Fe = F0 + F1 + F2 + F3 + ..., µ′

j

(66)

Now we have (m′ ≡ µ′ /µ)
⟨ ⟩ c ⟨ i ⟩ 3 eiθ ⟨ω ˙ ⟩ = −κ Fav = −κ m′ c , µe 4 a′2 − r2

1 r′2j +3 ¯′ r

,

F1 = F3 = F5 =

2j h1 j r

r′ r′2j +3 r ′2

,

(67)

j

where
⟨ eiθ a′2 − r2 ⟩ = a(1 − e2 )(A + B ) − a′ (A − B ) e ≡ D e, 2a a′ e2 AB (57)

2 2j h2 j r r

r′2j +5 ¯′2 r r′2j +7

,

j

¯r2j h3 j r

r′2j +5 r ′3

, (68)

F4 =

j

3 2j h4 j r r

,

j

2 2j h5 j r r

r′2j +7

, (69)

with A, B deﬁned as before, so that 3 ′′ ⟨ω ˙ ⟩int = −κ m′ c D. 4
C.

(58)

and so on. The coeﬃcients hi j (i, h = 0, 1, 2 . . . ) are rational numbers. After the average with respect to P ′ , e Fav will have terms proportional to
¯ ′k e a′2j +n−k r i r2j , e′ k a′2j +n−k ¯i r2j . r

Average force of a planet in an elliptical orbit

Let us suppose the orbit of P ′ elliptical. Then Fav depends also from the mutual geometrical disposition of the two orbits, i.e. from the angular distance ω ′ − ω of the respective perihelia. If P ′ is external, with ψ = f ′ − f + ω ′ − ω, γ = r/r′ , we have
F e = µ′ µ′ (r ′ − r ) r′ − r = ′ 3 |r − r | (r2 + r′2 − 2 rr′ cos ψ )3/2 µ′ (r ′ − r )

where n = 3, 5, 7 . . . . From this we are lead to the e following conclusions as regard to Fav : • All terms go to zero as the ratio i, j, k, n → ∞.
ai+2j a′2j +n−k

for

0 • Fav is a force of the central type that coincides e (with e′ = 0) with the expression of Fav already found for the circular orbit. 1 • Fav gives the contributes of order e′ , while all the successive terms give contributes containing the product of powers of e, e′ . 2 3 4 5 i • Fav and Fav , Fav and Fav , and in general Fav and i+1 Fav give contributes of the same order of greatness, so that they must be calculated always together.

( µ′ (r ′ − r ) 9 = ≈ 1 + γ2 ′3 3/2 ′ 3 2 r 4 r (1 + γ − 2γ cos ψ ) ) 15 2 225 4 γ + · · · + 3γ cos ψ + γ cos 2ψ + . . . + 64 4 ∑ = µ′ (r ′ − r ) Hj cos jψ, j = 0 , 1, 2 . . . (59)
j

C

Average force of a planet in an elliptical orbit

6
II III IV V N • Fav and Fav , Fav and Fav , and in general Fav and NI Fav give contributes of the same order of greatness, so that they must be calculated always together. ⟨ i ⟩ At last, we must ﬁnd Fav and loop integrals of the form

As regard to ⟨a ˙ ⟩, the demonstration that at the ﬁrst order this element does not have secular terms arising from the planetary perturbing force here is immediate only for the 0 i central part Fav . The other forces Fav will give little positive and negative contributes, that in the long run counteract themselves. e Now we insert Fav in the perturbation equations and we get, for the secular part, ( ) a2 n e ¯ , ⟨a ˙⟩ = R Fav dr (70) πµ ic in e e ˙ ⟩ = − ⟨Fav ¯ ) dr , ⟨e ⟩− I(Fav r (71) µ 2πµ and, dividing by e,
⟨e ˙⟩ in ic e ⟩− + i ⟨ω ˙⟩ = − ⟨Fav e µe 2πµe
e ¯) dr , (72) I (Fav r

, r2j +n−1 ( i− 1 ) r I 2j +n−3 dr , r
V.

¯ r i dr

, r2j +n−1 ( i+1 ) r I 2j +n−1 dr . r

¯ i dr ¯ r

(80) (81)

APPLICATIONS

e so that we must ﬁnd ⟨Fav ⟩ and loop integrals of the form

¯, r i r2j dr ( ) I r i−1 r2j +2 dr ,

¯i r2j dr ¯, r ( i+1 2j ) ¯ r dr . I r

(73) (74)

Considerations of the same type can be made when the orbit of P ′ is internal. Then we have the force Fi = F O + F I + F II + F III + . . . µ
FO = ∑
j o kj

(75)

After the formal development of the planetary perturbing force, we are left with the practical application of the formulas. For the secular perturbations in the planetary problem,10,11 we have some possibilities, each depending from the concrete problem at hand. So, after the choice of the order of approximation, we can proceed ﬁrst symbolically, and then we shall have the characteristic structure of each of the particular terms considered, and after the successive numerical determination we shall have the contribution to the secular variation of the same term. All this it is possible because we are in a linear environment: ﬁrst-order perturbations, integrations of a sum of elementary terms, real and imaginary parts determinations. We can verify the work done with a direct de˙ ⟩. For this it is required the termination of ⟨a ˙ ⟩ and ⟨e numerical computation of the double integrals
a m′ 2π 2 a′ b b′ ∫
0 2π∫ 2π

˙ ) r2 r′2 df df ′ , ¯ R(F r ˙ I( r ′ r ¯) r c (r ′ − r ) + |r ′ − r |3 |r ′ − r |3 )

(82) r2 r′2 df df ′ , (83)

r r2j +3

r′2j ,

FI =

j

I kj

1 r2j +3

r ′ r′2j , (76)

i m′ 4 π 2 a a′ b b′

0

0 ∫ 2π( 2π 0

F II =

j

II kj

1 r ′ r′2j , r2j +3 ¯ r r2j +5

F III =

j

III kj

r ¯′ r′2j , r r2j +5 (77)

2

where ˙ = r ) iµ ( r +e . c r (84)

F IV =

j

IV kj

r ′2 r′2j ,

FV =

j

V kj

r2 r2j +7

r ′3 r′2j , (78)

At last we get, for a century, a ˙ sec = 36525 · ⟨⟨a ˙ ⟩⟩ , ⟨ ⟩ ˙ e e ˙ sec = 36525 · e · R , e ⟨ ⟩ ˙ e ′′ . ω ˙ sec =κ·I e (85) (86) (87)

and so on, where are rational numbers. Then the i typical terms of Fav are e′k a′2j +k r , 2j +n−i ri ¯′k a′2j +k e ¯i r r2j +n−i . (79) for for

N kj

• All terms go to zero as the ratio i, j, k, n → ∞.

a2j +n−k

a′i+2j

O • Fav is a force of the central type that coincides i (with e′ = 0) with the expression of Fav already found for the circular orbit. I • Fav gives the contributes of order e′ , while all the successive terms give contributes containing the product of powers of e, e′ .

For a numerical veriﬁcation of the method applied to the motion of the perihelion, we preferred to consider the Earth instead of the usual Mercury, because the coplanarity of the orbits involved is best veriﬁed for the former planet. The results for the Earth to order α5 are given in the following tables, where we have employed the planetary elements given in the Appendix for the epoch January 1, 2000. In the second table, relative to the Earth’s perihelion in arcsec/century, the ﬁrst column is referred to the full classical approximation, while the others give respectively the contributions, computed by our method, of the circular (by Eq. (50)) and elliptical parts, and their

A

Other force laws tions to t, m, r:16
−1

7 ( ) ( 1 2 α) t → t0 γ ≈ t0 1 + β =t 1+ , (88) 2 r ( ) ( 1 α) µ → µ0 γ −1 ≈ µ0 1 + β 2 = µ0 1 + , (89) 2 r ( ) ( 1 α) r → ˙r0 γ ≈ r0 1 − β 2 = r0 1 − , (90) 2 r

sum. The diction ”Theory” is referred to complete published calculations, which however make reference to a slightly diﬀerent epoch.12
Planet Mercury Venus Mars Jupiter Saturn Uranus Neptune Total Theory (A) e = 0 e ̸= 0 -13.75 3.30 -15.01 345.49 508.73 -151.97 97.69 21.05 75.52 696.85 709.79 -12.95 18.74 32.82 -12.85 0.57 0.60 -0.03 0.18 0.18 0.00 1145.77 1276.47 -117.29 Total (B) -11.71 356.76 96.57 696.84 19.97 0.57 0.18 1159.18 (B-A) 2.04 +11.27 -1.12 -0.01 +1.23 0.00 0.00 +13.41

Secular motion of the Earth perihelion

We have a rather good agreement between the two sets of results. The discrepancies are mainly due to: 1) to having neglected the non-coplanarity of planetary orbits, 2) to having used elements related to diﬀerent epochs, 3) the order of approximation considered. As a further example we consider the secular advance of the perihelion of Mars. Here ”Theory” refers to the computation of Doolittle.13
Planet Mercury Venus Earth Jupiter ,Saturn Uranus Neptune Total Theory (A) e = 0 e ̸= 0 0.62 0.64 -18.01 49.48 47.52 -2.2 229.03 208.73 6.03 1247.24 1468.28 -214.38 66.77 63.29 3.40 1.20 1.13 0.07 0.34 0.34 0.00 1594.67 1789.90 -207.75 Total (B) -14.71 45.32 214.76 1253.90 66.69 1.20 0.34 1582.15 (B-A) -0.96 -2.2 -14.27 +6.66 -0.08 0.00 0.00 -12.52

to the order β 2 , where the last is a pure general relativistic eﬀect because involves the radial distance r, and ˙ |2 /c2 = (2µ)/(c2 ) = 2α/r, where we have put β 2 ≡ |r 2 with α ≡ µ/c deﬁning the gravitational radius of the mass M . We make these substitutions in the equation of motion by dropping the zero suﬃxes ( ) µ 1+ α ¨ r r r r ¨+µ 3 → ( r (91) )2 + ( )3 = 0, r 1+ α r3 1 − α
r r

and, to the O(α) we have: ( ) ( ) µr 2α ( 3α α) ¨+ 3 1+ r 1+ = 0, (92) 1+ r r r r µr 6µαr ¨ + 3 = − 4 = F. r (93) r r ⟨ −4 ⟩ The perturbing force is central, with r r = e/(2 b3 ), so that c 3αc 3αnab ⟨ω ˙ ⟩′′ = −κ ⟨F ⟩ = κ 3 = κ 2 3 µe b c b 3αn =κ 2 (94) c a(1 − e2 ) For the centennial motion we have ∫ T 3αn 6πα ′′ ∆ω = κ dt = κ 2 . 2 a(1 − e2 ) c c a (1 − e2 ) 0

Secular motion of the Mars perihelion

(95)

with the same restrictions as in the previous table. In conclusion, are worth noting the signiﬁcant corrections to the secular motion of the perihelion due to the presence of ellipticity of the perturbing planets and to the neglect of the relative inclinations of the orbits. It is also evident that Venus and the Earth, for their respective proximity to Mercury and Mars, would require the introduction of more higher-order terms in the development of their disturbing force.

We see from the above developments that the variations with the speed of the planet of time, mass and radial distance give a respective contribution of 1/3, 1/6 and 1/2 to the perihelion precession. Almost inverse-square law. Let us suppose that the gravitational law goes as r−(2+ϵ) , where 0 < ϵ ≪ 1. Then by expanding in powers of ϵ 1 ϵLn(r) − + ..., 2 r r2 and we have the perturbing central force r(2+ϵ) ≈ ϵLn(r) r . r2 r The secular perihelion’s motion is given by ⟩ ⟨ ϵnab Ln(r) iθ c ′′ , ⟨F ⟩ = κ e ωav = −κ µe e r2 ∫ ϵ nab eiω 2π Ln(r) 2 if ⟨ω ˙ ⟩′′ = κ r e df 2πab e r2 0 ∫ 2π nϵ Ln(r) eif df =κ 2π e 0 F = −µ 1 (96)

A.

Other force laws

(97)

We examine now the eﬀects of some other force laws,14,15 beginning from: General relativity (GR). The general relativistic correction to the gravitational law in the ﬁrst approximation can be obtained introducing modiﬁcations to the ˙ |/c, classical equation of the motion. If we put β = |r where c is the speed of light in vacuum (173.144 AU/day), √ and set γ ≡ 1 − β 2 , GR requires the following correc-

(98) (99)

A

Other force laws
( Ln(r) ≈ Ln(a) + e cos f − e2 ) + ..., (100) B. The lunar apse

8

Now

5 cos 2f + 4 4

so
⟨ω ˙ ⟩′′ ≈ κnϵ

(

) 1 1 1 4 5 6 7 8 + e2 + e + e + e + ... . 2 8 16 128 256 (101)

Let us calculate ϵ for a tentative explanation of the nonclassical perihelion shift of Mercury. We ﬁnd
42.95′′ = ϵ · 2.71937 · 108 =⇒ ϵ = 1.579 · 10−7 . (102)

As last application of the method of the averages, we apply them to the derivation at order m3 of the part of the motion of the lunar perigee independent from the eccentricity, where m = n′ /n is the ratio of the mean motions of the Sun and the Moon, in the hypothesis of the main lunar problem (the Sun in a circular orbit in the same plane of Moon’s orbit). The perturbing force ( ′ ) r −r r′ F = µ′ − , (112) |r ′ − r |3 r′3 with r ′ = a′ eiθ = a′ eiℓ becomes, with µ′ /a′3 = n′3 and neglecting the solar parallax,
( ) r F = n′2 (r ′ − r ) 1 + 3 ′ cos(θ′ − θ) + . . . − n′2 r ′ a r ≈ −n′2 r + 3 n′2 r ′ ′ cos(θ′ − θ) a ) ′ 3 ′2 iθ′ ( i(θ′−θ) ′2 = −n r + n e r e + e−i(θ −θ) 2 ′ ′ 1 ′2 3 1 3 ¯e2iθ = n′2 r + n′2 r ¯e2iℓ . (113) = n r + n′ 2 r 2 2 2 2
′ ′

Weber’s Law. Now we brieﬂy study a proposed alternative to the Newton’s law, Weber’s law,17 and apply it to Mercury perihelion. This law it is interesting, other than for historical reasons, also because it introduces additional terms, containing the temporal derivatives of r, to the inverse-square law. By equating the two expres˙ we ﬁnd sions of r c iµ iµ −iθ dr +i = + ee , (103) dt r c c and, taking the real part ) µ dr µ ( µ = R i e e−iθ = e sin(θ − ω ) = e sin f, dt c c c (104) d2 r µ µ = 2 e cos(θ − ω ) = 2 e cos f, dt2 r r because, from the area integral d c d = 2 . dt r df Weber’s law is [ ( )2 ] 2 d2 r r − 2 r 2 3, c dt r (105)

In the unperturbed motion at order e, we have for the Moon’s orbit in terms of the mean longitude ℓ r 0 = aeiℓ − 3 1 ¯ e2iℓ . ae + ae 2 2 (114)

In the ﬁrst approximation, by solving the perturbation equations, we ﬁnd the evection, given by the following terms18 δr = −
′ ′ 45 15 ¯ m e2iℓ + ae a e m e2(ℓ−ℓ ) . 16 16

(106)

(115)

1 F = −µ 1 + 2 c

dr dt

(107)

where c is the speed of light. Substituting for the derivative and introducing the true anomaly f = θ − ω we have ( ) e µ2 eµ 2 eµ F =− 2 2 − cos f + cos 2 f eif , c r 2c2 r 2c2 (108) and we ﬁnd
Fav = − e 2πab ∫
0 2π

In the second approximation we put r = r 0 + δ r in the equation } i { ˙ =− ˙r ¯ )F + r ˙ I(F r ¯) , e I(r (116) µ ˙ = ndr /dℓ. We cannot use Fav in this equation, with r ˙ are present terms containing ℓ′ . We ﬁnd, because in r , r at order e, the following constant terms ( ) i 3 2 135 3 ˙r ¯ )F = i − 2 3 I(r m + m n e, (117) n a 4 32 i 45 ˙ I(F r ¯) = i m3 n e, − 2 3r (118) n a 16 and periodical functions of ℓ, ℓ′ that give a zero contribution to the double average. Thus
˙ ⟩⟩ = ⟨⟨e 1 4 π2 ∫
0

µ2 c2

(

) eµ 2 eµ − cos f + cos 2 f eifdf 2c2 r 2c2 (109)

µ2 = − 2 3 e, c b

a complex constant. The secular perihelion motion is given by c nµ ′′ ⟨ω ˙ ⟩ = −κ ⟨Fav ⟩ = κ 2 , (110) µe c a(1 − e2 ) and numerically for Mercury nµ ∆ω ′′ = κ 2 = 14.32′′ . c a(1 − e2 )

∫ 2π 2π
0

˙ dℓ′ dℓ = i e

(

) 3 2 225 3 m + m n e, 4 32 (119)

so that from Eq. (23) ( (111) ⟨e ˙ ⟩ = 0, ⟨ω ˙⟩ = ) 3 2 225 3 m + m n. 4 32 (120)

A

Analytical Methods

9 The calculations of an average will be done by adopting the more convenient variable for the situation at hand. If n = 0, it will be used the eccentric anomaly
⟨m, 0⟩ = am eimω 2π ∫
0 2π

The m3 term is that found for the ﬁrst time numerically by Clairaut and algebraically by D’Alembert in their respective theories of lunar motion, and this solved an apparent problematic aspect of the lunar orbit evidenced since the publication of the Principia of Newton.
VI. ORBITAL AVERAGES - B A. Analytical Methods

√ (1 − e cos η )(cos η − e+ i 1 − e2 sin η )m dη.

If m = 0, it will be convenient employ the eccentric anomaly for n > 0, and the true anomaly for |n| < 2. So we have, for (0, n) and for (0, −n)
⟨rn ⟩ = 1 2πa ∫
0 2π

rn+1 dη = ∫
0 2π

an 2π

(1 − e cos η )n+1 dη,
0 n−3

Some averages are immediate. So 1 ˙⟩ = ⟨r T ∫
0 T

⟨ −n ⟩ r =

1 2πab

df

rn−2

=

a 2πb2n−3

0

(1+e cos f )n−2 df.

1 dr dt = dt T

dr = 0 ,

(121)

because this is a contour integral over a closed orbit. In general for the analyticity of the integrands we have ˙ ⟩ = 0, n ≥ 0. From the orbital expression of r ˙ we ⟨r n r have at once ⟨r ⟩ ⟨ ⟩ ⟨ ⟩ ⟨ ⟩ = eiθ = eiω eif = − e → eif = −e, r (122) ⟨ n⟩ ⟨ ⟩ r = − r n−1 e, (123) r ⟨ n ⟩ ∫ r einω 2π inf = e df = 0, n = 1, 2 . . . . rn+2 2πab 0 (124) Many averages we must compute are of the type ⟨ m ±n ⟩ r r ≡ ⟨m, ±n⟩ ,

For m + n + 2 > 0 the computation with respect to the true anomaly requires the integral ∫ eimω 2π imf m+n+2 e r df, (130) 2πab 0 with r = a(1 − e2 )/(1 + e cos f ), that it can be done by means of the repeated use of Cauchy integral formula2 for the derivatives. If we substitute in the integral eif → s, e cos f → we obtain
[ ]m+n+2 ∫ 2a(1 − e2 ) 1 s2m+n+1 (132) m+n+2 2πiab e |s|=1 [(s − p)(s − q )]

e(s2 + 1) ds , df → , 2s is

(131)

(125)

with m, n ≥ 0 positive integers. We can limit ourselves to take m ≥ 0, because every combination of r , r can be reduced to one of the precedent type employing the ¯/r2 : equality r −1 = r
¯ m r −m r ¯−m r±n = r ¯m r±n−2m = r m r±n−2m . r −m rn = r (126)

where |s| = 1 is the unitary circle centered at the origin and where √ √ 1 − e2 − 1 1 − e2 + 1 p= , q=− , e √ e 2 1 − e2 . (133) (p − q ) = e are the solutions of the equation in s s2 + 2 1 s + = 0. e e (134)

An important property of these average is that they are of the form ⟨ m ±n ⟩ r r = K em , (127) where K is real, because we have ∫ ⟨ m ±n ⟩ e eimω π r r = g (r) e±imf df, 2πab e −π

The pole within the circle |s| = 1 is p, of order m + n + 2. Then by Cauchy integral formula ∫ 2πi f (k−1) (p) f (s) ds = , (135) k (k − 1)! |s|=1 (s − p) we can write (m, n) = [ ]m+n+2 (m+n+1)¨ 1 2a(1 − e2 ) f (p) , ab e (m + n + 1)! (136)

(128)

where g (r) is a periodical even function of f, and the integral of the imaginary part of g (r) e±imf is zero, because this function is odd. In particular, if F is of central type, F = g (r)r , we have ⟨F ⟩ = K e. It is also evident that in general ⟨ m ±n ⟩ ⟨ ⟩ r r = eimω eimf rm±n ∼ em am±n . (129)

with f (s) = s2m+n+1 . (s − q )m+n+2 (137)

A

Analytical Methods ˙ we have from the orbital expression of r ⟨ ⟩ ⟨ ⟩ ⟨ ⟩ ˙ 1 1 1 ic r =− − r e r µe r √ 1 cn 1 − e2 − 1 =− + = . a e µe ae From the expression
⟨r m rn ⟩ = eiω 2π a ∫
0 2π [(

10

This formula fails for m + n + 1 < 0, and in this circumstance we resort to the calculation of ](m+n+2) [ ∫ einω 2π inf 1 + e cos f df e , (138) 2πab 0 a(1 − e2 ) for m − n + 2 > 0. Some results:
(√ e2 )n ( √ ) 1− −1 n 1 − e2 + 1 n ⟨n, −n⟩ = e , (139) e2 n (√ )n 1 − e2 − 1 ⟨n, −(n + 1)⟩ = en , (140) ae2n { ∫ 2π 1 n=0 einω einf df = ab ⟨n, −(n + 2)⟩ = 2πab 0 0 n > 0, (141) (142) ⟨1, −4⟩ = ⟨1, −5⟩ = ⟨1, −6⟩ = ⟨1, −7⟩ = ⟨0, −1⟩ = ⟨0, −2⟩ = ⟨0, −3⟩ = ⟨0, −4⟩ = ⟨0, −5⟩ = ⟨0, −7⟩ = ⟨0, −9⟩ = e , 2 b3 ae , b5 (12 + 3e2 )a2 e , 8 b7 2 3 (4 + 3e )a e , 2b9 1 a 1 , a2 (1 − e2 )1/2 1 , a3 (1 − e2 )3/2 e2 + 2 , 2a4 (1 − e2 )5/2 2a5 (1 3e2 + 2 , − e2 )7/2 (143) (144) (145) (146) (147) (148) (149) (150) (151) (152)
2

(160)

a cos η + ia

)]m √ 1 − e2 sin η − a e × (161)

× [a(1 − e cos η )]n+1 dη

with m, n ≥ 0, we ﬁnd
a (2 + e2 ) , (162) 2 2 2 a (2 + 3e ) (0, 2) = , (163) 2 3 (1, 0) = − a e, (164) 2 a2 (4 + e2 ) (1, 1) = − e, (165) 2 3 2 5a (3e + 4) e, (166) (1, 2) = − 8 3a4 (8 + 12e2 + e4 ) (1, 3) = − e, (167) 8 5 4 2 7a (5e + 20e + 8) (1, 4) = − e, (168) 16 9a7 (35e6 + 280e4 + 336e2 + 64) e, (169) (1, 6) = − 128 9 8 6 4 2 11a (63e + 840e + 2016e + 1152e + 128) (1, 8) = − e. 256 (170) (0, 1) =

15e4 + 40e2 + 8 , 8a7 (1 − e2 )11/2 35e + 210e + 168e + 16 , 16a9 (1 − e2 )15/2
6 4

(153)

315e8 + 3360e6 + 6048e4 + 2304e2 + 128 , 128a1 1(1 − e2 )19/2 (154) ⟨ ⟩ ⟨ ⟩ iθ if iω ⟨r ⟩ = re = re e , (155) ( ) 3 ⟨r cos f ⟩ = ⟨r cos(θ − ω )⟩ = R ⟨r ⟩ e−iω = − a e, 2 (156) ( ) −iω ⟨r sin f ⟩ = ⟨r sin(θ − ω )⟩ = I ⟨r ⟩ e = 0. (157) ⟨0, −11⟩ =

Sometimes it is possible to obtain the same result more easily by means of a clever use of already known relations. So, from the immediate averages ⟨ ⟩ ∫ 2π 1 1 1 = dη = , (158) r 2πa 0 a ⟨ ⟩ ∫ ˙ r 1 T dr dt 1 2πi dr = = = = ni, r T 0 dt r T r T (159)

(1, −1) = −e √ 1 − e2 − 1 (1, −2) = e, ae2 (1, −3) = 0, e (1, −4) = , 2 b3 ae (1, −5) = 5 , b 3 a2 (4 + e2 ) e (1, −6) = , 8 b7 ( ) 3 2 a 4 + 3e e , (1, −7) = 2 b9 ( ) 3 a5 5 e4 + 20 e2 + 8 e (1, −9) = , 8 b13 ( ) a7 35 e6 + 280 e4 + 336 e2 + 64 e (1, −11) = , 16 b17 5 (2, 0) = a e2 , 2 (2, −5) = 0, 3 e2 (2, −7) = , 4a5 (1 − e2 )7/2 (2, −9) = 5(e2 + 2)e2 . 4a7 (1 − e2 )11/2

(171) (172) (173) (174) (175) (176) (177) (178) (179) (180) (181) (182) (183)

A

Analytical Methods
⟨ ⟨ eiθ a′ + r eiθ a′ − r ⟩ = ⟩ a(1 − e2 ) − A + a′ e, a e2 A (184) (185) (186)

11 Some examples:
⟨ ⟩ dr 1 r = dt T

a(1 − e2 ) + B − a′ = e, a e2 B ⟨ ⟩ ⟨( iθ )⟩ r 1 e eiθ = − a′2 − r2 2 a′ − r a′ + r

r dr =

1 2T

1 r2 ¯+ dr r 2T r dr = −π i a b e,

r dr , (199)

r dr = π i a b e, r2 ¯=2 dr r r dr −

(200) (201)

a(1 − e2 )(A − B ) − a′ (A + B ) + 2AB = e, 2a e2 AB (187) ⟨ ⟩ ⟨( iθ )⟩ eiθ 1 e eiθ = + (188) a′2 − r2 2 a′ a′ − r a′ + r a(1 − e2 )(A + B ) − a′ (A − B ) = e, 2a a′ e2 AB with A≡ √ (a + a′ )2 − a2 e2 , B≡ √ (a − a′ )2 − a2 e2 . (190) (189)

r dr = 3 π i a b e,

and

⟨ ⟩ dr 1 r = n i a b e. dt 2

(202)

¯) dr and f (r , r ¯ ) dr Loop integrals of the type f (r , r over an ellipse in the complex plane are present when ˙ . This is ˙ or r ¯ the expression to be averaged contains r ˙ dt = dr in the avaccomplished by using the identity r ¯) r ˙ ⟩ and ⟨f (r , r ¯) dr/dt⟩, together with the erages ⟨f (r , r expressions
2π , T = n iµ r iµ ˙ = r + e, c r c ( ˙ =µ ˙r ¯ r 2 1 + r a ) ,

Curve rectiﬁcation. From √ |dr | 2µ µ = − , (203) dt r a √ ∫ 2π √ 2π µ |d r | = 2 ar − r2 dη n 2πa3 0 ∫ 2π √ =a 1 − e2 cos2 η dη = ellipse lenght.
0

(204) Some other integrals:
√ 2πi( 1 − e2 + e2 − 1) dr √ = e, (205) r e2 1 − e2 nπi dr e, n = 2, 3. (206) = rn [a(1 − e2 )]n−1 √ rn 2nπia(1 − 1 − e2 )n ((1 − e2 )3/2 + e2 − 1) √ dr = (−e)n+1 . n r e2n+2 1 − e2 (207)

(191) ) ¯ R(¯ r dr ) dr 1 ( ˙ ic iµ iµ r ¯+r ¯r ˙ = = rr =− + + e. dt 2r rdt r c c r (192)

These are easily computed when is is possible to use, in the integrand, the eccentric anomaly in orbit’s parametrization. We put r = (a cos η + ib sin η − ae) eiω , r = a(1 − e cos η ), dr = −(a sin η − ib cos η ) eiω dη. (193) (194) (195)

In these expressions we may assume that ω = 0, e = e, i.e. that the semi-major axis of the ellipse lies on the real axis.
( ) ¯ = π i a 2 b e2 − 2 , r r dr ( ) ¯ = π i a 3 b e2 − 2 , r r2 dr ( ) 1 π i a5 b 9e4 − 8e2 − 8 , 4 ( ) 1 6 ¯ = π i a7 b 25e6 + 30e4 − 72e2 − 16 , r r dr 8 ¯= r r4 dr ¯ = − 4 π i a 2 b e, r 2 dr ¯=− r 3 dr 15 π i a3 b e2 , 2 (208) (209) (210) (211) (212) (213) (214) (215) (216) (217) [ rn r ˙ dt =
0

It is immediate to see that, when the force F is central, ¯ is a pure imaginary number, because the integral F dr
¯= ¯= F dr g (r) r dr [ ] 2π (b2 − a2 ) g (r) sin 2η − a2 e sin η − i(abe cos η + a b) dη, 2 0 (196)

¯ = −14 π i a4 b e3 , r 4 dr ¯=− r 5 dr 105 π i a5 b e4 , 4

and, since g (r) is an even function of η , the terms containing sin η, sin 2η are zero in the integration. In general we have, with n relative integer, r r dr ∼ a
m n m+n+1 m+1

¯ = 2πia2 b e ¯, r2 dr ¯ = πia4 b (3e2 + 4) e ¯, r4 dr

e

m ≥ 0, m ≥ 1.

(197)
rn dr = 0 because

T

¯ ∼ am+n+1 em−1 r m rn dr

(198)

rn+1 n+1

]T = 0.
0

(218)

A

Analytical Methods and after we calculate the line integral of f . Some examples: (219)
∫∫ dx dy =
D D

12

Loop integrals of the form r n dr , rm ¯ r n dr , rm

∫∫

¯ ∂r 1 dx dy = ¯ ∂r 2i

¯ dr = π a b, r
∂D 2

can easily be transformed to an orbital average in the true anomaly we already have considered. For example, ∫ T n ∫ ˙ rn r 1 2π r n r ˙ r dt = d r = df m m− 2 rm r c r 0 0 ∫ ) iµ 2π r n ( r = 2 + e df c 0 rm−2 r ∫ 2π n+1 ∫ i eµ 2π r n iµ r df + 2 df. = 2 c 0 rm−1 c rm−2 0 (220) The independent variable f of each of these integrals can be transformed, if more convenient, in η by means of the relation df = (b/r) dη . This is an interesting by-product of the calculus of orbital averages: the possibility of alternate computations of certain integrals. Since an angular average of a ﬁnite two-body expression can be done in two ways, with f, η as independent variables, we can choice the most convenient for the calculation, and automatically we have the result of the corresponding integral in the other variable. We denote as dual the two correlated expressions. In practice, the method is the following: given the integral of an expression derived from a twobody orbital function, we can interpret it as an average in f or η , and after the computation, made with the more convenient variable, we have also the value of the dual integral in η or f obtained by using the relations df = b dη, r dη = r df. b (221)

∫∫
D

r dx dy = − r3

π (1 − e − 1 − dr √ = r e 1 − e2

(225) e2 ) . (226)

All these results are purely geometrical, without reference to the underlying dynamical problem we employ as solution device.

VII.

DISCUSSION

We give here only the simplest example, which non requires any integration at all. With α > β, β/α = e < 1 ∫ 2π ∫ 2π dθ 1 dθ = (α + β cos θ)2 α2 0 (1 + e cos θ)2 0 √ ∫ 2π 1 2πa2 1 − e2 2 = 2 2 r dθ = 2 ⟨1 ⟩ α a (1 − e2 )2 0 α (1 − e2 )2 2π 2πα = 2 =√ . (222) 2 3 / 2 α (1 − e ) (α2 − β 2 )3 Last, the Green’s theorem on the complex plane ∫∫ 1 ∂f ¯) dr , dx dy = f (r , r ¯ ∂ r 2 i ∂D D

We have given a rather consistent number of examples of application of the method of the orbital averages, so it is now possible to draw conclusions about its pros and cons. From the theoretical point of view, it reduces the secular perturbation of two gravitationally interacting bodies in complex geometric situations to a succession of ever smaller simple forces each of which provides a contribution that can be computed exactly. Having available a literal expression deepens our knowledge about the various factors that help produce the ﬁnal result. An important point to emphasize is that, with the symbolic formulas, the work can be done once and for all because in actual cases will suﬃce replacing the various symbols with the numerical values of the orbital parameters. We have also harnessed the power of complex analysis to obtain our results in an elegant way. This also paradoxically constitutes the major drawback of the method: it works under conditions of coplanarity or, at most, in situations of almost coplanarity, but we believe that in its scope it allows to obtain interesting results, especially in the calculation of the eﬀects of gravitational forces arising from alternative theories to general relativity. In conclusion, we think that this method represent a useful working tool that can provide valuable services to the researcher in a wide ﬁeld of study.

VIII.

APPENDIX

Tables of the planetary orbital elements.? The ﬁgures are rounded to the fourth decimal. (223)
Jan 1, 2000 a b Mercury 0.3871 0.3788 Venus 0.7233 0.7233 Earth 1.0000 0.9999 Mars 1.5237 1.5170 Jupiter 5.2034 5.1973 Saturn 9.5371 9.5231 Uranus 19.1913 19.1699 Neptune 30.0690 30.0679 e 0.2056 0.0067 0.0167 0.0934 0.0484 0.0541 0.0472 0.0086 ω (rad.) 1.351870079 2.295683576 1.796767421 5.865019079 0.257503259 1.613241687 2.983888891 0.784898126

where D is the simply connected domain bounded by the unperturbed elliptic orbit ∂D, gives immediately, for each computed value of a line integral, the value of a surface integral over the domain and viceversa, given a ¯), we ﬁrst integrate it function g (r , r ∫ ¯) dr ¯ = f (r , r ¯), g (r , r (224)

13
Jan 1, 2000 Mercury Venus Earth Mars Jupiter Saturn Uranus Neptune m′ = µ′ /µ 1.67 · 10−7 2.44 · 10−6 3.01 · 10−6 3.31 · 10−7 9.59 · 10−4 2.87 · 10−4 4.37 · 10−5 5.18 · 10−5 c 0.010473 0.014629 0.017199 0.021142 0.039236 0.053046 0.075179 0.094527 n 7.142 · 10−2 2.796 · 10−2 1.720 · 10−2 9.143 · 10−3 1.470 · 10−3 5.88 · 10−4 2.08 · 10−4 1.05 · 10−4

∗ 1

2 3

4

5

6

7

8

9

10

11

12

13

14

Electronic address: s.elmo@mail.com Battin R.H. An Introduction to the Mathematics and Methods of Astrodynamics. Revised Edition, Aiaa Education Series, (1999). Lang S. Complex Analysis. Fourth Ed., Springer, (2000). Moulton F.R. An Introduction to Celestial Mechanics. 2nd Ed., Art. 197, p.352, Dover Publications, (1984). Murray C.D., Dermott S.F. Solar System Dynamics. C.U.P. (1999). Brouwer D., Clemence G. Methods of Celestial Mechanics. Academic Press, (1961). Boccaletti, D., Pucacco, G. Theory of Orbits. Vol. 2, Springer, (1998). Vashkovyak, M.A. and Vashkovyak, S.N., Force function of a slightly elliptical Gaussian ring and its generalization to a nearly coplanar system of rings. Solar System Research, Volume 46, Number 1, Pages 69-77, (2012). Iorio, L., Orbital Perturbations Due to Massive Rings. Earth, Moon, and Planets, Online First, 8 May 2012. Price M.P., Rush W.F. Nonrelativistic contribution to Mercury’s perihelion precession. Am. J. Phys. 47, 531 (1979). Brumberg, V.A., Evdokimova, L. S. and Skripnichenko, V.I., Secular perturbations in general planetary theory. Celestial Mechanics and Dynamical Astronomy, Volume 11, Number 1, Pages 131-138, (1975) Migaszewski, C., Gozdziewski, K., A secular theory of coplanar, non-resonant planetary system. Mon. Not. R. Astron. Soc. Volume 388, Issue 2, pages 789802, (2008). Clemence G.M., Reviews of Modern Physics 19, 361 (1947). Doolittle, E., The secular variations of the elements of the four inner planets computed for the epoch 1850.0 G.M.T. Transactions of the American Philosophical Society, New Series, Vol. 22, No. 2 (1912). Baryshev, Y. and Teerikorpi, P., Predictions of Gravity Theories. Astrophysics and Space Science Library, 1, Volume 383, Fundamental Questions of Practical Cosmology, Pages 111-130, (2012).

15

16

17

18

19

20

21

22

23

24

25

Lin-Sen Li, Parameterized post-Newtonian orbital eﬀects in extrasolar planets. Astrophysics and Space Science, Online First, 28 April 2012. Ney P., Electromagnetism and Relativity. Harper & Row, New York, (1962). Roseveare, Mercury’s Perihelion from Leverrier to Einstein. Clarendon, Oxford (1982). D’Eliseo M.M. The eccentricity vector of the Moon. Chinese J.Phys. 3, (2012) Stewart, M.G. Precession of the perihelion of Mercury’s orbit. Am. J. Phys. 73, 730 (2005). Van Laerhoven, C. and Greenberg, R., Characterizing multi-planet systems with classical secular theory. Celestial Mechanics and Dynamical Astronomy, Online First, 23 April 2012 April 2012. Bombardelli, C., Bau, G. and Pelaez, J., Asymptotic solution for the two-body problem with constant tangential thrust acceleration. Celestial Mechanics and Dynamical Astronomy, Volume 110, (2011). Lin-Sen Li, Inﬂuence of the gravitational radiation damping on the time of periastron passage of binary stars. Astrophysics and Space Science, Volume 334, Number 1, Pages 125-130, Number 3, Pages 239-256, (2011). Lara, M., Palacin, J.F. and Russell, R.P. Mission design through averaging of perturbed Keplerian systems: the paradigm of an Enceladus orbiter. Celestial Mechanics and Dynamical Astronomy, Volume 108, Number 1, Pages 122, (2010). Dulliev, A.M., Evolution of Almost Circular Orbits of Satellites under the Action of Noncentral Gravitational Field of the Earth and Lunisolar Perturbations. ISSN 00109525, Cosmic Research, Vol. 49, No. 1, pp. 7281. Pleiades Publishing, Ltd., (2011). Seidelmann K.P. Ed., Explanatory Supplement to the Astronomical Almanac. p.316 (Table 5.8.1), University Science Books, Mill Valley, California, (1992).