The Gravitational Field of a Body with Rotational Symmetry in Einstein's Theory of Gravitation

Citation

Zhou, Peiyuan (1928) The Gravitational Field of a Body with Rotational Symmetry in Einstein's Theory of Gravitation. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/GH13-VA35. https://resolver.caltech.edu/CaltechETD:etd-06282004-092506

Abstract

Einstein's set of field equations in vaccuo

Gμυ = 0

is reduced to such a form that simple problems like the sphere (Schwarzschild's solution), the infinite plane and the infinite cylinder can be solved. The fundamental quadratic differential forms for the latter two cases are respectively

ds ² = - [(1+4πσz) ^-1 dz ² = - [(1+4πσz) ^-1 dz + dρ ² + ρ ² dφ ² ] + 1+4πσz)dt ² ,

ds ² = - c ² ₄ ρ ^-2 [(1+4mlogρ) ^-1 dρ ² + ρ ² dφ ² ] - dz ² + (1+4mlogρ)dt ² ,

where σ is the surface density of matter on the plane, z=0; m the linear density of matter on the cylinder, ρ=const.; (ρ,z,φ) the cylindrical coordinates; c ₄ an indeterminate constant and the velocity of light is unity. Setting g ₄₄ = the Newtonian potential + const., we can get the solution of the general gravitational problem for a body whose mass is distributed symmetrically about an axis provided we can solve

2δ/δψ[(1-2Mψ)δn/δψ] + δ ² /δθ ² e ²ⁿ = 0 (M = mass of the body).

The gravitational field of an oblate spheroidal homoeoid is characterized by

ds ² = - ψ ^-4 (1-2Mψ) ^-1 dψ ² - ψ ^-2 dE ² - ψ ^-2 cos ² Edψ ² + (1-2Mψ)dt ² ,

where ψ = k ^-1 cot ^-1 (sinhη), M = mass of the homoeoid whose equation is c ² ρ ² a ² z ² = a ² c ² , k ² = a ² -c ² and E,η are related to the cylindrical coordinates (ρ,z,φ) by ρ+iz = kcos(E+iη). Analogous expressions for a prolate spheroidal homoeoid are obtainable. The oblateness of the homoeoid causes a slight increase in the advance of the perihelion of a planet's orbit derived from Schwarzschild's solution.

Item Type:	Thesis (Dissertation (Ph.D.))
Subject Keywords:	Physics; theory of gravitation; gravitational field of a body
Degree Grantor:	California Institute of Technology
Division:	Physics, Mathematics and Astronomy
Major Option:	Physics
Thesis Availability:	Public (worldwide access)
Research Advisor(s):	Bell, Eric Temple
Thesis Committee:	Unknown, Unknown
Defense Date:	1 January 1928
Additional Information:	Author's name on title page is spelled P'ei Yüan Chou.
Record Number:	CaltechETD:etd-06282004-092506
Persistent URL:	https://resolver.caltech.edu/CaltechETD:etd-06282004-092506
DOI:	10.7907/GH13-VA35
Default Usage Policy:	No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:	2747
Collection:	CaltechTHESIS
Deposited By:	Imported from ETD-db
Deposited On:	29 Jun 2004
Last Modified:	03 Oct 2019 23:20

Thesis Files

Preview

PDF - Final Version
See Usage Policy.
1MB

Repository Staff Only: item control page

THE GRAVITATiONAL FIELD OF A BODY WITH ROTATIONAL SYMMETRY IN EINSTEIN'S THEORY OF GRAVTrATION THESIS by P 1 ei-Yiian Chou In partial fulfillment of the requirements for the degree of Doctor of Philosophy CALIFORNIA INSTITUTE OF TECHNOLOGY Pasadena, California 1928 THE GRAVITATIONAL FIELD OF A BODY WITH ROTATIONAL SYMMETRY IN EINSTEIN'S THEORY OF GRAVITATION (ABSTRACT) Einstein's set of field equations in vaccuo is reduced to such a form tbat simp~e problems like the sphere (Schwarzsebild' s solution), the infinite plane and the infinite cylinder can be solved. The fundamental quadratic differential forms for the latter two eases are respectively I a is the surface density of matter on the plane, z =. O;· m the linear density of matter on toe cylinder, p const.; (p,z,q>) the cylindrical coordinates;· c 4 an indeterminate constant and the velocity of light is unity. Setting g44 = the Newtonian potential t const., we can ·get the solution of the general gravitational problem for a body whose mass is distributed symmetrically about an axis provided we can solve ~here = 0 on '0 2 a, ae ~[(1-2M'fl)-l + - e n 2 o~ =0 (M =mass of the body). The gravitational field of an oblate spheroidal homoeoid is characterized by where 'I' = K- 1 cot- 1 (sinhr:i), M = mass of the homoeoid whose equation is e 2 p 2 +a 2 z 2 = a 2 c 2 , K 2 = a 2 -c 2 and ~,~ are related to the cylindrical coordinates (p,z,f) by p+iz = xeos(fti~). Analogous expressions for a prolate spheroidal homoeoid are obtainable. The oblateness of the bomoeoid causes a slight increase in the advance of the perihelion of a planet's orbit derived from Schwarzschild's solution. CONTENTS Page I III Introduction 1 Einstein's law of gravitation 1. The field equations 2. Reduction of the field equations 1 4 Remarks on earlier discussions of the problem 3. On Weyl's original solution 4. On Weyl-Levi-Civita's solution 7 9 Fields of sphere, plane and cylinder 5. Schwarzsehild's solution 6. Infinite plane 7. Infinite cylinder 11 13 15 IV General solution of the problem 8. Transformation of the quadratic differential form 16 S. Introduction of the Newtonian potential 18 v Fields of spheroidal homoeoids 10. Oblate spheroidal homoeoi4 22 11. Prolate spheroidal bomoeoid 26 12. Motion of a particle in the field of an oblate spheroidal homoeoid 27 THE GRh.VITATiot~AL FIELD OF A 88:JY VIITH ROTATIONAL SYMMETRY I~ EINSTEIN'S THEORY CF GRAVITATIO~ INTROJUCTION The present paper is an attempt to solve rigorously the problem of the static gravitational field of a body whose mass is distributed symmetricallj' around an axis in Einstein's thecr;>- of gravitation. In I 1 Einstein's field equations in vaccuo~ (~ .1) are set up and reduced in j~ to a forlli such that simple problems like the sphere (,5), the plane l~6) and the infinite cylinder (j7) can be solved. In the general problem there is a fundamental difficulty which will oe avoided oy the introauction of the Kewtonian potential (jil). The solution of the whole problem then jepends upon the solution of a partial differential equation of the second order which is non-linear. Finall~ toe gravitational fielis of spheroidal homoeoi~s (~l~, Ill) are giveL as illustratior,s of the present investigation and the motiou of a particle in trie fiel:i cf an oblate spneroi dal t10moeoid. is discussed( h~). Tte paper also ccntains a critical examination cf earlier works upon the prcblerri notatly tbcse ot fro±.'s E.V:eyl aud T.Levi-Civita (~3, ~4). I EIN3T~IN'J Lh: JF 33AVITlII~K 1. The fielo. egu~tions. Consider the static gravitational field outside of a ~oay wnose mass is distributed s~mmetrically about an axis. Hence the g~v's do not vary with respect to time. The most general fundamental quaaratic differential form in sucn a field appears to be (1.1) where x 1 ,x 2 are any two coorainates in the merideonal plane containing the z-axis, x 8 = ~, the azimuthal angle, x 4 = t, the time coordinate, the unit of tiffie being so cnosen that the velocity of light in vaccuo is unity. fhe gµv's in (1.1) are functions of X1 and X2 only. ~ A.$. ~k, • J.k ~ ~.~ ~ :· ~~ ~.OP-21-),y.81. ---. r- ~ ~ ~ ~~-~4-~~ ~~· . te assume that the values of the gµ v's exist. From a well-known ' theorell~ on positive definite quadratic differential forms of two variables in the parenthesis of (1.1), it is always possible when g 11 , g221 gt2 are explicitly given, to make a real single-valued, continuous transformation from x 1 and x 2 to u and v by (1.2) where J - ~!!!.~ !! "" Ci o(u,v) such that the following identity is true, (1.3) Hence (1.1) becomes where are functions of u and v to be determined. Let m, n, v u = x 11 v = x 2• = g22 = - e2m Then g11 (1.5) g = g t1g 22g33g4 4 = g11 = g22 g33 1 = - e-2m = - e2n I g44 = e2V I - e•m+2n +2v , I gas = - e-2n I g44 = e-2v • Now (1.4) is an orthogonal quadratic differential form. The general expressions of the Cnristoffel symbols of the seccnd kind for such forms are well known.~ In the present problem the non-vanishing symbols are Fl, 11 = mu tll, 2] = - 114, 11 {12, 2} = mu = !Ly \22, 1! = - mu e2n-2m 0 u ~33, 11 = (1.6) f22, 2i = my t33,2"J = _ e2n-2m 0 v {44, 1! = e2v-2mv u l44, 2} = e2v-amv v ~13, 3! = nu ~14,4} l43, 3} = nv ---~-- IDy -~-- It.~- ~, A.~.~v. = vu where for simplicity we denote partial differentiations by subscripts. tritten out in full Einsfein's field equations in vaccuo are (1. 7) of which the following five components are not identically zero: + 112,11112,2} + {12,2n22,2J +h3,3H23,3} + f14,4H24,4} (1.3) =- :: 0 122,1} - ~~22,25 + I21,11f21,1j + 2~21,~lf.22,1} OX1 CX2 + 122,21 i22,2~ + ~23,31~23,31 + 124,41~24,4} ~2 a ' IL \1..J + ~-;logt-g, OX2 .-~ ,. l -:--log ' ~ (-g - ~£GG,1 -::.. dX1 f - G4,4 ,, C·} . ., 1 (-gJ\ ) ~ 1 C' 0 ::--~og OX2 I;_ '7-J If we substitute tne three-index Christoffel symbols of the second kind from their values (1.6) into (1.8), we obtain (1. 9) ( l. lC) ( 1.11) "' 0 22 = muu + rnvv + nvv + V.vv + 02v + v2v + mu(nu+v) r mv(nv+vv) = "", ( 1.12) r, \.l33 - e 2n-2mrl nuu + nvv + nu (nu +v) + nv(nv+vv)J = v, r (1.13) 2. Reduction cf tae field equations. By Putting (2.1) an~ X = n + v, adding toe expressions in the square Lrackets of (l.12) and (1.13) we get 'Y NU U ~hich l2.3) +'Y /\JV V +'Y'2+X2_, /I. U V - " I Lecowes ~ap:ace's equation in the uv-plane, g?uu + ~vv = ~ .,; , on settir~g ~= e/l, = en+v. It is well known that the solution of U;.3) is unique, if the Doun:iary value of if be given in the uv-plane. Tnen U44 "' = v oecou:es r (2.4) which ietermines v. ~e obtain n by (2.1). To get the unknown function, m, we use ll.9), (1.1~) and (1.11). lrite (2.5) = -21 [- XUU +YAVV - n2-v2+n2+v2] U U V V - B. Then oy solving mu and mv simultaneously from (2.5) we get 1 mu = [X2u + 'l~-J- [AvA - ll.uB] / (2.6) Illy = [X.~ + 'X v2 ] - 1 [X u A + AvBJ. It will now be shO\Hl that (2.7) where mu and my are given in l2.6) is an exact differential and secondly that rn rnust satisfy (1.B) and (1.11). By (1.12), (1.13) and (2.2), we note that (2.8) Ii. v - Bu -- - ry 11..u ( !'y /\.uu + 112u + V2) u - 'Y I flvn., from which we obtain immediately (2.9) = - 2Xu'XvA -j{~(/{uu+ n~ + v~) (2.lC) - X~( X.vv+ n~ + v~). By 1ifferentiating mu with respect to v we get ~~mu = [ (2.11) x: .x: T2. [t~~ +x;) ~ A."Av-X...,Bir} t + XvlT c,A~ -X~) -.tXu.'.X1rXl4.11~A -i- + t Xu.v-0~-~ - 'X;) -t- ;:, .X..,X" Avlr ~ l3 ] , which, by (2.2) and (2.S), becomes [X~-+ A.~ r~ [ex.: 1" x;) ~XIA.A.._ + X" B~1 - ~ XllK. C{~ - x;) +z X~v Xv-u.i A + ~ x~"'- ex.: - x:) - z X~X" A"-tA- 1BJ Differentiating mu and mv with respect to u and v respectively we have (2.12) muu = [:X.~ + X: 1-1. [ex~+ A.~) tXv-A14, - It"" Bu.1 + tXir~ + lX111l-n: - (2.13) mvv cx:-x;-J -~X-.,X1r l"'v..\ A x; J+ z ~\A.Xv-Xu.if 1 BJ, = [A_~ + X~J-:t [C:X.~ +/{.;) 1/t~A" + 1"" B" 1- lX\A.v- L7~:.-X~)+.tl"'X1rXvvf A Tl 1.--.-v cx:--x;) - ~X"'!Lv 'il\A.V- ~BJ. On adding (2.12) and (2.13) ana by (~.2) and (2.IC), we obtain (2.14) By (8.14) and (2.2), it is easily seen that (1.&) and (1.11) are satisfiea. This completes the proof that the functions, m,n,v, thus obtained satisfy every component of ~instein's field equations (1.7). For future reference we collect the following inaependent equations: (2.3) 9?uu + ~vv = Ci where (2.4) (2.6) B(1.4) .! ·2 - v2u + n2v + v2J l X uu +'Y /Lvv - nu v / 2 7 II REMARKS ON I::AHLIDH :JISCUSSIOKS OF THE PaOBLEM 3. On 1fo3l 's original solution. The problem under consideration was first at tacked by H. Weyl'-¥. His result was sebsequently criticised b;y '.r.Levi-Civita°"' as being incomplete. The latter started with (0.1) and gave a complete though restricted set of solutions (for reasons see j4 below). The incompleteness of i'ieyl 's procedure seems to lie in tne fact that the formal statement of (C.l) b) the variational principle fails when certain gravitational potentials, gµV' vanish identically and some cf them are equal to eacn other. It is well known that"¥ Oh = (3.1) l-1 o G(-gJ(')) u~ = CH (a~ :;. ax1. · · · dx 4 ) gives the field equations ~ 'J.'~iV ;: _ ~((1J.l.V _ -g·'fA.VG) :;. 0, en proviaed the 6g~v's are entirely arbitrar~. If, however, (to fix our ideas) g1. 2 = C and g 11 = g 22 , .it is unlikely that we have tne rignt to concluae from ~3.1) In ~eyl's work the fundamental quadratic differential form for a body with axial s 1 mmetry is taken to be (1.4) in which gt1 = g221 and gt 11 g 88 , g,, are all functions of X1 and X2· He then constructes G(-g)~ from (1.4) and.varies the action integral, k, with respect to gi1, gss and g,. 4 , which are the only g~v's present in the integrand. The three Eulerian equations (3.2) thus obtained are indeed T11 + T22 :; u, (3.3) T33 = () I while the equation, T12 = 0, whicb is the only component of Tik (i~h) in the present problem not identically zero, is entirely neglected. Moreover it is obvious that the solution of (3.2) satisfies (3.3). • I But ~be converse is in general false. hf ter Levi-Ci vi ta' s critic ism Wey l~ replie<i in Annalen der Physik expecting to justify his previous result Ly assuming the relation (3.4) '3::,:) between the components of the mixea energy-momentum tensor density, which, in bis opinion, keeps the boay in equilibrium. This assumption, however, is inconsequential, because from the physical considerations every component of ~; should vanish in the gravitational field outsiae of the body, ani matbe~atically (3.4) in the present case is merely the first equation of (3.3). In short (3.4) aoes not lead us beyoni tne set (3.3). h plausible ~athematical justification of teyl's result seems to lie in the four Bianchi relations~ (µ = 1,2,3,4) (3. 5) (T 11 )1 + (T 12 )2:: (3.6) ~T21 ~ ,.,_,,22 1 , _ ; - - + ~-- + B.ia.,~}+,~,_P~-c-a)~JT:t7..+111.iJr '+133,~rr.u~ uX1 OX2 ~~0 -+ ~2J4,~ff' 44 -+ [i \;(,1,:t] + C>~.~(-a-)%] ,,~ = o. If equations (3.3) are satisfied, tnese relations become oT11 + ~T1: + OX1 0X2 [ \.11, 1\ + :::~ log(-gf~ - ~22, 11)'f11 OX1 ~. ~ + [2~12,1}+ =~-log(-g)~)T 12 OX2 (3.7) -ar 12 --- OX1 0T 11 --- + OX2 o ~ (l11,2J - --log(-gl' 0X2 = C, f22,2}JT <'l + ~--~O O , g( - 6o)~]cn12 + [ "Sl'' GL 4 1 4~ ~ - 11 r V 1 0X1 ~here the coefficients of T11 and T 12 are functions of x1 and x,. The - T12 --- r h1'ch, 1·r1 fact, is general solutions of (3· • 7 ) are no t ·I·11 = v .. only a very trivial case. Tnis shows that even if (3.5) is introduced here, solutions of te)l's eq~ations (3.3) do not necessarily satisf) 00 .......... every component of Einstein's field equations (3.2). Consequently the condition on the covariance of physical laws as required by the fundamental postulate of the general theory of relativity will not be satisfied generally. 4. On tekl-Levi-Civita's solution. Levi-Civita's solution of (G.1) is a special case of the results we obtained in ~2. Consider (2.3), ana set ~= p. Let ~ be the conjugate function of p. Then p + iz = f (u + iv) (4.1) where f(u+iv) (4.~) dP 2 is analytic in + dz 2 u+iv. From this it follgws that = f'(u+iv)f'(u-iv)\du 2 + dv 2 ); namely, one set of coordinates (u,v) is conformally transformed into tne other (p,z). In order to avoid cumbersome mathematical manipulations in J2, both teyl anj Levi-Civita assume initially that (4.3) p = u, (4.4) ds 2 = r..oreover, ~u44 = vr z = v. Then = p2e-2v e2n and - e2Ili(ap2 + dz 2) - p2e-2Va,q;2 + e2Vdt2. dm'W' become respectively and (4.b) otv 0 2ia 1 0\1 + -2 + - -- = 0 I -ap2 p 'Clp oz (4.6) dm = 0\1 0\1 ;av) 2 \,ov, - ) ]dp + 2p -- a.z • - d + p [ \;:c;p op oz oz 2 \I We recognize (4.5) as Laplace's equation in cylindrical coordinates (p, z, q;) indepenaent cf qi. V:eyl calls P, z in (4.3) the "canonical cylindrical coordinates"'3' which are apparently different from the ordinary cylindrical coordinates used in solving Newtonian potential problems. He tnen emphasizes the f act't' that if the distribution of ma•$ of a given body in our dpace-time manifold is known in terms of this set of configurational canonical coordinates, the probleffi is reduced to the solution 'I.~-~ ··~. ~. &,,..· ~" ,I vk ~, ~, ~- oi li~1~·). ~. w~, ··~. &u ~~." Slf, &. I 3'j (I~ I fS'). ~. w~, "~,~.~~·: CZJ. (r<P-3), ~- '-''· '.IO o~· (4.5). He shows~ that Schwarzschild's solution in isotropic coordinates of a body with mass m having spherical symmetry, corresponds to that of a finite line segment of length 2m, with constant linear density, lying on the z-axis of· the configurational cauonical space-time manifold. But he does not make clear that it is almost impossible to know the corresponding distribution of mass in this canonical-coordinate system when the distribution of mass in our space-time coordinates is given. This difficulty is clearl) brought out by the following argument. When we carr~ out the transformation-from (x1 1 x 2 ,,,t) to (p,z,,,t) by (l.2) and (4.3) we assume onl~ the existence of the values of g 11 , g1 2, g,, in (1.1) so that the transformation is possible, but their explicit forms are not given a prjorj and consequently (l.~) is not explicitly known. Although we know the boundary values of g~v in the original (x 1 ,x 2 ,,~t) system, we do not know the corresponding boundary conditions in the (u,v,f,t)· system oL account of the uncertainty of (1.2). Since (4.5) has an infinite number of solutions if the boundary value of v is not specified, the solution otitainable from (4.5) and {4.6) will not be unique, and consequently it is indeterminate. The same difficulty arises even if we do not assume tne solution of f in (4.3). Here we do not know• which solution of (2.3) we should take in order to solve (2.4). The complexity of the situation is further enhanced by the uncertaint.;- of the boundary conditions of v in the uv-coordinates. An alternative procedure to get a solution for the origiru1l physic al probleffi from ~4.5) acd (4~6) is to cnoose a solution v of (4.~) in terms of the canonical coordinates first and. trier; try to interpret it in ttie (x 1 ,x 2 ,~ 1 t) systelli ~y a transformation (1.~). The gµv's thus obtained must satisfy tne originc,.l bo1.a1a.ar;y ccna.i tious iu terms of (x1, x2, <r, t) given initially. l'ne questiou wnether this procedure will leaa to a unique transformation ~1.2) needs further investigation. It appears not to nave teen considereu in the literature. · ' · t · - t ·o s Following Weyl and Levi-Civ1ta s 1nves iga l n , several authors\8.' have given scli..tions of prolilems in wt;ich distribut._i_o~ masses in ~ >\. w~. "~. ~ f4.'' ~. s-z;.) t 11}0 (1,1v. ~~ ~ ~ ~ ~ ~ *-~ ·~-. -*~k_ ~M ~ {>-°" ~ ~ ( c-tr· -s. .,.. ~, "G>W. "'fl··· M1. 2.<H, ( I "I U ) ) • ~ T ~-~,q~.~.~~,., v.J.. vi, c. dt· 101 u~1~) _; ~. ~) -===er ~ "~. "~. ~~." 1· ~ ... &.&. J!t 6-J. 3'3, rfJ'· 131./- (/~2.Z.):; '1'· '113 (iq")...!y-). l'J) vi:f.i. terms of the canonical coordinates is given. The corresponding mass distribution in our space-time manifold has been never discussed. Hence these solutions are academic and without ph~sical significance. II FI~LJ3 GF 3PHERfu, P~bNE ANJ CYLINJ~R o. Scnwarzschilu's solution. As tbe first applicatio11 of the results in !2 let us ccnsider Schwarzschi:d's solution. The arc element in the gravitational field outside of a body with spherical symmetry is (5.1) :is2 = where (r)e,,) denote spherical polar coordinates ana ~,µ,v are functions of r onl~. (5.1) may be put in the form of (1.4), (5.2) v = 8, (5.3) m,~,v being then functions of u. (= en+v E Rsinv (say). Then d2 R (5.4) a.u2 (2.3) is - R = c. Integrating (5.4~ we obtain = reµ+v = c 1 sinn(u+u0 ) = c 1 sinhu (by fixing u properly in (5.5) H Hence 7v = log~ = loge 1 + logsinhu + logsinv. Then (5.3)). (~.4) becomes d2 v du dv du ~2 + cothu·~ = G. Hence (5.6) (5.7) exp(v/c 2 ) = c 8 [cothu - cschu]. Eliminating u between (5.5) and (5.7) we obtain (5.8) By using the boundary condition on µ ana.' v th~t both of zero as r increases indefinitely we get get .. them tend toward ca = 1. Solving v from (6.8) we or (5.&) From (5.3) we see that m is a function of u only. In (2.6) we must have mv = 0 which gives (5.10) coth 2 u - 1 - vu = c. 2 Relations (5.6) and (5.10) determine , t:. 1) \t,;.1 Take C2 C2 = + 1. = 1. Then from (5.9) we have (5.12) 4ct - ..... e 2 v = 1 - --e r I Eliminating u between (5.3) and (5.6) we obtain (5.13) Tne second case dv eA = - rcschv ~ e ... dr c2 =- 1 only changes Ct to - Ct• Relations (5.12) and (5.13) connect the three unknown functions, A,µ,v. 8onSequently an infinite number of solutions arises. To obtain Schwarzschild's solution we set ~=~.Then (5.12) becomes (5.14) c 1 may be identified. as the mass of the bod~ from Newton's theory. From (5.12) and (5.13) it follows that A = - v. The same result can be also obtained b~ assuming that is 1-2V to start with wnere V is the ~ewtonian potential of the body. Y1here g.. A second solution of interest is the one in isotropic coordinates where the velocity of lignt is independent of direction. Putting A = µ in (5.13) and integrating, we get (5.15) To determine the constant of integration, c., we use (5.12) and let r tend toward infinity. This gives 4c. = - c 1 • (5.lt) Solving fpr ev ana rejecticg the negative root of tially positive from (5.15~ we get ev which is essen- (5.17) Tbis result was also obtained by a transformation of r in Schwarzschild's solutior~. 6. Infinite plane. Let the xy-plane be the given plarie. From symmetry considerations a.round any line parallel to the z-axis the most general fundamental quadratic differential form appears to be (6.1) where A,µ,v are functions of z only. (6.1) can De put in the form (1.4), (6.2) p : (e.3) In the present case n> = i VI m = A., pel\+v _ pP~ arid (~.3) becomes Hence (6.4) (Ly setting (c.5) Iher l:2.. 4) Lecoo.es A.'l. ~, Uo = ~). 14. (c. n 1 dv + - u du =0 which gives ( 6. 7) Cc2,cs = constants of integration). From (6.3) and n = /t- v, v1e find (6.8) mv = A v = t·v, Hy (2.6), we get (6.&) Consider c2 = 1. If we choose the unit of length properly, and e1... = 1. (6 .10) Then c3 = c1 ~ ' ( c.cJ ,. becomes (6.11) ~ifferentiating l6.ll) and on using (6.3) we find (6.12) Here we have au intfinite number of solutions of v and i,;,. To avoia this indeterminateness we use Kewton's theor). d) setting (6.13) g 44 a e 2 v = 1 - ~V = 1 + 4naz and identifying c 1 as 4ilo where cr is the surface density of matter on the given plane, we get µ + v = 0 and the final forffi of (6.1) is (6.14) The additive constant in (6.13) is chosen to be unity. Here we are dealing with a body whose ffiass extends to an infinite distance and ds 2 is not Galilean at infinity. The latter condition, however, can be replaced by the one that space surrounding tne plane is flat if the den- .1.v sity of matter on the plane vanisnes. This is satisfied by (6.14). The solution (6.14) can be regarded as the limiting case of 3chwarzscnild's solution of a spherical shell when the radius of the snell becomes infinitely great (neglecting the infinite constant obtained in this limiting precess). In fact ~hittaker~ uses this metnod to obtain his "quasi-uniform" gravitational field which, as we see in the present discussion, is tne field outsi~e an infinite material plane. The ease c 2 = -1· and hence eA =c 4 u 2 has been treated by Levi-Civita~ and the result extenued to the gravitational fielC1 of a chargeJ plane bs Kar~. 7. Infinite cylinder. Take z as the axis around which the mass of the cylinder is syffimetrically distributed. The most possible fundamental quadratic differential form of such a field in rectangular coordinates is (7.1) where µ,n,h,v are all-functions of p. By using polar coordinates (p,~) in the xy-pla:Je, (7.1) becomes (7.2) ds 2 = _ e2Adp2 - 2 + e2Vdt2 p2e2µd\P2 - e 2n az where e2A = e2t'-+ h2. ( 7. 2) can be put in the form, (?.3) 2 ds 2 = - e2W[du2 + ClV ) d u = p- 1 eni _ ...... d_ p , e 2ndxs 2 + e 2Vd t 2 v -- m t' , e2m -- p2e2tc. ,.. • with X3 = z. Kow m,n,v in (7.3) are functions of u culy. In the present case (2.3) is (7.4) d2i = v du 2 (7.5) ~= Ct (u + Uo) ana r = C1U• Then (2.4) becomes dv -u1 -= \). du r (7.6) E.T~, 'f. ~-~ I s_c_ ~, Hence '' f /\-crt.. ~- ~c." 11 ~. L__\ CA-), ~ "1 ' f-lt. ~." '( Y~. II,, J'- '12-'-- (/'~7). .r-e. 2.'7 > 0 L I;, ~· 240 ( 1~ l'g). VJ. '-~. ~P- ~E u, z-l). ( 7. 7) From (2.6) we have (7.8) From (7.ti), (7.7), (7.8) ana du= p- 1 eA-i...dp we can eliminate the auxiliary variable u and obtain three relations between the four functions, 4,µ,n,v. Consider the case n = 0. Then comparing (7.5) and (7.7) we have (7.9) Ct = cs, c, = 1 and from (7.8) There is still one degree of arbitrariness in the present solution. To avoid this difficulty we resort to Kewton's theory once more. Let ( '1. 10) where m is the mass per unit lengtn of the cylinder. Identify Ct as 2m. Then from \7.3) ana (7.10) (7.11) The final form of ds 2 in the gravitational field outside an infinite cylinaer is coniequently ~ as 2 = - p- 2 c 4 2 ((1 + 4mlogp)-tdP 2 + P d~ J - dz 2 2 2 + (1 +· 4mlogp)dt 2 • Asid~ from the indeterminate co11stant C4 the above solution is quite similar to that of an infinite plane, or of a body with spherical s~m metry. IV GENERAL SOLUTION OF THE P30BLEU 5. Transformation of tbe fundaillental quadratic differential form. The foregoing tnree special cases are solvable from (2.3), (~.4) and (2.6). J..'/ This is because (2.3) degenerates into an ordinary o.ifferential equation in all these ciases. In reality when ~ is a general function of u and v, on account of the uncertainty of the boundary conditions of ~ in the lu,v,f,t) aanifcld as we have pointed out in J4, tne problem can be hardly solvable. In tne following section we shall avoio. this difficulty by introd1..<cin~: tte l\eivtonialJ potential intc tne presellt for1}. . , s t a ;i..1C . u . ·f..~ · - '1','Al..J -z:<i< S'l-..a· ... -°'..._ see• pr""S"'Ylt'\ . ~- ..,. .l.J, -'-t'"' t- ... t"'t'rt_:··' t-' '"" • ... cu 01" t"ne geuera.i. P..""Oi-·.._ gravitational field of a finite ~ody ~ith rctational symmetry can be solvea provided we can solve a nor:-.l.inear partial differential equation of the secona. crder. Ie start with the q·lindrical coordinates (p,z,;p), the z;-axis Leing tne axis of symmetry of the given bca~ which is finite in extent. Consider the merideonal plane containing the z-axis. 011oose in this plane as in ordinc.1.ry potential theory a more general set: u~, T\) which is conformally mapped upon lp,z) by 1 1 z + ip = F(E + iT\) (8 .1) where F(E+ifl) is a monogenic function of f+if\, so that Let v(E,T\) = const., B(f,T\) = const. be two orthogonal (in the Eucliaean sense) families of curves in the plane, to be determined. Jenote partial :iifferentiations by subscripts as in ! 1. Then where (5. 3) - "• - (3.4) I' Choose tne Jacobian ot transformation of (S.3) to be (3. 5) J ~ ~(,,b) (5.6) ef = p. -a(t., ri) 3olving \8.7) = iEBT\ - ~T\eE = ef(•E + w~), where and (3.5) simultaneously we obtain eE, eT\ from ( ~.4) ' Since d6 is an exact differential, (8.7) must satisfy the necessary and sufficient condition, 0 0 (8.5) afeTJ = -eE oT) ' (B.9) '¥t,C, + 'f ri TJ + feo/E + fY\ By (8.1) and (8.6), giving "'ri = ~- f is a known function of E and TJ. 3imple veri- fication shows that (8.9) is Laplace's equation in the (E,ri,~) coordinates independent of f • Kow solve (S.3) simultaneousl~ for dE and dT) in terms of d' and d6 and put the results in (8.2) which then becomes (5.10) Conaequently the f~ndamental quadratic differential form for a flat spacetime continum in the present ('1',8,~ 1 t) variables is (8 .11) ihen matter is present, ds 2 is no more Galilean. We suppose that in such cases (8.11) is replaced by (8.12) where we write (3.13) and A,.,.,y,v are functions of 'I' and 6 to be determined according to Einstein's law of gravitation, ·with the condition that at infinite distances from the body all four approach zero as a limit. 9. Introduction of the Kewtonian potential. Kext transform (2.3), (2.4), (2.6) into the (,,e,,,t) system. Consider the following expressions from (8.12), (9.1) where we put eg = peA-µ. Let (ex+ip)- 1 ~ 0 wnere both ex and p are real be an integrating factor of d\fl + ie-gde, so that d'i' + ie-gd9 = (ex + i~)(du + idv)., (3. 2) and (8.12) becomes (9.3)· C6mparing (9.3) and (1.4) we obtain (9.4) Equating real and imaginary parts in (5.2) we get (9.5) from which the con~itions of integrability ~or 1'1' and cte give av + .3u = v I r'. (S.6) 'l'u = ex I (9.7) au '¥v = eu = eg8 ' I - - Pv = '3g v - agu, -:i. ':-'I ev = egcx; and furthermore "1vv = 8 uu = eg(~u+~gu), 6 vv = eg(cxv+cxgv)• Hence by (B~6) and (S.7) '¥uu + lflvv (&.8) B~ (o.5), (9.7) and (9.3), equation (2.3) becomes ") ( r· ~.~ In the like manner we get (2.4) in the form, (9.10) - ~v;· 'l'uu = au, ov ciL CJ'I' o\fl + _,_ -- + Save the intrduction of A,~ 1 y,v into l3.12) the foregoing dicussions in ~Sand }9 have been purely mathematical. We now use Newton's potential function. Silberstein-Vhas shown tnat (~ .11) where £,'. is tne mass of tl:e Lojy c.nd i1 tl1e rev:tc:,nic,r1 ro te1d,ic;::. pT unit mass, holds rigorously in local or geoaesic coorainates and approximately iq all other coordinate systems at distances large when compared with the dimensions of the gravitational bojy under investigation. Here we assume that (9.11) holds rigorously not onl~ for geodesic coordinates but also for all other systems of reference. (9.1~) then becomes (9.12) 02: ov ol{I oif -~ - -- -= giving G, or (5.13) where e(e) is an ar~itrary function. At infinity wnen o/ = ~I ~ = µ = y = v = 0 for all values of &. Hence we have 9(8) ~ 1 and (9.13) can be rewritten in the form, (~.14) >.. + v = y + µ. This condition is evidently satisfied by 3chwarzschild's solution and ty those of the infinite plane (6.14) ana the cylinder (7.12). B;y (9.11) and (e.12) and since i= en+v, (9.15) (&.9) becomes or (~ .1€) Equation (9.15) is also obtainable by transforming (1.12), as we did with (1.13) in (&.lC), and u~ilizing (S.11). Consider {2.6). B~ (2.2), the exact differential (9 .17) can be integrated into the form, 21 ( :::l v. 1~' b). 2m = log(/{,~ + :{~) + ~ + 2 f Pdu + Qdv, where and (9.20) E~pression (9.18) containds only first partial derivatives and is simpler than (2.6). By (3.5) and the inverse relations, (&.13) becomes (~.21) where we define (9. 22) C' = [X; + e gX~l- eg(n,ne + v,ve), 1 2 J' = ~ [X~ + e2g~91-1(n~ + v~ - e2g[ne + v6J). and Between (9.4) ana (9.21) we can eliminate the auxiliary functions m a 2 +p 2 • A recapitulation of results gives (9.23) ~ _ e2V = 1 - 21~1qi' (&.li) 544 \5.&) \721Ji = o, (e .14) A + v = y + µ., (9.1£) '/-~-l"l _ "'' ·)onJ, + ~e2n "'o\jl ' Gfo\jl ~ (36.2 :::; ~ e" = pe Y, (8 .12) (8.13) -. -. 2 =r "I 22 Frow the aobve list we see immediately tnat the solution of the whole problem depends upon tne solution of the non-linear equation (9.16) and a quadrature. Instead of using A in (9.23) it is sometimes more convenient to deal with its differential forms corresponding to (2.6). The following expressions are obtained from differentiating (9.23) and further simp!ified by {C. O)• '"•V • oH = (~.24) + where oA -06 = A' = (9.25) B' = ii FIEL:JS OF SPH~ROIJAL HC~OEOID3 l:J. Oblate spheroidal homoeoid. Let the equation of the homoeoia be (10 .1) p2 z2 -a2 + -c2 = 1, Use spheroidal coordinates, (10.2) ~,fl, defined oy p + iz = Kcos(E + if)) or p = KcosEcoshf), z = KsinEsinh11, Then fl = cosnt. represents a f amil~ of oblate spheroids confocal with (10.1), which is Kcoshfl = a in the famil) and E = const.,- a family of hyperboloids of one sheet confocal with and orthogonal to the spheroids. The Newtonian potential for an oblate spheroidal homoeoid with unit mass is ( 10. 3) sinhfl = coth.ip. fhe function, e, defined by (3.7) may be taken as 23 e = sinl;. ( 10 .4) From (lG.2), (lC.3) ar,d (10.4), 00. ti) In the present case (8.12) is (1~.e) where an::i A,~, Y are to oe deterrrinea, v Leing given t,y (~.11). Ihe equation (B.16) that y must satisfy becomes u~. 7) in whicL we assua:.e that Y is a function of 1t1 alone. If we write ( 1~. 7) becomes which is a, particular case of one of Painleve's irreducible types~ But (10. 7) is solvable in the present case by tne following cnanges of va- . riables, (l:J .8) where the negative sign in du is chosen to make u tend toward positive infinity as ~ approaches positive zero. (l~.7) then becomes i 2R R = (;, ( 1~. ~) -:-2 - ( 1:-,,. lC) R = c 1 sinh{u + uo) = c1sinhu, au giving the solution (uo = 8). 8) (10.8) and (l~.l~), we have t. l. .,~~~,·· 24 ( lC .11) from which, and by the choice of du in (10.8), we see that Ct must be a positive constant. Eliminating eY between du in (10.8) and (10.11) we get = cschu•du, (10.12) - (10.13) 2M'log(l - 21.~iv) = log(cothu - cschu) CtK(l - ZMo/)-td'V which gives c t.K + c2• Eliminating u tetween (lC.11) ana llC.13) we find (lC .14) l,Ci.X: ), / ) Y = -ZlM+ 1 .iOgll- "-:M\f/ + 5 2'(( l-2Mo/ ) logt(e - CtSiDK'V J+ ~ + Ct 2 sin 2 1\'V]N C2, the general solution of (10.7), involving the two arbitrary constants, c 1 and C2• To :letermine C2 let o/ approach zero. Then y tenas toward zero and C2 = Ci. The constant Ct can be identified with lvi/K, which is also a constant of integration in Newton's theory. Putting (lCi.15) in ll0.14) and solving for eY, we obtain (10.16) Obviously ( 10 .16) approaches unity as 'fl tends toward zero. ~ Last, we must obtain" Knowi1~g >..we can getµ.. by (9.14) and (10.16). In order to avoiu cumbersome aifferentiations in integrating (e.23) directl~ we use the transformation of ' in (lQ.8), and furthermore set (10.17) dv = dE. By (9.14), (10.3), (10.5), (lG.E) and (10.16), (l:;.6) can oe written in the form (10 .18) 25 which has the same .form as (1.4), provided (10 .19) By (10.8) and (10.16), we see that ljl can be expressed as an explicit function of u, and (2.6), that must be satisfied by µ, can be computed with the aid of R in (10.10). The quadrature in terms of the u,v variables is quite si~ple. Coupled with the condition that at infinite distances from tne body ~ must vanish, ~ is found to be (10. 20) F'rom (9.14) and (lC.20), f.. is given t~ (1(.,. 21) Again, ds 2 in (lu.e) becomes <10.22) ' 2 :l.S = K- 2 1fi- 2 [1j1- 2 (1-2l/;1j1)- 1 secn 2 TJdrt 2 + K 2 c.tE 2 + K 2 cos 2 ~d<p 2 ] + (l-4M'l')dt 2 • Solving for sinbT) and sinE trow (lG.2) we get 2K 2sinb 2T\ = r 2 - K 0 .,2(p2-z2) 2 + [r• _ ~A + .,.•J-~, n (10.23) where r 2 = p 2 + z 2 • Tohen K is small, these expressions can be expanded in the following forms: 2s1·nn 2n1·= r 2 [l - -p2K 2 + (1-W2) K 4 + • •. '· J, r2 4r• z2 1-w2 w(l-t.>2) • -K + • • •., sin 2 E = -r2 + - ' K4. 2 + 4r 4r 4 K (10.24) It is interesting to o b serve f rom (lr,v • ~'J3) tt1at When K approaches zero, namel), when the spneroilal nomoeoid tends toward a spherical snell as a lin:i t, the line element (lC .22) becomes 3chwarzschild 1 s solution. Furthermore, (10.22) is also the solution of an infinitely thin material disc with mass M and radius K. 11. Prolate spheroidal homoeoid. The treatment of the proalate spheroidal homoeoid is analogous to the preceding protlem. Here the equation of the surface of the body is given by (ll.1) P2 z2 + a12 c2 = 1, c2 > a12. with The spheroidal coordinates E,ri used are aefinea by z + ip = ~cos(~ + iri), giving ( 11. .G) z = Kcosfcoshri, p = hsir1Esi11hq. The Newtonian potential fer (11.1) with unit mass is (11. 2) ljl 1 coshri + 1 = -log-~-~---. 2K cosnri - 1 Solving for ri in terms of ljl from (11.3) we get ( 11. 4) sinhri = cschhljl. 'I'n' e 1·unct·1·or· . ,, e, ...~et"1"11ed b~,, (5.7) tecornes (11.5) e = - cosE ana as 2 given in (5.12) is then ( 11. e) where Ll e- 2 l' = K 4 s i n h ri l s i n h T) + s l n "t::) , 2 e 2 v = 1 - 21/iljl, . 2 . M = mass of 2 (11.1), and y is.assumed to be a function of ljl alone while both A. andµ, are func• yI v we have the relation, (9.14), namely, tiODS Of \jl and e o uuetweeli l\l~I ( 11. 7) A + v .= y + .i.. The equation (9.16) for y in the present case is similar to ( 10 o 7 ) I so the remaining analysis will be similar. Hence (11.e) is (11. 8) wfiere = K- 1'- [sinn 11 + sin ~)- , e Y = K- '41- csch T). e ~" 2 2 2 2 2 2 2 2 1 2 We note that (11.6) is also the solution for a rod of length ~ and mass 1.; l~ing on the z-axis. Similarly when the prolate spheroid.al homoeoij approaches a spherical shell as a limit, (11.5) degenerates to Schwarzschila's solution. 12. Motion of a particle in the field of aa oblate spheroidal hGmoeoid. The fundamental quadratic differential form (lC.22) for an oblate spheroidal homoeo±d also can te written in the form (l~.1) If, for convenience, we put (12.2) IJI = l/r, where it must be remembered that r and 6 are not the r,e used in previous sections, then (12.l) becomes (12. 3) which nas the s~e form as Schwarzscbild's solution. The results worked out in the latter case are immediately applicable to the present problem, provided we interpret the symbols in (12.3) appropriately. The four differential equations, (12.4) dx dx = O' l ' ds ds + ~µv a -~v 1. defining the motion of an infinitesimal particle in the four dimensional continum characterized by (12. 3). areW d 2r -ds 2 (12.5) 2 - re- 2A.(d6) - 2- rsin 2ee- 2A.(.!!!) + e2V-2Av' (~)2=0, ds ds ds ,dcp) 2 = 0, - - - - sin6cos6\ds d6 aq; .. 2cotEI -- -- = c, --r ds ds ds ds + A I(~) 2 ds 2 .... c, dr db d 8 + 2 Cis r ds ds (12.6) d 2q; (l2.7) ds 2 + ~ :dr~'1'P + (12.8) d 2t " dr dt + :::;v'- -- = u, 2 ds ds d.s where e2V = 1 21.~/ r, r A + v = v, v' = dv a.r Instead of using (12.5) we can take (12 o 3) I which can be writ ten as (12.&) . dq; 2 e-2vcdr,2 - ) + r 2 (dEi' - ; 2 + r 2sin 26(-) - e2V(~)2 = - 1. ds ds ds ds Equations (12.7) and (12.6) are immediately integrable, giving respectively (12.10) (12.11) Let the constants of integration c1 and C2 be positive. Eliminating d,lds between ~12.e) and (12.10) we get 2 (12.12) (12.13) ds 2 + r cis ds C2 ~ SL r. -cost1csc v = v, r4 giving (Take Cs > 0) Eliminating ds from (12.lC) and (1~.13) we find (12.14) v;e choose the negative sign in (12.14) to make e decrease when q; in- creases. Put ( 14.15) By (12.13) since r,6,s are all real, we see that quently p is real. Integrating (12.14) we get (12.16) cote = psin(qi - Ji,) c: t c: and conse~ ( &, = cons t. of integration), where J1 is the node, and e is taken to be n/2 when q:i = JJ.. The geoir.etrical meaning of e = n/2 is that z = ~ where the particle crosses the equatorial plane of the oblate nomoeoid (cf.(12.2) and ll~.2)). By using (12.i;i), (12.1~), (12.11), (12.13) and (12.16), we obtain the following relation between t (or r) and ' ' (12.17) (12.15) f ( lji) :: ljl 3 - 1 2 -.~~,'¥ "t'l ..,, 1 c ~-1 ... 3 2 ..,. <;• ' 21 .:.L<.C3 + --:4' + - - · the negative sign i11 (12.17) will ue explained presently. The right han1 side of (12.17) is immediately integrable in terms of circular functions. The rigorous integration of the left hand side in terms of elliptic functions has been discussed by ForsythW' and subsequently by otners. Let cx., ~' y (ex > i3 > y) tie the three roots of f ('!'). Then ~ can lie only within the interval ~ ~ ' ~ y. rnen ' = ~, we have the analogous "perihelion" and when ' = y, the "aphelion". Let ' = , 0 , wnen ~ = ~. Integrating, we nave (12. lG) C2 f f ~ , I Cf (2Mf(1j1)J-~dljl = - Cs f [l + p 2 sin 2 (qi-~)J- 1 d,. t Here we see that since ~ aecre~ses after • = ~, but that ' continues to increase after ~ = , 0 , the negative sign in (12.17) must be taken. From (12.19) we obtain , 1 + enµ. (12.2:;) ljl = y + (~ - y)-1 + dnµ' where µ is defined b~ the equation, ~·~."(A) rj. 14-S- ( 11 u). 30 (1.G.21) in which cr = c3/c2 1 P = [2M(a - y) J-Y~ tic integral of the tirst kind with (12.22) Le.t a""'1 K = the complete ellipmodulus k given by k 2 = (~ - y)/(a - y). From (lC.2)., (10.3)., ll2.2), (12.16), (12.20) and \12.Gl), we obtain the equations of the orbit of the particle in the following forms, p = ~ [1 + p 2 sin 2 (qi-&,) j-~CSC11.o/, (12.23) z = Kpsin\qi-Jb) [1 + p 2 sin 2 (qi-a,)]-.tcotKIJI. The equation 8 = const. (cf. (12.2) and (10.2)) represents the family cf hyperboloids of one sheet orthogonal to the family of spheroids o/ = const. Then (12.16) shows that the maximuE and minimum latitudes of the particle in its orbit are invariable for given initial conditions. The function, ljl, in (12.20) is a Jacobian ell;iptic function of <p. Hence the analogous "line of apsides" of the orbit precesses about the z•axis. The amount of this precession fer the particle to prescribe the orbit once can be calculated in the following manner:Win (12.20) we have so chosen ' ' ' that at perihelion o/ = ~ . ' = ,o.• Then at apnelion o/ = Y, let ' = , 1 • From (12.20) and (12.21), I (12.24) At the next perihelion let ' ~ ~ 2 • The relation analogous to (12.24) is (12.25) ~ading (12.24) and (12.25) we get (12.26) The precession is given by (12.27) 6 = ~2 - ~o - 2n. Solving ' 2 from (12.26) we get (12.28) ~ = tan- 11[-~~E4P~ + crtan(,o: -Jl,)] _ ( cr 1 - crtan ( (fo - Jl,) tan4PK _ $b) _ <¥Jo n 2 • It is interesting to note from (12.13), (12.15) and (12.23) that if the particle lies initiall~ in the equatorial plane of the homoeoid, i.e. d81ds = v when e = n/2, then subsequently e = n/2 and the particle will continually lie there. The approximate formula for 6 in this case c~n be calculated as follows:. Regar ct ( r, ~) as configura tional polar coordinates of the particle. Then (12.3) shows that the motion of the particle in these coordinates is the same as the motion of a cooresponding particle in Schwarzschild's solution. Hence the constants Ct and c2 in (12.11) and (12.lC.) are given by'V •'/r o, i'i (12.29) ...p 2 = (ro2 -: 2 it2)/ ro:, where M is the mass of the homoeoid, ro the semi-lliajor axis, rt the semi-minor axis, ard e the eccentrici t~ of the orbit in the configurational coorainate system. The advance of the perihelion, 6, is given approximately by 4n·- ( 12. 30) -;-) • ro:Cl - e From (l~.3), (l~.23) with \12.31) 1 1 z = C, -1[1( - p2 - - = I.Ji = -cot r K K and (12.2), we obtain 2)~] • " ~hen p2 - K 2 > K 2 , which is obviously satisfied by large values of P, we can expana (12.31) in ascending powers of Kip in the form 1 1 K -1 = -[1 + r p t; 2 p2 4 J K + -- -+ .•• 4 4::, P ]. ~quation (12.31) shows that p is a mcnctcnic function of r, and consequently the value of 6 in (1~.3~), which is primarily for the orbit in the lr,f) configurational coordinates, will hold also in the (p,~). system. Knowing the "sellii-majcr" and "semi-minor" axes, Po, and P1, of the particle• s orbit in the latter sys ten: we can compute the corre.spond. g v a.i.• ue s 0 1· r o ari"'u. r 1 bv,, r12 32). Then (12.38) shows tnat the oolate1n " • 32 ness of the central body causes a small increase in the advance of the perihelion of the orbit predicted from Schwarzschild's solution. This increases vanishes when K = 0, namely, when the oblate spheroidal homoeoid degenerates into a spheri~al shell. In conclusion I wish to thank Prof. E.T. Bell heartily whose interest in this problem and encouragement in the course of the investigation have made this paper presentable.