# AN INTRODUCTORY TREATI E

ON

BY

H. C. PLUMMER, M.A.

LATE PROFESSOR OF ASTRONOMY IN THE UNIVERSITY OF DUBLIN

DOVER PUBLICATIONS, INC.

NEW YORK I.A Lib .•

24

All rrghts reserved

Published In the Umted KIngdom by Constable and Company Lnwted, 10 Orange Street,

London W C 2

ThIS new Dover editaon, first published In 1960, IS an unabridged and unaltered repubh-

cation of the work orlgmally published In 1918 ThIS edmon IS published by special arrangement wIth the CambrIdge UniVersIty Press, the

origmal publisher of the work

Manufactured In the United States of AmerIca Dover Pubhcanons, Inc

180 Vanek Street

New York 14, NY

PREFACE

THIS book is intended to provide an mteoductson to those parts of Astronomy

W ich reqUIre dynamieal troatment To cover the whole of thls wide subjeet, even m a. prehmmary wa.YJ within the Iimrts of a single volume of moderate size would be mamfestly Impossible Thus the treatm.ent of bodies of definIte shape and of deformable bodies 18 enttrely excluded, and hence no reference WIll be found to roblems of eodea or the man as ota of tidal

theory Already the study of stella.r monons IS bnngmg the methods of

eo IS both too recent and too dlstlO<lt In Its subject-matter to find a place here.

Nevertheless the book covers a wider range of su bJ act than has been usual In works of the lund Thereby two advantages may be gained For the reader is 8 the re titlon of ve much the same mtroduoto ma.tter

wluoh would be neoessa.ry If the different branohes of the subJeot were taken

see these brarnohea In due relation to one another Mdld will realIze better that he 11 dealing not wlth several distinct problems but With different parts of wha.t 18 easentl&lly a amgle problem. In an mtrociuotory work It therefore seemed desuable to xna.ke the 800pe as WIde &8 was oompatlble with a rea.aon-

a. e uDlty 0 met J t • more 80 OD aooount of the almoet oornplete abeenoe

1 t

The ftm aU: cha.ptel'l are devoted to preliminary matters, ohiefly connected with the undi.turbed motion of two bodies. Thee. are followed by :fly. ohapters VII to XI dealing with the determination of orbita. ThIS eection 11 lutended to ft,m1l1arize the ree.der with the propertIes of undisturbed motion

whioh are en1arely devoted to this subject. Otbeme it would haTe bttn neceeae.ry to deeoribe in dewl suoh admirably effectl va methods ... P~or Leueohner'. and to lllolude fuUy worked numeriOfol eXNDplee. Her., Q .a... where, the aim haa befm to siva .uoh &n aooount of princlpl. u will be

VI

Pre/au

mstructrve to the reader whose studies III tlns branch go no further, and at the same time one which will help the student to understand more easily the teohmeal details to be met with In more special treatises Though the actual detaIls of practical computation are entirely excluded, the fact that all such methods end In nnmeneal apphcetion has by no means been overlooked A mstmct effort has been ms.de to leave DO formulae In a shape nnsmtable

for translatIon mto numbers The student who feels the need will have no dIfficulty m findmg forms of computatIOn In other works At the sHlme tIme

the reader who Will take the trouble to work out such forms for himself will

be rewarded WIth a much truer mastery of the subject, though he should not disdam what 18 to be learnt from the tradItIOn of practIcal computers

An outhne of the Planetary Theory 18 given in the seven chapters XII to

XVIII The first of these deals exclusl vely wlth the abstract dynamlcal pnnciples which are subsequently employed It IS hoped that thia SynOpsIS

will prove useful m avoidmg the necessrty for frequent reference to works on theoretJ.e&l. mechaw.cs The reader to whom the methods a~ unfarmhar and

who wishes to become more fully acquamted with them may be referred to

Professor Whlttaker's Analytu;al Dynam'&C8, where he wlll also find an mtrodncnon to those more purely theoretical aspects of the Problem of Three

Bodies wluch find no place here To those who are famihar WIth these pnnClples m theIr abstract form. only the concrete apphcatlODs m the follow-

mg chapters may prove mterestmg A. chapter on special perturbations l8

mcludedo Here. as m the determmatlon of orblts, the need for numenoaI examples mdty be felt To have mserted them would have mterfered too

much wrth the general plan of the book..aad they will be found in the more spec18l treatases But It wa.s felt that the subject could not be ormtted altogether, and a concise and fmly complete account of the theory has therefore been given It may seem curious that with the development of a.na.lytl(~al resources the need for these mechanical methods becomes greater rather than. less, but so It 18

Chapter XIX on the restncted problem of three bodies 18 mtended a.s an mtroduetJ.on to the Lunar Theory conta.med m Chapters XX and XXI The dlVlSlon of these two chapters 18 partly arbItrary, for the sake of preservmg a faIr umfomuty of length, but It coincides roughly Wlth the distincnon between Hill's researches and the subsequent development by Professor Brown In. the second a low order of apprexrmation 18 worked out, and It IS hoped that tins Will serve to some extent the double purpose of mahng the

qu a pro y a war

the time has come when he should uire If he has not done so a r

Preface

Vll

whole method clearer and of pomtmg out the nature of the prmcipal terms,

W IC are apt to e entirely 1 en by t e complicated machmery of the systemanc development

The rotation of the Earth and Moon IS discussed In Chapters XXII and XXIII The treatment of precession and nutation IS meant to be SImple

nomical methods of reckoning tame In actual use In the final chapter of the

00 t e teary ate a mary met 0 s of numerical calculation IS explamed ThIS IS necessary for the proper understanding of Chapter XVIII, but It also bears on vanous POInts which occur elsewhere Numerical appheanons find no place in this work But let the mathematacal reader be under no nnsapprehension The ultimate aim of all thea In Astronom IS seldom

attained WIthout comparison WIth the results of observation, and the medium

as complete tall they are reduced to a numerical form ThIS IS a process which often demands Immense labour and In Itself a quite special land of skill It IS Just as essentaal as the manrpulanon of analytIcal forms

Ongmahty In the WIder sense IS not to be ex eoted and Indeed would

defeat the object of the book, which alms at making' It easier for the student

to the ongmal memoirs .A. certain freshness in the manner of treatment IS possible and, It IS hoped, will not be found altogether wantIng Few duect references have been given as a guide to further reading, and this may be regretted. But the opInIon may be expressed that for the reader who IS

faculty of consultmg the hbrary for what he wants WIthout an apparatus of special direenons Sign-posts have their uses, and the expenenced traveller IS the last to despise them, but they are not conducive to a spmt of ongmal adventure

absence of mathemencel ngour may sometames be noticed attention IS given to such questaons as the convergence of senes to be mferred that these pomts are ummportsnf or that the modern astronomer can afford to disregard them But apart from a few SImple cases where the

In

• •

VIll

Preface

reader will either be able to supply what IS necessary for himself or would not bene:&t even If a cntacal discussion were added, such questions are extremely difficult and have not always found a solution as yet It 18 preeisely one of the aams of this book to Increase the number of those who can appreciate this aide of the subject and will contribute to Its elucidsnon

The reader who WIShes to proceed further In any parts of the su bJect to which he 18 introduced In this book WIll soon find that the number of

systematIc treatises available In all languages IS by no means large He must tum at an early stage to the study of orlg1Da.l memoirs It IS not

difficult, to find assistance In such sources as the articles 10 the Encyklopad'&8 der Matl£8maiJt8chen WisselZschaften, whICh render It unnecessary to gtve a

biohogrsphy The subject IS one which has received the attentIOn of the

majonty of the greatest mathemaucians dunng the last two centuries and In which they have found a constant source of inapiranon TheIr works are

generally accessible in a convenient collected form

For the benefit of any student who WIshes to supplement his readmg and has no means of obtaInmg personal adVice, the follOWIng works may be

specially mentIoned

Determ'£natwn of Orbtts and SpsCf,al Perturbasume 1 J Bsusclnnger, Baltwestllmm'lJ/Tl;g iRr Hf/mmelsko'per

(A source of fully worked numerical apphcataons )

2 PfiJjl1CiitilYii8 of the LiC/C Observator!f, V 01 vn

(Oontams an exposraon of A 0 Leuschner's methods)

Planeta'I"!J atnd Luna» TheO'MeB 3 F 'lisserand, Tm",te de mecan'ItqUe oeleste

(The most complete a.ccount of the classical theones) " R Pomcare, l£fOTUI de mican't9.'U8 clleate

"

H Pomesre, M etlwdes '1l.ou'fJelles de mican~ue oeleste C V. L Charher, 1Ae Mechantk des Htmmels

6 7 E. W Brown, An 'I,'n:Qroductory treatue on the lunar theory

(GIVes full references to all the earlier work on the subjeet )

The great examples of the classicel methods in the form of practical apphc&tlon to the theones of the planets are to be found In the works of La Venter (Annales de l'ObaeMJatovre de Pans), Newcomb (Astronomical

Preface IX

Papers of the .d.mencan Ephemens) and Hill (Oollected Works) The most

<:In rl,",...,.",l---~..,,+ .... ~ .. .+_ ~yv\ i.1-.o ,~ ____ 1.._~ ~J:' "0,, ____ ,.( ,_ ____ ..J

-00 '1: , -I:' w __ ~~ , .............. ~ v ..... '''', ...... ''' "''''''v

m the work of H Gylden (Tra~te analyf!tque des orlnte« absoluee des hu~t

planetes pY"IIMtpales) and P A Hansen A.1I these works will repay careful

study, but the suggesaons are not to be taken many resmcnve sense

The author of the present book has the best of reasons for acknowledgmg

his debt to most of the writers mentioned above and to others who are not

mAnt.lnnAfI Rn't'Y\o nf t,ho. n,.roroof <:Iho.",h. h"'~" 'L."",,~ .... __ , 1~_~ .. -n79 .......... ..1 1.._ .... 1.._

s: ~""""'J ~J ~...,"""'" "".1 ~

Rev P J KIrkby, D Se , late fellow of New College, Oxford. Acknowledge-

ment 18 also due to the staff of the Cambndge Dmversrty Press for their

care In the pnntmg It IS not to be hoped, in spite of every care, that no

errors have escaped detection, and the author will be glad to have such as

~. . ..J 1. _1. 1.

ao~Q ~vu. .... u. IU~VU.O.uU uv u~a llVU.L\JC

H. C PLUMMER

DUNBINK OBSERV A.~ORY7 Co DUBLIN,

20 February 1918

. CHAPTER I

THE LAW OF GRAVITATION

SBOT PAGII

1,2 Kepler's laws 1

3,4 The field of force ('entral 2

6 4

8 The apsldal a.ngle 6

9 Oondrtaon for consta.nt apsidal angle .,

10 Bertnmd's theoreDl On closed orbrts 8

11 Summary of results 8

12 Newton's law 9

13 Gravity and the Moon's motaon 10

14 Dnnensions and absolute value of the constant or graVl't.a.tlOIl 10 CHAPTER II

INTRODUCTORY PROPOSITIONS

16 Forces due to a gra.vltatlonal system 11

16 Potential of sphenoal shell 12

17 ttraotJ.on of a sp ere

18 Potential of a. body at & dIstant point 13

u o m

20 The same referred to the centre of mass Hi

21 A theorem of Jacobi 16

22 The Inva.nable plane 16

23 Relative coorcbnates and the disturbing function 17

24 Astronomloa.l umts 19 MOTION UNDER A CENTRA.L A.'fl'RACTION 26,26 Integration in polar coordmatea

27 The elhpao anomahes

28 SolutIon of Kepler's equatlon (fig 1)

21 23 24

xu Oontesu«

SECT PAGS

29 Parabohc monon 26

30 HyperbolIc motion 26

31,32 HyperbolIc motion (repulsrve forc.e) 27

33 The hodograph (fig 2) 30

34 Special trea.tm.ent of nearly parabolic monon 30 OHAPTER IV

EXPANSIONS IN ELLIPTIC MOTION

30 RelatIons between the anomahes 36 True and eccentric anomahsa

33 34

37 Bessel's coeffiOlE"nte 38 Recurrence formulae

36 36

39-41 ExpanslOJls lD terms of mean anomaly

~ Transforma.tIOn from expanSIon In eecentne to mean anomaly 43 Cauchy's numbers

44 An example

45 Hansen's (.oeffiolents

37 40 4-1

46 Convergenoy of expa,USIOm. In powers of e 47 ExpanSJ.on of Bessel's coefficients

43 44

46

CHAPTER V

**REL.ATION6 BETWEEN TWO OR MORE POSITIONb IN .AN ORBIT
**

AND THE TIME

48 Determmateness of orbIt, given mean dIstance and two pOInts 49

49 Lambert'l'! theorem 50

50 Exa.n:una.tlOD of the amblgUlty 51

51 Eu]er's theorem 5~

52 Encke's transformation 63

53,64 Lambert's theorem for hyperboho motIon 51

M Ratlo of focal trIangle to elhpnc sector 57

56 RatIO to parabolIc sector 58

57,58 &tl0 to hyperbohc bector 59

59 A general theorem In approxllnate form" 61

60 Two applIcatlOns Formulae of GIbbs 62

61,62 ApprOXImate raotJ.Qs of focal tnangles 63

CHAPTER VI

THE ORBIT IN SPACE

6~, 64 Defimtlon of elements 65

66 EclIptiC coordlqates 67

66 Equatonal coordInates • 68

67 Change ID the plane of reference 69

68 Effect of precesSIon on the elements 70

69 The loC'UI fict?U 71 Oontents ...

Xlii

CHAPTER VII

CONDITIONS FOB THE DETERMINA.TION OF AN ELLIPTIO ORBIT

SEO! PAGE

70 Geocentric distance and lta derrvatavea "i3

71 Denvabrvea of drrectaon-cosmes 74

'79 T\-'. ,,~hAI. .,.,...,lo ....... _ ._ .~._ .~ ___ .,.J .11 .1. .~ . 17,..

~A._ .. """ ... ~ .... - .. - .... ...... .., .... uJ

73 The elements determmed . 76

74 The NllIJlLt.1nn In the ],,,.11 ._ .~-." .-1 .... +.<'1 .... " .. 76

75 The hmrhng curve (fig 3) 77

76 The smgulsr ourve 80

77 The apparent orbit. Theorem of LR.mbert . 81

78 Theorem of Kltnkerfues 82

79 The small circle of closest contact 82

I 80 Geemetmcal mterpretanon of method 83

CHAPTER VIII

DETERMINATION OF AN ORBIT lIE'l'llOD OF GA.USS

81 Data. of the problem 80

~_,.,_, 10 .

Oli '-JVUUl"lVU Vl ......... u.vu lU ~~ 00

83 The middle geccentric distance 86

All. rrh .. .A ..l. .+ .... 1 A1'1l1".4-, " .... ,..~ 1:1 ....... " QI'7

. .,.. -

80 Fust and last geooentne d18ta.nces 89

86 FLrst approximation 90

87 Treatment of aberra.tlon 91

88 True ratIOS of sectors and tna.ogles 91

89 The solution completed 93

I

OHAPTER IX

DETER1tUNA.TION OF PA.RABOLIO AND Cm017LA.R O:BBITS

90 Data for a parabobo orblt 94

91 (Jonnlt.lon 01 In .. ",1<>. ...... lIA

I:

92 U Be ot Euler's equstaon 90

93 DeduotlOD of -pa,ra,bohc elements ~A

94 The second place as a. teat . I 97

96 Method for ClrCUla.r orbit 98

96 Method of GaUS$ 100

97 Olrcular elements den ved . 101

• · Oontents

XlV

CHAPTER X

ORBITS OF DOUBLE STARS

UCT PAGB

98 Nature of the apparent orbrt 103

99 Apphca.tlon of proJectlve geometry (fig 4) 104

100 Five-point construenons (fig 6) 106

101 Other graphical methods 107

102 Altet DatIve method 107

103 Use of equation of the apparent orblt 108

104 Elements depen<hng 011 the tlme 110

105 SpecIal cases no

106 DUl'erentlal correctIons 11~

107 Ratio of masses 113

108 Use of absolute observatrons 113

CHAPTER XI

ORBITS OF SPECTROSOOPIC BIN ARIEH

109 DoppIer's pnnclple 115

110 Correotlons to the observations ll6

111 Nature of Speett08eOp1C blD&l'1es 118

112 The VelOCIty ourve (ftg 6, a and b) 118

113 SpecIal pomts on the curve 120

114 Analytical solution for elements 121

IUS PropertJ.es of focal chords 1~2

116 PropertlelS of cha.meters 1~3

117 Integral properties of Ve1OC'lty curve 1:a!4

118 DdferenttaJ. propertIes 1~6

119 DIfferentIal correotlODS to elements 126

uo DunecBlOllS and mass tunctJ.ODS of system 1~6

121 ApphcatIon to VISual double stars 127

CHAPTER XII

DYN AHIOAL PRINCIPrJEFI

122 Lagrange's eq uatlons 129

123 The mtegraJ. of energy 130

124 Canomoal equations 131

125 Contact tra.nstolmatlon lSi

126 The Hamllton-Jl;Loobl equation lai

127 Val"Jatlon of a.rbltrary constants 133

128 Hamtlton's pnnciple 134-

129 Prmeiple of least actIon 136

130 La.gre.nge'e and POIsson's brackets 136

131 Oonditaons satIsfied by contact transformation 138

1'32 Infimtesimal contact tra.naformatlon 139

133 Disturbed motton related to an mtegral 140

134 Theorem ot P018BOn 140 Oontents

xv

**VARIA. TION OF ELEMENTS
**

135 Hamilton-J acobi form of solution for undisturbed motion 142

136 Inberpretataon of constants 143

137 'Lagrange's brackets 144

138 POIsson's brackets 145

139 Equanons for the vanatlOllt:! 146

140 Modified definrnon of mean longrtude 147

141 AlternatIve form of equanons for the variataons 148

142 Form involving' tangential system of components 149

ystems 0 canomca VarI8, es

144 Delaunay's method of mtegratnon 153

145 Subsequent tranaformanona 155

•

146 Effect of the process 157

CHAPTER XIV

T.HE DISTURBING FUNCTION

158

148 Formulae of recurrence 159

149 Newcomb's method of ealculating coefhoienta 160

160 Direct calculations required 161

151 Oontmued fractaon formula. 162

152 Jacobrs coeffiolents 163

153 Parttal differentIal equation for coefficients 164

155 167

167 Symboho form of complete development 170

168 N ewoomb's opera.tors 172

159 Indirect part of disturbing funetnon 173

160 Alternative order of de" elopment 174

161 Expholt form of dlsturbmg function 175

CHAPTER XV

162 OrbIt 10 a. resistmg medium 177

163 Nature of the perturbations 178

164 PerturbatIons of the first order 179

165 Seoular and long period mequalraes 180

166 Perturbations of higher orders 181

168 JacobI's coordinates 184

he areal mte Is Ehmmation of the nodes 185

170 Equatlons of mobion 186

171 EqUS.tlODS for disturbed monon 187

172 POIsson's theorem 188

173 Etfect of cornmeneurabihty of mean rnotl(.)tll:l 190 XVI

Oontents

CHAPTER XVI

SECULAR PERTURBATIONS

SEO'!'

174 The dlSturblng funchon modified 175 Form of expanSlon

176 Effect of symmetry

PAGlil 192 193 196

177, 178 ~phclt form of secular terms

179 Orthogonal transformataon of van ables

ISO Solution for ecoentnc vanables 181 Solutl.on for oblique variables

196 199

182 Other forms of the mtegrals

183 Upper lmnt to eeeentnerties and mchnations

~oo 202 203

" 204

CHAPTER XVII

SECULAR INEQUALITIES METHOD OF clAUSS

184 Statement of the problem 185 AttractJ.on of a loaded nng

207 208

186 Geometncal relations between the orbits 187 EquatJ.on of the cone

209 210

188 The fin&! quadrature

189 Introduetaon of elliptic functl.ons

212 213

190 Integrals expressed by hypel'geom.etnc senes 191 The potential In terms of lDvanants

] 92 Transforma.tJ.on of COor<ilDa.teS

214 215 216

CHAPTER XVIII

SPECIAL PERTURBATIONS

193 Nature of apeeial perturbatlons 218

194 The drft'erence table 219

195 Formulae of quadrature.e 220

196 Apphcatlon to a dIfferentIal equation 221

197 .An example 221

198 Method of rectangular coord mates 222

199 EquatIons of monon 10 <,yhndnca.l coordma.tes 224

200 Treatment of the equations 22e;

201 PerturbatIons In polar coordmates deduced 226

202 EquatIOns for va.natlOD8 In the elements 227

201 CalouJ.a.tlOD of dlsturbmg forces 228

204 PerturbatIOns m the elements 229

.20e; Case of p&ra.bohc orbits 230

206 Necessary modlfica.tlon of coeffi(nents 231

207 Sphere of mfluence of a. pla.net 234 Oontents ..

xvn

CHAPTER XIX

THE RESTRICTED PROBLEM OF THREE BODIES

SJIICT PAGE

208 Jacobi's Integral 236

209 Tlsserand's eriterion 236

210 Curves of zero velocity (fig 7) 237

211 Pointe of relative equihbnum 239

..

;nLZi .JYJ,ulIlon In me nelgooouroood 241

213 StabIlIty of the motion 242

" • '1'1. ..l , 243

-...... ... •• ' , c:w. ... "'''' v.lOW

215 Elementary theory of the dIfferential equstron 245

216 Vanatlon of the action 247

217 WhIttaker'S theorems 248

218 Use of cOIlJugate functzons . 260

219 Apphca.tlOIIS . 252

CHAPTER XX

LUNAR THEORY I

220 Choice of method 254

221 MotlOn of Sun defined 254

222 Force function for the Moon 256

223 Eq uaaons of motion 257

224 HIll's transformatIon • 'IF.A

2215 Further transformatIon 21S9

226 Vanatlonal curve defined 261

..... 262

;nZiI ~"l IOr ccemcients

228 More symmetrIcal form 263

229 Mode of solutaon 263

230 Polar ccordrnates deduced 265

231 Another trea.tment of problem 265

232 Bquanon of veried orbit 267

233 HIll's determmant ~-

;lIiQO

234 Properties of roots 269

231S Development of assocIa.ted determllltl.nt c)'7l'\

236 Adams' deternllnatlOn of g 272

OHAPTER XXI

LUNAR THEORY II

237 Small dIsplacements from varIatIonal ourvs .......

.. , ...

238 Fmlte dlsplaoements 274

239 Terms of the first order 9."K

240 The vana.tIon . 276

241 FIrst terms calculated 277

242 MotIon of the pengee 278

243 Pnnclpal elhpae tenn The Eveotlon 279

244 Terms dependIng on sola.r eooentrlclty 280 • • Oontents

XVlll

SECT PAGE

245 The .Annual EquatIon 281

246 The ParallactIc Inequality 283

247 The third coordinate 284

248 MotIon of the node 285

249 Further development 286

250 Mode of treatment 287

251 ConsIStency of equahona 287

252 lhgher parts of motion of pangea 288

253 DefinItlons of arbltra.ry constants 289

254 Rem&lnIng factors ill the lunar problem 291 CHAPTER XXII

PRECESSION, NOTATION AND TIME

255 Euler's equetaons 292

256 Mutual potential of two distant tnasses 293

257 The moments calculated 294

258 Steady state of rotaotlOn 294

259 EquatIons of motion for the &XIS 295

260 Cba.nge of &us for the Mooll 296

261 ExpanslOns for elhptac motion introduced 298

262 Mode of solution 299

263 LUOl-SOlar precession 299

264 General precession (fig 8) 300

265 Nutation 302

266 Nutatlona.l ellipse 303

267 Numenca.l values for preees810n 304

268 Results for nutation Moon's mass 305

269 Annual precesSIOns In & A snd declination 306

270 SIdereal brme 307

271 Mean tame 308

272 TrOpIcal year 310

273 General remark 310 CHAPTER XXnI

LIBRATION OF THE MOON

274 Cassun's laws

275 OptIcal Iibration 276 Equa.tJ.ons of motIon

277 F11'8t eondition of stability 278 Iabra .. non In longrtude

279 EquatIons for the pole

280 Second condmon of stability

281 Thud eondmon for moments of inertia 282 Second order tenns

!S3 Ans of rotatIon

312 312 313 314 316 316 318 319 320 321

Oontents .

XIX

ua A PTER XXIV

FORMULAE OF NUMERICAL CALCULATION

SBOT 1'",om

284 Representation of a. function 323

285 The opera.tors A, a 324

286 Stlrhng's formula 325

""no. T~ , " ,.., 326

"'<;11 ~..,~_ ...... Ul UitIoUlSlS

288 Bessel's formula. 327

C:UIQ T ... ,--,~ ~. _, 328

--g.... - ...................

290 Mechamcal dlfferentIatlOn 329

291 Inverse operstaons 330

292 The first LDtegra.l 332

293 The second mtegral 333

294 Properties of Fonner's series 333

295 Mode of solution for ooefficlents :t.'!t:I.

21:16 .Ifunaamental formulae 335

297 Slmphfioa.tlons 336

--- ~ 337

.:i110 -s;- case _~8-12)

299 Property of least squares 338

300 Periodic function of two vana.blel:l 339

INDEX 341

AN INTRODlTCTORY TREATISE

ON

DYNAMICAL ASTRONOMY

• 1

CHAPTER I

THE LAW OF GRAVITATION

1 The foundations of dynamical .Asfnonomy were laid by Johann Kepler at the begmmng of the seventeenth century H1s most unportant work, .A8tronom~a Nova (De Motlbus Stellae MartIS), published in 1609, contains a profound discussion of the motaon of the planet Mars, based on the obsern

kmemstical relataons of the solar system 1S for the first time revealed

ep er s mam resu s may e summanze us

(a) The heliocentric motions of the planets.u e their motions relative to the Sun) take place In fixed planes passmg through the actual posinon of the Sun

(b) The area of the sector traced by the radius vector from the Sun,

e ween any wo POSI Ions 0 a pane in 1 s or 1 , IS propor rona 0 occupied in passing from one pOSItIOn to the other

(0) The form of a planetary orbit IS an ellipse, of wbich the Sun occupies one focus

These laws, which were found m the first instance to hold for the Earth and for Mars, apply to the individual planets In a later work, Harmonsoes Mundt, published In 1619, another law 1S given which connects the motions

o tel erent p anets toget er IS IS

U e f the OdIC me ro ortional to the cube of the

mean distance (1 e the semi-axis major)

These deducnons fl0m observanon are glven here in the order m which they were discovered The third (c) IS generally known as Kepler's first law, the second (b) as his second law, and the fourth (d) as hIS third law But the first statement IS of equal Importance In the Ptolemaic system. the I( :first

mequality " of a. planet, which represents Its he iocentnc motion, was aeeigne to a lane assm throu h the mean OSltlOD. of the SUD Even In the

Oopermean system this U mean posmon " becomes the centre of the Earth's orbit, not the actual eccentnc position of the Sun In consequence no astronomer before Kepler had succeeded 10 representmg the latitudes of the planets WIth even tolerable success

2

The Law of Gravdat~on

[OB I

2 It 18 undeniable that in making hIS discovenes Kepler was aided by a certain measure of good fortune Thus hIS law of areas was ill reality founded on a lucky combmanon of errors In the first place It was based on the bypotheeie of an eccentric circular orbit and was later adopted in the elliptic theory In the second place Kepler supposed (a) that the trme m a small ,trc was proportional to the radius vector, (b) that the tune In a finite arc was therefore proportions] to the sum of the radn vectores to all the

pomts of the arc, (0) that this sum IS represented by the area of the sector Both (a) and (c) are erroneous, and indeed Kepler was aware that (0) was

not strrctly accurate MathematIcally expressed, the argument would appear thus

hdt = rds, ht = (t ds = 2 (area of sector)

Both the supposed fact and the method of deduction are wrong, yet the result IS fIght But If It should be supposed that Kepler owed hIS success

to good fortune It must be remembered that thib fortune was SImply the reward of unpa.ralleled mdustl'y In exhaastmg the poesrbrrrnes of every

hypothesis that presented Itself and m checking the value of any new pnnerple by dIrect compa.rison With good observatIons It must also be remarked that

Tycho Brahe's observations were of the proper order of accuracy for Kepler's purpose, beIng sufSetently accurate to dlScrrmma.te between true and false

hypotheses and yet Dot so refined as to Involve the problem In a mase of

unmanageable detaJl Another factor m Kepler'S success was hIS knowledge of the Greek mathematrciaua, In particular of the works of ApollonlUs

3 Kepler had no conception of the property of inertia and he was therefore unable to ma.ke any progress towards a correct dynamIcal View of

planetary motaon It IS interestang to analyze hIS results and to see what IS Imphed by each of the above sta.tements taken by itself

Accordmg to the first statement the planets move In a field of force which IS such that every tra.Jectory IS a plane curve If we suppose that the acceleration at each point IS a function of the coordinates of the point, an immediate deduotion can be made as to the nature of the field of force For let A, B be two pointe on a certain trajectory, and let P be a tlurd point not in the plane of this curve Then P can be joined to .A and to B by plane traJectones The planes in which AB, PA and PB he meet In one point 0 (which may be at mfimty) The acceleration at A 18 In the plane OAB and also m the plane OAP Hence It IS along A 0 Similarly the acceleration at B 1S along BO, and the acceleranoa at P IS along PO But the point 0 IS determined by the two pointe .A. and B Therefore the acceleranon at every pomt of the field IS directed towards the fixed point 0, and the field of force 18 central (or parallel) Now the planes of the orbits all pass through the Sun Hence the Sun IS the centre of the field of force In which the

The Law of Gravitat~on

3

planets move For an analytical proof of the general theorem see Halphen

4 To this the second statement adds nothing with regard to the nature of the forces, and might Indeed have been deduced from the first For It tells us that

the Sun being the orlgm of coordinates and h being a. constant By differen-

or

xy-yx = 0

Thus y/x = yl.!, which proves that the acceleration IS towards the Sun at ever oint, 1 e the field of force IS central Cleai l the ar ument mi ht be

reversed, and the law of areas deduced from the fact that the acceleramons re directed towards a fixed centre which has alread been obtamed from the

first statement Both this theorem and its converse are given In Newton's Prvncvpvx, Book I, Props I and II

5 We shall now mvesngate the law of acceleration towards a fixed pomt under which elliptic motion IS possible In the first mstance It WIll not be

h nx 1 t ~ cu f teA art from the

mterest of the more general result, this 19 the more desirable because many

revolve around one another in apparent elhpses, but the plane of the motion being unknown It IS only a matter of inference that either star IS In the focus of the relatrve orbit of the other. For It IS the projection of the motion on a plane perpendicular to the line of SIght which IS observed Let then the ellipse

be descnbed freely under an acceleranon to the fixed point (1, g) .AIJ.y point on the elhpse can be represented by (a oos E, b sin E) The angle E which 18 known m analyncal geometry as the eccentric angle 16 called In Astronomy the eooentno anomaly of the point The accelerationa being

- a, SID E E - a cos E EJ, b COS It' E - b SIn E EJ

along the two axes, we have

whence

E _ ag.!?o~E_-:: blSIn E E

E- ab- agslnE- bfcos E

(1)

'Phe Law oj Graouauo» Ttw. 1'" Il.ll HIt.fIR! ab lo form, gIvIng Immedlately

A" = h (ab - agsmE-bfcosE)-l

[OB I

r

(2)

(tbE + (to COS E - bf eu: E = h (t - to)

}lIlft· h ."uri t" art· <,~()JlMtn.ntb of IntegratIon If we put h= ob»,

It} - f SIn E +g cos E n(t to) (3)

Ct b

ul "u- uan .• \ 1 ... (OOllHlderad a. generalIzed form of wba.t ]8 known as KepleI's

lu .• tl,m H~ lLeldlng 27r to E It 1S evident that 27r/n = T IS the penod of a hul.- 1"1 volut,um K('plor'", form applIes when the motIon. IS about a focus of

If~ .·lhl'~(·' , .. nd <)fl.U ho obttl.lned by puttmg f= ae, 9 = 0, so that

•

E - e SIn E n (t - to) (4)

fUN c·.,IULtlun IIi Hf fundn.ln~ntd.l Importance The pOint for which E - 0 18

I" .. ··,. ... ··.t pUIIlt nn the.' orbIt to the attractmg focus and IS sometames called It~ J~jt ,,.#',,, re, 'rht) oorrespondlng time 15 to and n IS called the mean

(it.,,,, By t I ) ru u J (2) Mu, ('onlpon ents of the acceleratlOn beeome

•• rl HUI It: wE - a cos E E» ab(f -acos E)hP

( ab ag sm ~' bf cos E)8

b (~'IJoj IC E __ b sIn E E' ab (q b sin E) h'J

(a.b - ag sm.E - bj cos E)8 ttimt. t.he· t uta1 uc"oolt)mtlon ]8 equal to

(5)

'if'rt' j* If!! tin .. diKtn.noo of the pomf on the orbit from (f, g)

8. n.·fh,·.~ t;XIl,uuruIlg thIS result more closely, 1'1, may be noticed that the Iltlwd 1·. 'I\Ut,t· gurwut.l and r.nay be applied to any central orbit For If the Imlm"tc'f04 uf n, puult (Q1, '1/) on the curve be expressed In terms of a SIngle nUllnh·r «, W(> hn.v .. , sirmlarly

ala. + x" ci 2 y' (l + '!i" a a

aJ-f =----:Y=g-

a: ai' (y - g) - y" (m - f)

eX - - u/ (y - g) - '!i' (m - f) IX

1.·r~1 .t', s' ,. dttuut;,·o derrvatavce WIth respect to rl, and e, tx denvataves wrth IJMtt-tJ tu th .. t.urw. Hence on rnbegration,

a 1m - h {a:' (y - g) - '!i; (x - f)}-l

J(u dy - ydID) - !y+gx=h(t-to)

The Law of Gravitat'ton

5

By takmg the last integration over one revolution In a closed orbit It IS

components of the acceleration become

~9 (!llyN - m"y') (x -f) and h2{llly" - o)"y') (y - g)

{x' (y - 9) - y' (x - j )}8 tx' (y - g) - if (OJ _f)}8

and the total acceleration IS therefore

R = h?r (x'y" - al'y') {al ts - g) - '!I' (a; - f)}-a

where p IS the radius of curvature at the pomt and p IS the perpendicular from (f, g) to the tangent at the pomt Th18 of course IS the well-known expresSIon for the acceleration towards the centre of attractlOD

The same orbit will be described In the same periodic tnne under the

I

that IS, If

R'IE = p8r'/p'8r

'I'his result IS equivalent to P'Mnc~p'/,a, Book I, Prop VII, Cor 3

7 We now return to equation (5) which may be wntten

(6)

where q and qo are the perpendiculars on the polar of (f, g) from the POInt (11), y) on the orbit and the centre of the elhpse respectively Hence the ellipse represented by the general equation

~ + 2ha;y + by? + 2gl1) + 2fy + 1 === 0 (7)

acceleration follows the law

(8)

where a and a have bheir usual meanmg for the conic ('7) Conversely, If the law (8) lS gIven, the trajectory IS always a come whatever the untial condmons may be For (7) 1$ a posaible orbit, and f and 9 are determined by the law, while a, band h are three arbItrary constants which can be chosen so as to

satIsfy a.ny given condrtione, sue as t e imtia ve oeity gtven In magmtu e and direcnon a.t a artaoular oint

There now ansea the mterestmg quesnon whether any other form of law besidea (8) esists, for which the traJectones are always OODles (Bertrand's problem) Let

6

The Law 0/ Gravitation

[OR I

be such a law Then If (7) is to be an orbit,

f(flJ, y)=(l +gllJ+/y)8

must be aatasfied by the coordinates of every point on (7),1 e this equation must be equivalent to (7) But (7) can be wntten ill either of the forms

1 + gflJ + fy = i (1 - aar - 2h:cy - by2)

(1 + gil) +1y)· = (99 a) a'J + 2 (fg Ji) my + (]2 b) y2

and clearly In no other way which does not Introduce a great&!' number of

Independent cons tan ts on the right-hand SIde The first of these forms gIves an expreSSIon for/(a;, y) which IS (lIke an mfimte Dumber of others) eompatlblt

WIth (7), but only under restricted conditions For It fixes the constants a, b and h and leaves only/and g arbrtrary, and these are not In general ButnClent

In number to satIsfy the imtial conditione On the other hand, the second form gIVes an express]QD for the acce-leratlon whIch may be written

R = m~ (ate2 + 2f3o;y + ry'!/) - i (9)

ThIS only requires the constants In (7) to satISfy the two relafnons

and thus three other relatrons can be satisfled which are required by the mmal condlinons Hence motiOn under a central deeeletatlon gtven by (9)

IS always In a conic which by the two relataons found touches the lines (real or J m a.glDary)

a.a;2 + 2f3a;y + 'Yy2 = 0

The laws (8) and (9) are the only ones under whIch a CODlC 18 always descnbed In a given plane whatever the initial conditions may be 'I'heir

ehara.eter was first establIshed by Darbollx and by Halphen (Oornptes Rsnd'U8, LXXXIV, pp 760, 986 and 989)

8 A pomt on a central orbit at which the motion 18 at ngh t angles to

dr

the radius vector 1S called an apS6 .At such a point de = 0 and the radius

vector IS ill genera.l either a maximum or a minimum Since the motion IS reversible the radius vector to an apse IS an axlS of symmetry ID the orbit and the next apsidal dietancea on either side are equal There can be therefore only two distmct apsidal distances recurnng alternately and the angle between any two consecutive apses 1S constant and IS called the apsrdal angle

The differentaal equation of a central orbrt IS known to be

diu P

d~ +u= h2u2

7-9J

The Law oj Grav~tation

7

where u = l/r and P IS the force to the centre If we wnte P = u'U the

shghtly disturbed, so that we may write u + e Instead of u, where U IS constant and fI} IS so small tha.t only the first power of a; need be retained Then

d2x U' uU' ~ dU

dB2 + a; = 12 IC = U OJ, U = du

If we put

the equation becomes

and the solutaon IS

(J) = a cos m ((J - (0) The apsidal angle IS therefore

K = 71' m = 71' (1 - 11, U' U) - ~

(10)

For example, If P = p.rP, U = p.u-'P-2 and

ThIS result IS gIven m the Prvnoqna, Book I, Prop XI~V, Ex 2

9 Let us push the approximataon further III order to see, If possible, under what conditions the apsidal angle remams unchanged by a higher order of the increment fI} The equation of the disturbed circular orbit

becomes

11

and we assume a solution

:L = ao + al cos m(J + a~ cos 2mB + as cos 3m9

If al IS of the first order. ao and Il,a must be of the second order at least, and It will become clear that as IS of the third order Hence

a? = !u,.1 + (2aOal + al~) cos m + a1100S m + ~o" cos m

rc' = a' cos mB + lala cos 3mB

All terms of order higher than the third have been omitted and products of the cosines have been changed Into Simple COSInes of the mulmple angles. We now subsntute In (11) and equate coefficients Thus

1 »o: a,l~

m'o,o=i'U

1 uU" 1 uU'" a1'

0= - (2ao~ + al~) +-

1 -ar

-SmRas== - U ('-12

4

1 uU" 1 UU"' 0,;3

-Smia.- - U C£,.as + 24 --rr-

2 8

The Law oj Gravitat~on

[OR I

The last of these equataons confirms the statement that as IS of the third order, but WIll not be needed here The :first three after the elmnnanon of

or

5u U"2 + :3 U'" ( U - u U') -= 0

(12)

ThIs squation expresses a necessary condition which must be satIsfied If the apsidal angle IS to remain constant when the dIsplacement from a CIrcular

orbit IS considered finite

10 Let us oonmd.er any closed orbit to be datal mmed by a central

acceleranon under d. fimte rd.nge of mitaal velocities The number of apses In a complete orbit must be fimte and (10) shows that 'In must be a com-

mensurable number It must be a constant therefore, for otherwise It would ch&nge dlscontlDuously as u ehttngas ContlllUOUSly Hence

IS an equatlOn giving' the form of V, and the solution IS U - lev}_""

But If all the orbits are to be re-entrant, so that K IS constant, the equatIon (12) must also be sstasfied Hence substItutIng the form Just

found, we have

or

Smce K IS finite, m IS not zero and we have

1 mil 0 or 1 m'J - 3

gIvmg

U-le

or U -ku-'

R = lc/rt or R = kn

Thus" e have Bertrand's remarkable theorem (Oom,pte8 Rendsu, LXXVII, p 849) that these are the only laws, expressible as functions of the distance, which always gIve rise to closed orbits whatever the imnal CIrcumstances may be (within A certain range) In these two cases m = 1 or 2 and the apsidal angle K = 'Ir or 17T

11 The results obtained can DOW be brought together Accordmg to Kepler's law the planetary orbits are elhpses WIth the centre of attraction, the Sun, situated In one focus The polar of the focus being the correspondmg directns, we have 1D (6) qo = ale and q = rle, so that the acceleration towards the Sun IS

(13)

When the centre of attra.ctlOn IS an arbitrary point and It IS merely known that the orbits are elhpses, the ecceleramon towards the centre must

9-12]

The Law of Gravitation

9

follow one of the two laws expressed by (8) and (9) These are not In general

simp e uno Ions 0 Infer from the apparent orbits of double stars that these bodies obey the law gIven by (13) But the law (8) provides a SImple function of the distance, It = m2r) when f = 9 = 0, 10 which case the centres of all possible orbrts are a.t the ongm, 1 e coincide WIth the centre of attraction Similarly the law (9)

rovides a srm 1e function of the distance, R = 'Tn'). r2, when Cl = ry and fJ = 0

In this case every orbit touches the hnes a.ll + y'A = 0, showmg that the centre the focus for ever ath These are the onl

two laws of central acceleration which gIve rise to elhptic orbrts In general and can be expressed In Simple terms of the distance But we have also seen that the same restncnon IS Imposed when It IS merely required that the paths shall be plane closed curves of any kind It IS moreover ObVIOUS that the law of the direct distance, which makes the attraction of a distant body

more e ecta ve an a 0 a near one, eanno only alternative IS that the acceleration vanes inversely as the square of the

distance, and this law can t ere ore e ase upon ese SImp e SUppOSI Ions (a) the planets descnbe closed paths In planes pasSIng through the Sun, (b) the centnpetal acceleration towards the Sun, required by (a), 18 a simple function of the distance and does not become mfimte when the dietance IS mfimte

12 We have now to consider ep er s aw connecting t e peno IC nnes of the lanets WIth their mean. distances from the SUD ThIS states that T2

vanes as as But T = 27T'/n, so that n2a,8 IS constant for all the planets ence by (13) the acceleration of each planet towards the Sun IS lJ./r'A where II- IS constant The force of attraemon actmg on a planet IS therefore mp./r'A where m IS the mass of the planet And observanon shows that the same form ot law holds for the sstelhtes of any planet, e g the satelhtes of Jupiter Thus

to attract their satellites In the same way It IS but natural to suppose that

e orces 0 a rae Ion 1 ,

and that a stress exrsts between the Sun and a planet. or between a planet and Its satellite ActLOD and reaction being equal and opposite, we must suppose the force proportional not only to the mass of the attracted body but equally to the mass of the attractmg body Weare thus led to Newton's law of vitataon that the mutual attractnon between two masses m, m' at

a distance r apart 18 measured by

where G IS an absolute constant, Independent of the masses or their distance It must be noticed that the law has been amved at from the consideration of

--

cases In which the drmensions of the bodies are small In comparIson WIth the

diatances separatlDg them But since the action 10 these cases IS proportional

10

The Law 01 Grav~tation

[OH I

to the total masses, It IS to be supposed that It applies to the individual elements of the matter composmg them ThIS IS the true form of the law of universal gravitation When It IS a question of bodies whose dimensions are not negligible in relation to the dtstances of surroundmg bodies, a modtfication of the SImple statement must be expected The examination of all consequences of the law of gravitation, including a companson WIth the results of observatlOn, practlcally constItutes the complete ftmetaon of dynamIcal

Astronomy

I

13 SInce the Earth possesses only one satellite, It IS impossible to venfy Kepler's third law In our own system But It IS of histone interest to calcu-

late from the observed motion of the Moon the aeeelerataon towards the centre of the Earth which a body would have at the Earth's surface

81 erea peno IS 27d t» 43m lIB 5 01 23605915 sees Let a be the Moon's m.ean distance and b the radius of the Earth The reqUIred aecelerJ,tlon IS

n2a3 = 471"J (~)S

b':J T' _b_ b

The ratio alb 18 602745 and b may be taken to be 6378 x IOscm The result ofsubstltutlng these numbers IS to give for the acceleratlon 989 em Isec II

In point of fact the acceleration of a body at the Earth's surface 18 10 the mean g 981 em /sec ~ But the discrepancy IS not surprising The Moon

descnbes ItS orbit not only under the attraction of the Earth but also under the wstUlbmg mfluenee of the Sun Moreover g IS a varIable quantity over

the Earth's surface, owmg to the Earth's rotation and figure The above calculatIon 18 altogether too rough to gIve really compctrable results But It

suffices to show that the quantity IS quite of the same order CtS g, and to this extent supports the IdentmeatloII of the force whIch retains the Moon in Its

orbit with tha.t which in the case of tenestnal objects IS known as weIght .As stated, the pOint 18 of histoncai mterest bectluse It presented a wfhculty

to Newton who was long misled by adoptmg erroneous numerical data

14. The numerical value of the constant G depends upon the units adopted Its drmensions are given by

G M2L-I=MLT-I

or

G=M-1LaT-2

In C G S units It IS the force between two parmelee each of 1 gramme placed 1 coo apart The :first determmanon of the force m absolute units by a. laboratory experiment was made by Cavendish Several determmataons have smee been made, of which perhaps the two best, those of C V Boys and K Braun: agree ill gIVIng

I G = 6 658 X 10-8

corresponding to 5527 for the mean density of the Earth and 5985 x lOW gr

for the total mass of the Earth •

CHAPTER II

INTRODUCTORY PROPOSITIONS

15 As we have seen, the simple facts of observation lead us to assume tha.t be w

P (m, y, a) and P' (<</, y, i) there exists a force Gmm'jrr2, where r IS the

I ,

(1/ - (1)

z' -z

1 r

and hence the components of the force acting on the particle 'Tn are

,m'-fl'

z'-z

'--

or

where

U=- Gmrn'/r

If 'In IS attracted not by a SIngle particle m' but by any number typrlied 'In, a.t «(1)" h z,) the com onents of the total force are similarl

au

aU

au

where

It IS evident that U IS the work which the system of a.ttracting parncles will do If the particle rn ]S moved from Its actual posinon by any path to

, ,

due to Its poeition relatrve to the attractmg system If we put

V IS called the potential of the attractmg system at the pomt P When the potential IS known It IS evident that the components of the attractaon can be easily calculated

['!II. II

16 'I'lw (OjWI uf .L hmuogtlufltltUl 'ph.'ril'.LI ",h. II 1'. ul' .,1. UU'lItIU,\' 1111

pClrtJ"tu*11 lA'" 111 1». tittl mn"H pC'1 urut lUlin tt tllil t,ulm" nml,. flU' .I!."UU'

of t,lw P"tn~ [J froilt thl' ("'titre', If' (J 1"1 tl..- (I ul,..' lif til,· ·.ph1It.\ t\\tI ("IIUf'. wlt,h "'PUll Veil tWILl nngh'''I ~ .mel cf! l tlc/-, ,olU·1t lllwmg Ihl ,,'ft. \ lit ti ,,, .. I (11' 110M lt~ U'UH. WIll ('untum lUlt.w,·t'U ttWIIl un ,lUlUI1Uti on th. lolil'f'Wf II' thl' HlllWft' '('ill' l'Htl'n~ULl of' f,hlt~ mmuh,·, ILt J~ In

or

PI J ,. + tI, PI

[f ,. > it, p& .,~ ,. .. " 'I.nd p. .. Pa ... :.!u , Ahl(~ tilt' wholel rnAHt4 Hi tlw Hlwll 1-4 /.,

rf r- tI, /II " " 1\.11«1 "~ PI ~,.

.hr,I11", It"rn't; wh.,u /i lit Ii 1.,mt.

I' /1 OMl,.

or tIle' pntp.ut,ut.l,l,ml tll..- thl'('I'H .lflll\t·,l frmu it, flU' Ul.1IU4 nf till' ttlwll WI'tt' ('unt'l'nt,rnt" .. l I'" 1.1w ",·ut,N ..

1,h.· NUt'" 'V", If ,11* wbuh· ()u t,l,,· utlwr lu ... ,', wiwn

v

(UII"

17 "'rum thlll C.1NtWlltIL'Y prllpc"'ttlltU fuUuw UlUlttith'l.h'l) l ~II' flflilhul«" (1) A Hl.twr(· nr unatuI'1il d,·uNlty. or tttlf' C·lIlIIpl" •• d H' f·lIw·!·nlrw utrllt" Hi' umfhnn ch·lUut.y. UULY 1 ... tr,·uh·.l~ ,,"I fnr liM ltM ".·tum Itt nu .~lCh,lnl') 1"'111', it! (mwt'rrwcl, aM 4"ClUivul.'nt ttl n fuuJ.tI" PUft,wlt- nf "Iuul "IJUt"' 1'1.",,·,1 ,,' I'M C'f'uttr I',

(2) t'·nr n. pcnnt with", Mwh n, "l'h,·r.', ttHo ., .. t,,·r.' limy 1",,0 tin 1,1,>,1 mtn t.wo purt," hy t.ll(' ('HtW"ut rw "ph.-u· ll'n ... cihg thruuJ(h t I ... Ijl,auL 'I"h, Hllh'r Imrt. 1M inopc.mt.lv.t futtl III"Y ht· IJ,ClUlr.·d. '" h.lf' t h.· m..... tI .. ,> 1fl' r.·l,I., .. ·.·tI b,)· n l)'''lh(~l., uf «(pml flU'"" HltU.,t,.'cl ILt, theA ,·"'~tJr.·.

'I'} ... lWn.VN11y lKHh"H nr., tet, t tu' nu l pRrt nPlnmcatrl.".4v ·., ·fWfl) UI

Hhn.I)t., lIurt t.lwugh nut utuiuJ'1U lU ,1('1 1',)1 ttWlt' ('...,"', utn,· ·~t h' 'Uf m

J(f'lwm.l fmrly hmtlug"nHtntl 'I-I"IY lUll)" t h.-u,fu,,· ht' t ft1dt. ,I Ifl turf'.r "M4". UK rt'Klt,niM tJhrltr I\f't.wu Ill. uti .... hcwlu',.. '''''' "'UlI.h' 11.IutU'I~I".

"}w fUutwn Hf n. hmlv within .1 ~l,h. fc' .. ut.~ ttt sllu"ltr:t.tlf·«l h) thll .""" •• " (If Ia. Uw!.4·Ul' wltlun " "lth"r 1('ul "Yr.n fIJ, HI 41f I' ht,lr Ul Ii "pt"'flf'ul ,·h •• " t U

16-18]

Introductory Propositions

13

the swarm fills a sphere uniformly the mass operative at any point vanes as

t e cu e 0 t e istance om e cen re enee tee eotrve orce towa s the centre valles directly as the distance As another example It may be

proved that If the density of a globular cluster varies as (1 + '1.2) - I, r being the distance from the centre, each star moves under a central attraction varymg

as r (1 + rs) - i

18 An approximate expreSSIon can be found for the potential of a body

at the centre of gravity of the body and the aXIS of x be drawn through the POlDt P, the distance OP bemg r Let drn be an element of mass at the POlDt (x, '!I, $) The corresponding element of the potential at P IS

Gam Gdm

dV==

-i

Gam { p (1]) (P)S (1&)

== -r- 1 + ;; PI P + ;: r, p +

}

where PI' P2' are the functrons known as Legendre's polynomials

Hence If the expanSIOn IS not earned to terms beyond the second order,

V = G r am (1 + ~ + ate' - p2)

r r 2r2

But If A, B, 0 are the pnncipal moments of inertia at 0, and lIS the 2

A +B+ 0= f2pid'fJ1., I= J(ps-or)am and since 0 IS the centre of gravIty,

Hence

v = r + 2r' (.A + B + 0 - 31)

and we see that the potentnal of the body at P differs from the potential of a particle of equal total mass placed at the centre of gravity by a quantIty depending only on l/rB Except In a few cases this quantity IS negligible

14

I ntroductory Propoeuums

[oa II

In astronomical problems not only by reason of the great distances which separate the heavenly bodies In companson WIth their hnear dimensions, but because they possess m general a symmetry of form which makes ~ + B + 0 - 31 Itself a small quantity

19 We see then that In general a system of 11, bodies of finite dimen-

SlOns can be replaced by a system of 11, small partIcles of equal masses occupYIng the pOSItIOns of their centres of graVlty The total potential

energy of the system IS

where """ m, are two of the masses and r~ their distance apart For If we start WIth anyone of the partICles this sum, WhICh consists of in (11, -1)

terms, represents the potential energy of a second in the presence of the first. of a thrrd 1D the presence of these two, and so on The eqUd.tlons

of motion are an in number and, according to § 15, of the form

au au au

Hence

or

IfflOiY, - yl:T11I\ ast + bg I1nt,z, = i I~= ast + ba

where (x, y, z) IS the centre of gravity of the system Thus we have the SIX Integrals correspondmg to the fact that the centre of gravity moves WIth uniform velocity In a certain direction

AgaJ.n, we have

~ (y~ 0aU _ $, ~ U) = I~m\m, {Y\ Z, -JzJ _ $, y." -aY"}

~ s; vY", t 3 r,,, r. v

Hence

or

~m." (y,z, - z"y,) = 0 ~m" (y,z~ - z"y,,) = C1

and snmlarly

18-20J

Introductory Propositions

15

whence, on integration,

t ~m, (x,'l + '!Ie + Zt2) = h - U

where h IS constant ThIS IS the Integral of energy

There are then m all ten general mtegrals for the motion of d. system of particles movlDg under their mutual attractions and It IS known that no others exist under certean hmrtations of analytical form (Bruns and Poincare) They are In fac~ SImply those which apply m VIrtue of the absence of external forces actmg on the system

20 Let the centre of gravlty (x, y, s) of the system be now taken as the ongm of coordmates If (~'" '1'/" ~,) are the new coordinates of m,.,

and

~m", !m,~,-~m,~,=O The equatIons of motion become

where U IS the same as before, but rid IS now given by rtd' = (f, - Ej)1 + (~, - 'T/j)" + (~" - ~j)t For the Integrals of area we have

c" = ~nt{, ('}JiS, - z,y,)

= ~m, {(y + ~~) (z + ,,,) - (z + ~,) (1; + ~,)} - l:~ ('1'~ - ~,,,,,:) + (yz - zy) ~~

and slIXlllarly

l:'11li (~'~i - E"ei) :;::: C9 + (a.abl - a,.ba)/Im, - c: ~m, (~,'1Jt- 'I1t.Et) = Os + (~b9 - agb1)/'Itm, = c:

16

Introductory Proposusons

[OR II

The mtegral of energy becomes

h- U =i~~ {(il: + E,Y~+(y+ "1,)2 + (z+ t,)9}

= iIrtlt, (,,2 + "1,. + ~,2) + t (~S + asS + a,2)/~m"

or

where

21 An mterestmg equatIon Involvmg the mutual du~tcmces of the masses

can be deduced We have

2 S "","11.1 (E .. - fi)2 I.Im,11lJ (E,· + El 2E'~J)

',.1 , :J

= ~'Y!h~\. ~m, + Imt "iim,E/ 2~m,.ft !m,f.1

WIth simllar equatlODs for the other coordmates Hence ~m"mJr"l ~~ ~m,. (t2 + 1],2 + ~ .. II)

It follows that

=4(h' -U)+2U=4k' -2U

Slnce U IS a homogeneous functIOn of the cooldmates of degree 1 The

form of the result IS due to Ja.cobl Now U is eseentaally negative Hence

If 1£' be poSltlve the second denvcLtlve of !.m,m.11",' Will be always positive and the :first derivative will Increase mdefimtely WIth the time Thus the first

derrvatrve; even If nega.tIve Imtially, wlll become POSItIve after a certam tune and therefore ~m"l11..Jr,l wrll mcrease WIthout limit Th18 means that at least one of the distances WIll tend to become infinite We see therefore that a necessary (but not sufficient) condimon fOI the stabihty of the system IS that h' must be negatave

22 The angular momenta whose constant values are Cu C~, Cs are the projections on the coordinate planes of a single quantity They are therefore the components of a vector which represents the resultant angular momentum about the aXIS

(1)

~

For this aXIS, which IS fixed In space, the angular momentum 18 a mazrmum

The plane through the origin 0 which IS perpendicular to this axis and therefore fixed 18 called the ~nvarlable plane at 0 About any hne through 0 In this plane the angular momentum IS zero, and about any hne through 0

20-23J

Introductory Propositions

17

makmg an angle e with the invariable axis (1) the angular momentum IS

t; Cg Ca os

the pOSItIOn of the chosen ongm of reference

Here we have considered the angular momentum as anBlDg purely from the translataonal motions of the bodies treated as paaticles In reahty the total angular momentum of the system Includes also that part which anses

If the system consiets of unconnected, rigid, spherical bodies whose concentric

ayers are omogeneous n er s

point, as determmed by the translatronal motaons of the system alone, remains permanently fixed The condrtions hold very approxrmately in a planetary system But precessional movements and the effects of tidal fnction cause an mterchange between the rotational and translatronal parts f the an ular momentum WIthout disturbm the total amount, and to this

extent affect the pOSItIOn of the astronomical mvanable plane as defined

The centre of gravrty of the system may be taken instead of an orIgIn fixed 10 space The invariable plane IS then

(2)

and this 18 the mvanable lane of Laplace Its permanent fiXIty IS subject

to the qualifications Just mentioned

SImp e proposl IOn app ies 0 the planes through a fixed pOInt. 0 and contaming the tangents to the paths of the two bodies intersect the mvanable plane at 0 In one hne ThIS IS easily seen to be true For the :first plane passes through the OrIgm, the posinon of the first body (~, 'l}l, Zl) and the consecutive point on Its path

t dt z, + i dt Hence Its e usmon 1S

SImilarly the equation of the second plane IS

IJ1 (ysSs - '!I1$2) + '!I (SlIllJg - $slJ1s) + Z (lJ1a'YlI- a;~'!I'iA) = 0

The equamons of these planes togethet WIth that of the invariable plane may therefore be wntten

23 When we deal WIth the motacns m the solar system It 18 convement to refer them to the centre of the Sun as origin Let M be the mass of the Sun, m the mass of the planet specially considered and let there be 11 other

18

Introductory Propoeuume

[OH II

planets, of which the typical mass IS m" Then the total potential energy of the system IS

U (~ 'lnt,'mJ M~ 'In, 'C' 'In, 'TnH') G

=- ~--+ J..I-+m~-+~

r'J p" ~"r

where p", IS the distance of m, from the Sun, ~, the distance of 'In?, from 'In and r the distance of m from the Sun, so that

r,,l = (Xl. - (l,J)" + (y", - '!J,)D + (z, - z,)"

p,,2 (0:" X)2 + (y, - y)2 + (Z" Z)2

ax' ar: OZ

and of the planet cOIlSldered

au

au

au

mm-

ate '

mz oy ·

oz

If (E, 'I}, n are the relatIve eoordmates of the planet,

Hence, If (E" 'th, s,) are the eoordmates of m; relative to the SUD,

= {_ ('Tn + M) ~ _ I m, (E - E,,) _ s ~E,,} G

103 ~,,3 p,3

If then we put

R = () {I ~ - ~ ;:; (EE" + 'fJ7'J, + ss~)}

we have for the equations of relative motion

(3)

and SlIW.larly

E = - (m + M) G I + a~ r' 'OE

'IJ = - (m + M) G 1 + oR r '0'1]

t - - (m + M) G i + aR r a~

(4)

(5)

(6)

23, 24]

Introductory Propoeitume

19

The function R IS called the d~8turlnng functwn When, as m the solar system, the masses of the planets are small In companson with that of the central body, M, we see that the forces derived from this function are small In comparison With the attraction of M Indeed a first approximation to the motion of the planet considered, which may now be called the disturbed planet, IS obtained by puttmg R = 0

240 A double star, or system of two stars physlCally connected and at the same time Isolated from external mfluences, may be considered to present a

case of the problem of two bodIes In the solar system the disturbing effect of the other planets IS always operating Since, however thIS effect IS small

in comparlSon WIth the attractIOn of the Sun It IS useful to neglect R and to consider the orbit which a particular planet would have If at a given Instant the disturbmg forces were removed and the planet continued to move as part of the system formed by Itself and the Sun alone, its velocity In direction and amount at the given instant berng that which It actually possesses Such an orbit 18 called the osoulat~ng orb~t corresponding to the gIven Instant The actual orbit from the begmmng will depart more and more from the osculatmg orbit, but for a short interval of time the divergence between the two WIll be so small that an accurate ephemeris can be calculated from the elements of the osculatmg orbit The usefulness of the conception of the osculatmg orbit

goes much deeper than this, as WIll appear later

Now the equattons (4) to (6) show that ill the problem of two bodies, SInce

R = 0, the relative motion IS that whICh IS determined by an acceleratIOn (m + M) G/'f"J towards the body M which IS considered fixed But by § 11

(13) a law of thIS form leads to an elliptIC orblt With mean distance a and periodic time T, where

nT _ 27T, nllall_ ('Tn + J.lt) G

We can now introduce the usual system of astronormcal unite PrOVISion-

ally they are taken to be

Unit of tame one mean solar day

Umt of length the Earth's mean distance from the Sun Unit of mass the Sun's mass

Correspondmg to this system G IS replaced by the constant Jo2J so that

k = 2w/(1 + m)V T

which dIffers httle from the Earth's mean motion Here T IS the SidereaL year expressed In mean solar days and m IS the mass of the Earth expressed

as It fraction of that of the Sun The numerIcal va1ues adopted by Gauss

were

T _ 365 256 3835 m = 1/354710

20

Introductory Propositions

[OR n

which lead to

k= 0 01720209895, log Ie = 82355814414 -10 It may be useful to add that

1800 kj7r = 3548" 18'761, log (1800 kjn) = 85500065746

which differs httle from the Earth's dally mean monon expressed m seconds

The number lc 18 c.alled the GaUSSIan constant The numerical values of m and T on which It IS based are no longer conSIdered accurate Never-

tiheless It would cause great practioal Inconveruence to adjust the value of Ie to more modem values WhICh themselves could not be rega.rded as final

Hence It IS agreed to adopt the above value of Ie as a. definrte, arbItrary constant and to recognIze that the correspondIng umt of length IS only an

approXImatIOn to the Earth's mean mstallce from the Sun According to Newcomb the loganthm of true distance IS 0 000 000 01S

It IS also pOSSIble to put the constant Ie = 1 by adopting as the unit of time l/k- 5812244087 mean solar days

For breVIty we may often put

p. ~ /c2 (1 + m) = n2a3

1D the case of a. plaolletary orblti, and for a double star

where M, m are the masses of the two components when the mass of the Sun IS taken as umty

CHAPTER III

MOTION UNDER A CENTRAL ATTRACTION

25 If the attraction of the Sun alone IS considered, the relative motion

of anvother body of snherical shane IS conditioned bz.the __ centm.La.CCAl~l'\n

- - "

",rt., p, being a constant the value of which has been explamed The equations

nf Tnn+_,nn ,J Tn nl'\l..... ,,J'T1<l+.OQ a ... o

-r s:

r - r(J2 = - p./r2

rO+ 2rB =0

The latter equation g'lves imnredistely

".211 h

where h 18 the constant of areas Let v be the velocity In the orbit, P the

perpendicular from the orlgm on the tangent and '0/- the angle which the

tangent makes With the radius vector Then

r9 P

- = Sln "" = -

v r •

so that ~A

p~, J.

- " v

or the velocity IS ill versely proportional to P The result of elimmabmz 8

from the equations of motion IS

r = hair - p./rfl

whence

r" = 2p.Jr - halr2 + 0 (1)

and from these again

d2

';iia (r2) = 2 (rr + r2) = 2,."Jr + 20

The equation of energy IS

v2 = r~ + "tell = 2,.,,/r + c , " (2)

The geometncal mea nlng of the constant c has yet to be found. 22

Motton under a Gentral Attractton

[Oil III

26 From the second equation of motion

d d

dt =hu'J dfJ

where u = 1/r Hence the first equation of motion becomes

the Integral of which IS

~ ; II + e cos (fJ ry)}

(8)

where e 8J)d 'Y are the two constants of mtegninon But thIS IS the polar

equation of a conic section of which the eccentricity IS e and the focus IS at the ongm The semI latus rectum 1ft thiS connexIOll 19 more usually called

the parameteT and denoting It by P we have

Also

But by (1) and (3)

Hence

0= - p.(1- e"")/p

Thus If 2a IS the transverse aXIS of the orbit, 0 a parabola and c-+p/a for an hyperbola

p./a for an ellipse, e = 0 for l'he eguatIOn of energy (2)

becomes therefore

vJ = 2P./T - p./a, vJ= 2P./T,

tl~ = 2p./r + p./a,

(a-c 1)} (8= 1)

(8) 1)

(4)

Aga.m, "'" beIng the angle which the directaon of motion at (r, fJ) makes WIth the rachus vector (drawn towards the ongin),

p,e

'V cosy = -'1'= - Ii: SIn (fJ- "1)

'V sin t = r(J = 1~'U ='i {I + e cos (fJ - 'Y)}

are the components of the velocity along the radnis vector (Inwards) and perpenwcular to It The form of these espressiona IS to be noted For they endently represent (a) a constant velocity V = p./h = J/(p.!p) perpendICula.r to

Motion 'Under a Central Attraction

23

Since h = vr SIn y, the preceding equataons may be written p.esm(8- "I) =- 't)2rsmy cos".

g'lvmg e and ry when 't) and yare gIVen at (r, 8) Thus

JL' (~-1) = 't)2r ('I)2r - 2JL)sm2 t

27 In finding the relataons which subsiat between positions 10 an orbit and the time It IS necessary to consider separately the three kinds of CODIC sectnon The closed orbit, or ellipse, will be discussed first

hehon The angle I) - ry IS measured from this lme and IS called the t'1"U6

11,8 r ilw

= JL" J 0 (1 + e cos W)I

The corresponding result m terms of the eccentnc anomaly E has already

e n

the radius vector and the true and eccentrrc enomehes In the forms which are most frequently required We have

(C =r cosw= a (COB E -e)

'!J == r SIn w = a ,.)(1 - 62) Sin E

en.ce

(5)

rcos1iw=a(1-6)cos2iE rSUlB1W=a(l t-B)Slua iE

tan iw= J (~+:) taniE

(6)

'I'his last eq uataon ma.y be regarded 80S the standard form of the relation

I;)

ta.n tw = tan (450 + i¢» tan iE tan ,E = tan (4150 -l¢» tan tW

24

M{}fifll1 lIiU/t'r It ('fl/lirat .4.lIrtlrt"r~1I

I t'ff III

•

("Wi It: ,tI I f t"t.H h"

f'Ul'! A'

" f IIU", Ii' I t I' ,"", UI

1.' \ ( I f j t lUi f"

'tl rt ,f', ,,,

I • r ,'I Ii '1'

-/(1 tlwm 1

t .... t/ J.: "t'l~ N •

,II t'Il/if' J f , ,0'. ~I'

If nnw Wt1 tlrupluy (:;) :md (7) Wt' uhtlllll

t

I~' I tift·

IJ;' ,Il 1 t " III ,.'1

I

mean wlCwml.If '"Id mny 1H' ,h'lIf1h,d 11\' ;\f ubtrurwd KC'lllor' .. "'Iuntwn

At, n t! t, ) f,'

thf' ,..ngh~ M "u<l Ir1 hNUK .-'lCl'f'c;II4It.',1 au ~'n'·IIJ,., 1t1.',. IU, , If' s, .\1 U.k' " M~'

C~J: JrtlHHf'd In cln I' I ",..

H 'rlw ,loUlpltlfAl HulutUlu of' lh,· I"UM.,w lit ,U'Itf'" IW.t114tf ,J ( III f!i m .. I

rn thu (lflUo.tiuzlll I(lVN' "hU'tfl Nil 'htfirllltV III ttltW. rtf .1 ,"h''''''4 ,tt.'t"1

Cxoollt In tile! t',.,..\ (If K"I,IcIl'" t'C11IJ~lhl" wIlt II At "4 I ....... '''lIlt,1 tn. ~'1"', valutlH of tf and It, 'I'hcl g"rwl'nl IIIf·t,h'MI "PIJ1 ... 111.' •• ItI IUh"1 " f.'",,. f', '.tf 'It llluKtrat.,d hN"'. Uy ~mm nwmlH nn UJlfUH\ "' .. tot. ,~,.JUh"fI r • ~ t~ d,wl I.,., B. t fJ.N~ btl t}w llX~'t, KHlutum. nwl

M. N ... " jll!Hl A~.

Th(~f1

Itl .JI &/11 + (I "CI,m ,,: ••• J.1\.j • Uo

when A' - ~ HUt A' i .. ~'xIH"ldtl(l in n, Jluw"r Nf~rl'i" au ,l."I~q ' •• \1 1,I,t.) 'c.tr~~ ,lW14,' W Nogl(;otmg highnr POW'-I'H "t AA~ w.' bn.v,'

A1J" I (M ... M,.)/< 1· ... " f"'''' HI;}

and }wnro a. Hf'(x.nd rLJ.I,rHxmmtlUn All N~. 4/(11' J( tlull ylIdu_ It It,,' Huflicumtly JI,oC'umt .• • t,h(1 l)rorp"" tnu.y bfl r.·p.·,.",d tuttll n "".t.fat,wlttr\, II ~Hh tIt obtf\lIloo.

27, 28]

Motion under a Oentral Attraction

25

In order to obtain a good approximate solution at the outset a great vanety of methods have been devised These depend upon (a) the use of special tables, (b) an approximate formula or a senes, or (0) a graphIcal method Thus to the first order In e,

Eo=M +esmM

and to the second order In 6

tan Eo = sec 4> tan 2X

where

tan X - tan (450 + icp)tan {M the ~.renficatlOn of whICh may be left as a,D exercIse

Among graphical methods we can refer only to one, given by Newton (Pnn()'l,p",a, Book I, Prop XXXI) Consider a CIrcle of unit radrus and centre 0 rolling on a straIght hne OX Let E be the pomt of contact and .A the point on the CIrcumference InItIally COInCIdIng WIth 0 Let P be a point on the radius O.A such that OP = e and M and N the feet of the perpendiculars from P on OX and OE Then If E= L AOE=arc AE= OE,

OM= OE-ME= OE-PN=E- esmE

x

o

Flg 1

Hence If the CIrcle IS rolled (WIthout shppmg) along OX until the point P IS on the ordmate PM where OM = M, the pomt of contact gIves OE = E, which can therefore be read off when M IS grven, The locus of P IS evidently

a trochOId It may also be noted that the ordmate PM OE- ON=] -ecosE

which 18 the correspondIng value of ria or of dMjdE, and so grves the factor

required for the Improvement of an approximate value Eo For references to practIcal applIcatIons of the above principle Bee Monthly N ohces, R .A S,

LX.VII, P 67

26

, i 'it II f

29. In the CBfIO of' IHLrnhnllt' mULIOU

1t3] (/u;

t - tel - 1-''' II (1 t (OUH w)'

.,. J (~) '" ~ (1 t t.m tilt I d ",if I ~ II'} .. A ,,/( ~) (tlflll ~ III f ,\ t ImA i ill)

menca actor Wit t II urgnuwnt. /II All lU\I'I'" t .1.h· I{I\'HI,' It! "\att. 'f~,

argument M will be found in Hnu'if.,bmg.'r'H 'llut"''' (Nfl \. \, C tr 'I' '''d) l ...

deduced when t - to 18 81VUll thUH. 'I'hp p(tUnt,hln f H ~ III •• ,) I .. , r 1t'lIlH.",.f Ii tf fa the ldentlty

Hence

I '

A.)

Let

tim ,III .. A 1

A

aM 1,\1 A.l

X.., .. I,nn "'/, A. ) lau; I

j.M IID:~ vi (:'1) (i fu i "Ill .',1

(,all tI tUII', Then

and

tlllLU ~tu ,2 ("ut, :lty

By theso equatlOnM W run bu (·tLl(luh~tf'd .l1l1"·O,)' ~hnl , HI "~H ,TIa

30 Byperbohc tnotiotl n,lrml( t}w C'Urlr,Wt' hr ."",h uf ,'.,. '·IU",' 11,' r

attractlon to the fOOUH lllILy b4' tn"nttltl iu lUI ILfI.llngn .... wn\ h., IhJtf ••. , , hm

by uSlng hyparbohc functwlu. UlHt .. ,,,,tI uf clln·ul.~r hmJ'hHI1, nf t h .. fl4'4 h' ",. anomaly Thus we have

tI) - ,. 0(". to I t (f ("... ('t m1l /I')

11 - r sm w ... a. v(tA ~ t) IUtl'. I'

so tha.t

t --lil(" - t) _ II (l'f'ulIlh If' It + t ('()H W

29-31J

Motion under a Oentral Attracuon

27

r cos" tUl = a (e - 1) cosh" tF

~2 1"" ,. I ~ .J..!LLli't

11- - \ v ... I ..,............. 5!"'~

tan tw = J (: ~ D tanh kF

(II)

e- coshF

cos w = ---=--=-~

e cosh F_l1

JL.!)._ _'l '\ .1. m

hF e+ cosw cos = ::---:---

1 +ecosw

_L ~ nw

sinh F = v \~ .l} sn I+ecosw

SIn w = ~~ ... ) I:U.uu .J.'

ecoshF -1 '

J(e9 - 1) dF dw = e cosh F - 1 '

By employmg (10) and (12) we now obtain

er.: J(e2-1)dw 1 +ecosw

(12)

t t _;'3 r dw

- 0 - IJ-~ J 0 (1 + e C08 w),4

ecosh F-l

e~ -1

= \/(~) (e smh F-F)

(13)

which IS the analogue of Kepler's equation for this case

Analogy suggests the use of hyperbolic functions, but full and accurate tables of these functions are not always available Hence It IS convenient to

Introduce f, the Gudermanman funotron of F, where (Log denotang natural lo_g_arlthm)

or

•

sinh F = tant, cosh F = sec f, tanh t.H' = tan if

We may also put e = sec""". The pnncipal formulae (10), (11) and (13) then become

r -a (s seo f· n

(14'1

~ ,

tan !w = cot t+ tan if

r./(JJO;-S) (t - tu)= e tan! - Log tan (450 + !f) The last equation may also be wntten

J(IJ-a-a) ).. (t - to) = A.etanf-Iog tan (450 + if) where log denotes common logarithm and log A = 9 6371843

~ ..

(16)

vome1JS mOVIng m nyperoonc OrOl1iS are lew 10 number, and. in no case does the eccentncrty greatly exceed unity

31 There are certain astronomical problems which require the conaideration of repulsive forces according to the law p'r which are of the same form as gravrtational attractlOn but differ In sense The small particles which constatute a comet's tall are apparently subject to such forces and

I

28

Mot~on under a Central Attractlon

[OR III

finely divided meteoric matter 10 the solar system must move under tho pressure due to the Sun's radiation Hence we shall consider the effect (If replacing + f£. the acceleration at umt distance, by - f£' The drfierential equation of the orbit becomes

d'u p/

dfJ2 + u + h,J = 0

the Integral of which IS

,

u = h2 {e cos (8 - ry) - I}

P 1 (8 COSW I)

(17)

If we restrict w to such a rclJlge of valnee tha.t u (or r) IS POSItIve, bhis

equation gives only the branch of the hyperbola. convex to the centru of repulslOn at the focus, Just as under the same resmcnon the equatlOn (10)

gives only the branch concave to the centre of attraotaon As compl-.r<>d WIth § 26 the SignS of p and e, as well as of p.. have been changed HN1CO

the constant em the equation of energy becomes

G - 11" (1 &)/p +p//a

so that the equation of ellergy IS now

VII = p.'/a - 2p/jr (18)

Also. If '*' IS the angle which the drrection of motlOD at (r, 8) ma.kes WIth the

radius vector drawn towards the ongm,

du 'e

v sin ,y = rO = hu = ~ {e cos (8 - 'Y) - 1}

are the components of the velOCIty along the inward radrus vector and perpendIcular to It These are evidently equivalent to (a a constant

ve 0Cl Y - = -}£ = - J(p. p) perpendicular to the radius vector, the negative SIgn meamng that V'IS drawn m the sense opposite to that m which the radius vector IS rotating, aad (b) a constant velocity 8 V' In a direcnon makmg an angle !7r + e - 'Y WIth the radius vector, that IS, perpondicular to the transverse axis Thus at perihelion the velocity IS V' (8 -1) as compared WIth the veloctty V (6 + 1) at perihelion on the concave branch under an attractmg force

If the Circumstances of projection are given In the form of v and" n.t tho point (r, ()), we have

p/p = h2 = vir2 sm2 t

p,' e SIn «() - 'Y) = - V21 sin "" cos " p/ e COS «() - 'Y) = tIT sm"" + }£'

which determme p, e and 'Y in terms of given quantrnes In partrcular p/1 (8' - 1) = V'1 (v~ + 2p/) sm2 ""

31, 32J

Motion under a Oentral AUractton

29

32 Expressmg the coordinates In terms of hyperbohc functions we now have, Since the centre IS at (ce, 0).

$= rcosw= a(e + cosh F) y=rsmw = aV(e2-1) smhF

Hence

a(ell-1)

r a(ecoshF+l)

(19)

ecosw 1 r cos2iw= a (6 + 1) cosh2!F

1 sm2! w = a (e -1) smh! iF

(20)

6+coshF

cos W = e cosh E + 1

J(elI-1) sinh F sin W = -ecoshF +1 - ,

dw = v(elI- 1) dF

e coshF+ 1 '

hF e-cosw

cos =

ecosw -1

hF J(e2-I)smw

sin =

ecosw-1

dF = J(&-l) dw eeosw -1

• (21)

It then follows tha.t

= J(P3) J dF e cosh F + 1

,_,,' o.v(e'-l) e2-1

= J(a:) (e smhF+F)

(22)

which corresponds to Kepler's equation for tlns case

AB ill the case of an attractmg force we may now put tan!f=tanh!~ secf=coshF. tanf=smhF

and 6 = sec y Wlth these transformatrons the principal formulae of the solution become

r=a(6secf+1) (23)

tan!w = tan!+tan!f (24)

v(p. 'a 8) (t - to) e tan/+ Log tan (450 + If) (25)

or, as the last may be written,

.J(p.'a-S) A (t - to) = Xe tanf + log tan (450 + if)

m the notation previously explained

30

Motton under a Central Attracium

lOH In

83 The simple and important representation of the velocity in all cases as the resultant of two vectors both constant In magmtude, and one constant In direction also, may be rllustrated by consrdenng the hodograph of the motion ThIS curve IS clearly a CIrcle of radius V and centre at a distance e V from the ongm The four figures g'lVcn COl respond with the four disnnct types of motion, (a) elhptic, (b) parabolic, (0) hyperbolic, under attractIOn to the focus, and (d) hyperbolic, under lepulsIOn from the focus In all cases 0

IS the ongm, 0 the centre, and OP represents the velocity at penhehon If Q IS a.ny pomt on the hodograph, OQ represents the velOCIty In the orbit at.

one extremity of the focal chord which IS at nght angles to OQ The radius OF being V, 00 = e V and as the eccentncit mcreasea

ra IUS opposite to OP from the POSItIon 0 for a circular orbit to a point on the CIrcumference for a p.u-abohc orbrt As e increases beyond the value 1

p

~-I- ......

'1

\ / \ I \ /

o 0

(a) (b) FIg 2 (c) (d)

the POlDt 0 passes outside the CIrcle But the hodograph correspondmg to hyperbolic motion IS no longer a com lete cncle since the oss

o motion are muted by the asymptotes If OA, 0 B are the tangents from 0 to the circle the angles Oo.A, OOB are each equal to sm-1 e-1 and It IS easIly

seen that OA, OB ale parallel to the asymptotes of the orbit, that AOB IS equal to the exterior angle between the asymptotes, and that the arc APB constitutes the whole hodograph When the attraction ]8 changed to a repulsion and motion takes place along the convex Instead of the concave branch of the hyperbola, OP = V' (e-1), and the hodograph IS confined to that arc of the CIrcle which IS at all POInts convex to 0, whereas in case (0) It was everywhere concave to 0

34 From the POInt of view of practical calculation there are points connected WIth orbits nearly parabolic In form which require special attention Kepler's equanon for elliptic motion may be wntten

M= E-'ilnE+(I-e)smE

When 1 - e 18 small the accurate calculation of M depends on that of E - SIn E But If E IS small the latter expression 18 the difference of two nearly equal quantrties and cannot be calculated directly unless each IS

33, 34J

Motion under a Gentral Attraction

31

expressed by a drspropornonate number of srgmficant figures Hence the need for special tables (e g Bauschmger's Tateln, No XL) or an approximate formula Under the latter head may be mentioned the function

i E3 (cos ill E)14 4

which IS so close an approximation to E - em E over the range of E from 0° to 70° that the logarithms of the two expreseions never differ by more than

2 In the seventh place

It IS evident that In the parabola Itself E IS evanescent and generally In

the elhpse of great eccenrncity E IS small at all points near the attractmg focus The method gJven by Gauss In the Theorui Motus for the treatment

of Kepler's equation IS a particularly mstructrve example of the constructaon and use of special tables and as at the same trme It brmgs out clearly the relation to parabolic motion Its principle will be explained here

Kepler's equation may be Written in the form

M = (l-e)(aE + 13 sin E) + «(3+ ae)(E -SIn E) If a + f3 == 1, or

M= (I-e) 2AiB+(,8+a.e) .AIB

(26)

If

.A. = 3 (E - am E)/2 (rAE + f3 sin E)

:& = «(I.E + f3 Sin E)3/6 (E - sin E)

= (E8 - tJ8 E" )/(E8 nEB)

WhICh differs from umty by a quantaty of the fourth order only In E If f3 1/10, a - 9/10 With these values it 1S readily found that

B == 1 + TJfuE4 -

Hence log B IS a small quantity of the fourth order which IS tabulated wlth..A, Itself of the second order, as argument

We DOW put, 10 VIew of (26)~

.Ai = J(~ +~) tan iWl

so that

But

M 'V~ (t - to) - V rf (1 - eT (t - to)

whGl'e q IS the perihelion dlStance, 1n the present problem a. more convenient

- I(u) /e~) t

element than the mean distance a Hence

J(P- 1 + ge) t to

t 20 ,- = tan iW1 + i tan' twl

32

Mot~on under a Gentral Attractwn

[OH. rrr

the analogy of which with (9) of § 29 IS evident Here B IS unknown, but the supposition that B = 1 will lead to a good first approxnnation to tan ,WI and hence to A, and a nearer value for log B can. then be taken from the table ThIs In turn will Isad to a second appromnatron to tan iWl' and so on untal the correct value 18 reached Now let

'1' = tant iE = (1E + -rhEa )2 = iE'+ nE'

or

where 0 18 a function of the second order 1D .A, I e a. small quantity of tho

fourth order m E, which like log B can be tabulated with the argument A Hence

. r: 1(1 + B) 1(1 + e A )

tan tw = 'V 'T 'V 1 - e = V 1 _ e 1 - f.A + 0

Fmally, by § 27,

or

r cost lw = a (1- e) cost iE - qf(l + '1')

l-!A + a 2

r I +i:.tt + a qsec iw

•

so tha.t the problem of findmg w and r IS soh-ed by the ald of the tables

gI.vmg log B and a wrth the argument A without mtroduemg E explicitly Into the ca.]cnlatwn The method wIth very httle change IS adapted equa.lly

to hyperbolic orbits The tables WIll be found in the Theona Motus of Gauss, or lD an eql11valent form In Bausehmger's Ta/eln, Nos XVII and XVIII

II

I

I

OHAPTER IV

EXPANSIONS IN ELLIPTIO MOTION

35 The fundamental equataons of elliptic motion found In the last

M = E - e Sin E, e = sm cp

. (I)

(2)

r I-e!l

-= =1-e

3

a + ecosw

for many purposes It 18 necessary to express them as periodic functions in the form of senes Some of the more important forms of expansion will now be Investigated

But certain changes III these equations are sometimes useful Let

Then from (2)

fJ)- y-

-= ----

x+l I-f3 y+l

y-f3 a,+/3

0;= l"::fjy' Y = 1 + f3a;

.Also by (1)

log z = log y - 16 (y _ y'-l)

x+i3 [-(3 (aP-l)(l-,Bt)]

= 1 + ,sx e:x:p 1 +,& (I) + ,8)(1 + ,8x)

= x(1 + (30]1) (1 + /3(1)-1 exp [,8 cos q, {(f3 + X)-l - ({3 + x-l)-l}] (5)

34

Expans~on8 ~n l!Jll~ptic Mohon

[OH IV

The equation (3) grves

~ ~ 1 - 1 : #' ts + y-') = 1 : #' (1 - .811)(1 - ,By-') ! (6)

- _1 _ 1- ,8~ IX(l- fJ2) = (1- fP)2 (1 + ~X)-l (1 + (3X-1)-1{

- 1 +..& 1 + .Bx IX + f3 1 + ,811

It IS evident tha.t some expanslOns mIl be mQ.de more sImply In term.s of

~ than of e Hence It will be useful to have the development of any positive power of f3 In terms of e Now

or

,8 + ,8-1 = tan tel> + cot iel> = 2 cosec eI> :;: 26-1

,8 = 0 + ie (1 + ,/:J2) Hence by Lagrange's theorem

_ m I (ie)Sp±m [d2.P+m 1 (2.1> + m) arJl+m.-l]

p (2p + m) 1 drJ;2P+m-l P Q;-O

for the only terms which survive arise when q 2p + m Hence pm - m £ (ie)lp+m (2p + m 1) 1

p=o pl(p+m)1

I~ m. {I ~..d m m + 3...4 m em + 4) (m + 5) 66 +

\.)"8) + 4 1:'+ 42 21 c:;-+ 4* 31

}

(7)

and It IS readIly seen that thIS series IS absolutely convergent 36 SInce

It follows that

IX = ts - (3) (1 - ,8y)-1

log a: = log '!I + log (1 - (Jy-l) - log (1 - (3!J) = log Y + fJ (y - y-l) + t,82 (~_ y-9) +

Hence

w=.E+2 (.8smE+l,BSsln 2E +i,B'sm BE + ) (8)

But III and y can be Interchanged If the SIgn of ,8 IS changed at the same time Therefore

E=w- 2 (,8smw-t.&sm2w + kfPsm :Jw- ) It 18 also easy to express M In terms of w For, by (5),

log Z= log IX + log (1 + fJrr,-I) -log (1 + fJre) +,8 cos cp {(re + fJ)-l- (17)-1 + ,B)-I} = log 0; - ,8 (x - a;-I) + i,82 (x' - ,rll) - i/J3 (w _ !Ji8) +

+ f3 cos 4> {- (re - ail) + .8 (ret - a;-ll) - (32 (w _ ar) + = log a: - f3 (1 + cos~) (11) -re-1) + fJll(i + cos 4» (aP _ (D-ll)_

35-37J

Expansions ~n Ell'tptic Mot~on

35

and therefore

M = w- 2 {,8(1 + cos cp)SlD W - ~2(!+ cos cp) SIn 2w + (33(i+coscJ»sm3w- } By this expansion the equamon of the centre, w-M. IS expressed as a serres In terms of the true anomaly

37 We have now to consider the expansIons In terms of M, which are of

the greatest Importance because they are required In order to express the coordmates as perIodlC functaons of the time And first we take the case of

r-1 Now

dE a (1 _ e cos E) 1

'I' dM

'I'his IS an even periodic funcmon of E and consequently of M Hence

"Q, 1 JTT 2 rTT

- = - (1- e cos E)-l dM + ~-" cospM (1 - e cos E)-l cos pM dM

r 7r 0 7r 0

1 JTT 2 J""

= -- dE + - ~ cos pM cos (pE - pe sin E)dE

sr 0 7r 0

IX)

= 1 + 2:S Jp (pe) cos pM

P""l

(9)

were

lr

Jp (pe) = - cos (pE - pe sm E) dE

Jp (pe) IS called the Bessel's coeffi~ent of order p and argnment pe We shall

briefly study the properties of these coefficients so far as they are reqUIred for our immediate purpose

Let

+00

• F (t) = exp Hz (t - t-l)} = ~ ap"tP

-00

For t wnte exp (- tV) Then

+00

exp (-uvsm'o/)=:£ Q.pexp (- £pV)

-00

ThIS IS a Fourier expansion, showing that

ap == _!_J2tr exp £ (pt - f1) sin '0/) dy 27r 0

and combmmg the parts of the Integral which are due to t and 27J' - '0/ we

(10)

'U 1\

Thu« Llw (·ol'fluolt1Iltlol. IU tlu' I "Mil llUllit ,.'". ~I~ IH4Ij' l\ ,J~. II~ rI,l I' lIf

whreb WP IIn\I' tH~tucl, SU"

}t'(f) •• 1' j' 1"* 'It f ~., j

,. I" "

... (}, ,I );!o I ....

«

S8 'l'hl' t'tli·pt of' ('huUglUK thf' H'K"'" tlf r ,md, 1*1 tH t"llh Jt."llltulh ., .• ' Htm(w

Slmllatly ;'(t) I .. HrWl1ltllgtld If' ,I IN "'j1l1l'ltI111"'.1 fin' J1.'W'1

J/I (.r) t )".t t' (tt) II :h

Aga.m, th(, f,ttmlt ot ,htn'wntllltJt1,Lf /,'(t) \\11"1 '1 -Jl'" t til' ~I\' i .... (f I t ~)~,;/d I )/" '!P",,,,. P I

!..r f·'" I ( I) t .I", I ( I • p.,. tit

On tlw (libe·J llllUd,lt'wc Iftf.i'I'.'Ulllltf' "'(I) ""Ia J. I'" r til I ,\, "'"

,(t I 1):::.1" ( I H" l.l,. t I HI' or, tl(Illn.tmg t,hp ('odfiftU'lltbl ul'II',

4 :.1, I ftl) .I,IH" I '/J/ (II , 1ft!

rJ'h08(1 HlmpI(1 rpI'UrrNIt'(' f,tU'lUhuo ~huw thaf, ",,,It tW,' J.(UI h UIJU"H (It It. 'tr I I OC)Pfh(}WllLIoI of IUly ortltlJ\ urad ttt.·u d.,I\ r~"'J' I.m I" • 'IH~ ptH., 4' IH'~ ~t functWl1H of Mit! t ClCltli"INlt'1 ut' ..,.,. t~(. , ... rtlf.t.t H JlI,I. f,. .,' til 111\ "'., COUfflclon~ 1L1le! It.'I dt1rlVII,hvr,. 411(, .1, (of) luul t/. ( t~. Itt pM' II I*J n

J/I (11') .. , i [.f',. dtl" J 'In" I H

f f·',. ) (,.~ 2,'" (J1 ~ .. ~"lj d.d I

"" (.I) f., 1< p J ~.! II d." f (/~ , r •• 1" I' tI

... tl

I I '

.. ,. (.1 )

..

37-39J

Expans~on8 in Elliptic Motion

37

or

Jpll(X)+~Jp'(X)+ (l-:~Jp(X)=O ThIS shows that Jp(x) IS a particular solution of the equation

d2y 1 dy ( P~)

dafl+a;dx+ 1- X2 y=o

(16)

The general theory of Bessel's functions, defined c3.S solutions of this differentIal equatlOn, 1S not reqUlred for our purpose We need only the

solutiona of the first kind, with mtegral values of P, and the definition given above 1S suffiClent

89 The desired expansIOns In M can now be resumed We take sin mE which IS an odd function of E and M Therefore

smmE= ~ ~smpM('II' smmEsmpMdM

7r 0

= - ! ~ sm pM ('II' ! sin mE d tCI')8 (pE - pe sm E)}

7r op

=

2 I'II'm

- ~ sin pM - cos mE cos (pE - pe sm E) dE

'rr oP

(by mtegration by parts, the Integrated part vamshing at the hunts)

_ !.. ~smpMJII" Tn {cos (p - mE - pe sm E)

'T(' op

+ cos (p + mE -pe SIn E)} dE smpM

(17)

In particular, when m = 1, by (14)

2 smpM

smE=- I J;(pe)

e p

(18)

and therefore

E = M + 2 ~ !._ID_P_¥ r, (pe) p

(19)

Slmllarly, SInce cos rl'l,E 18 an even fUIlctlOn of E and 1l1,

2 t-

cos mE == ao + - ! cos pM) cos 1nE cos pit! dM

'IT' ()

= ao + -~ cospM J - cosmE d tsm(pE- pe Sin E)}

'T(' op

2 f'lf''1n

== ao + - ~ cos pM - sm mE Sin (pE - pe SIn E) dE

"IT' of

38

ExpanstO'nS ~n Elliptw M otum

[OB IV

(lntegratmg by parts as before)

1 (7rm-

=0..0+ -~cospM - {cos(p-rnE-pesmE)

7r oP

--

- cos (p+ mE-peslnE)} dE

oospM )

= 0..0 + m ~ - - - {Jp-m (pe) - Jp+m (pe)J

(20)

The constant term has not been determmed It 11:1

'IT' J 0

1 (1r

cos mE (1 e cos E) dE

and thus

- J {cos mE t8 cos em + 1) E i6 cos (m 1) oE} dE

'IT' 0

1 ror

ao- 11fm-O is Urn 1

- Olfm>!

The partICular case of m - 1 IS simphfled by (15), so that

cosE ie+2!cospMJp'(pe) (21)

p

40 From the last expansion It follows that

r 1 e cos E - 1 + te2 2s'I cos pM Jp' (ps) (22)

a p

:Axry posItIve power of 1 can be expanded by means of (20) For example ,.:

a'= (1- e cos E'1

= 1 + iet - 2e cos E + l~ cos 2E

-e cos pM COB pM

= 1 + tel + e' - 46.. P J,/ (ps) + e2 t p {JP-Il (ps) -JpH (pe)}

Now, by (14) and (15),

Jp-,(pe) - Jp-+2 (pe) = 2 (p - 1) Jp_1 (pe) _ 2 (p T_1) J. Cps)

pe pe 1'+1

::: ~ Jp' (pe) - p~ r; (pe)

Hence

(23)

39-41J

Expansions ~n Elliptic Motwn

39

The expansions of the rectangular coordinates can be wnt'ten down at once by means of (18) and (21) Thus, If x, y have this mearung and not as m § 35,

m = a cos E - ae

(24)

y = .j(1 - e2) a sin E

= 2a cot </> I sm 13M Jp (pe)

p

(25)

Other rmportant expansions can be denved from those already obtained by dlfferent1atlOn or mtegration For Instance, the equanons of motion give dlrectly

whence

::c 2

;:a = al "£pJp' (pe) cospM

(26)

~ = :9 cot q, I- pJp (ps) sm pM

(27)

41 The expansion of functions of the true anomaly in terms of the

mean anomaly 1S in general more d1fficult But sin wand cos ware readlly found For (§ 27)

sm w = .;:.....!...-~--=l-eeosE

4/(1 eI) SIll E

d dE

- cot cp dE(I-scos E) dM

= cot cp d~ (~)

= 2 cos cp I J,/ (pe) SID pM

(28)

by (22)

And

cosE-e cos w - -- -- - i-BOOSE'

e r

. .

(29)

by (9)

40

Expans~on8 in Elltpttc Motton

[011 IV

Hence a.lso for the equation of the centre,

BlD. (w - M) = 8 sin M - ! e '! I Jp (pe) ! S10 (p + 1) M - SIn (p - 1) M}

+ .v(I- (1):S J/ (pe) [sin (p + 1) M + sin (p - 1) M} = {e + 1- et J'I. (2e) + ..)(1- e2) Jg' (2e)} slnM + i apsln'PM (30)

e ~=2

where

l-e~ -- - \1

{Jp_..l (p 1 e) - JP+I (p + 1 en

e + ""(1 e2) {J'p..-l (p - 1 e) + J'P+l (p + 1 e»)

ThlS expeasion for the equation of the centre In terms of the mean anomaly IS lmporta.nt, although the coeffiCIents are rather compbcated

Hence. as far as if, am (w - M) e (2 lei) sin 1ft + 162 SID 2M + HeB sin 3M

w - M = e'(2 - ie2) Bin M + tel sm 2M + tiel sin ·~M

as can easily be vsnfied,

-42 For some purposes Laurent senes In the exponentl8.Is 11). '!}, z of

§ 35 are more convenient tha.n Founer senes in W, E, M Clearly

Let

fijI dID - " dw, y 1 dy " dE. Z-l de c dM

8 ao + ~ (ap coe pO + bp Sin pO)

Q(t. + l: {l (ap thp)"p + i (a." + tbJ» .,-P}

where log T = ,,8 By Founer's theorem

r2n "2"

Hence

where

2'71" Ap = J:'" S.,-1' ao

ThIS well-known form. mtermedrats between Founer's and Laurent's. IS general and Includes the case p = 0 It has been used alrea.dy In § 37

Formulae have been found which make It poasible to pass from any Fourier's expension III E to one In M The general result ma.y be expressed m a shghtly different way Far, SInce '1/ has the same period as e,

yP = I.Am.t""

* The readlng of §§ 42-46 can quite convenlelltly be deferred till after Chaptel XIII

41-43J

Expanswns 1n Elliptw Mot~on

41

where

27TAm= I:"" yPz-mdM = £m-1 f yPd(z-"'L) = [- £m-1ypz-m] - 'pm-l J yp-lz-m dy

r""

= p7/~-l exp {'pE - om (E - e sin E)} dE

(111 =t= 0) But when m = 0,

f2,," r2,,"

2'17" Ao = 0 yP dM = J 0 yP (1- e cos E) dE

= f:7I"(yp - ieyP+l - ieyP-l) dE

= 2'17" (p = 0) , - 7Te (p = ± 1) , 0 (pi > 1) Hence generally, for any function of Y,

:teo

S = l: Bpyp = ~ I B,Amz'rt& + ~BpAo

""CO

= Bo - ie (Bl + B_1) + ~ Spm-1BpJm-f) (me)z"&

m-=l 'P

43 There IS another form of calculation, due to Cauchy, In which Bessel's

coefficlents do not appear explicitly Let S be any perrodic function, such that

Here, by (4),

2 'iT Ap = to"" Sz-1J dM

= J:"" Sy-p exp [!pe(y - y-l)](l- e cos E) dE

= J~ Sy-tJ {1- !e(y + y-l)} exp [tpe (y - y-l)] dE

[211'

= Uy-pdE

... 0

where

(31)

the coeffiClent Ep of U expanded In powers of y±l beIng thus IdentIcal wIth the coefficient Ap of S expanded In powers of pl

**·1~ 1~'.rPtllllt"(lIIH • NlIiJI'lt· ,J!tif;,,,, : ( U 1\
**

III

AJ(mtl,

!, I ~~~lf ,I,ll I, d

II) I I 'Ii .Ir{. ,t 'II I I .. ,..; II U d If

.. '1r. "

r~ }' tiN I U • 1/8

.... p I • riM' j '/' I J 1111,tll

u II

}I I (" dN I '

11'1 I t I·.

It ' If

111 I tiS HIlI'( " /1 • q;( I:

'1 III I "'11

"I rI'l

C·

".11 JI I I t! R

Ii

wlwrtl

V I '/8 (\pfCI' 'I I II ' :~ .. ,

l'tl'

II till th.· ('O(ltlIc'lNtt, A" of N ~·~I .. uulc·,I1U Jl0'AP"l ot I" 'I'll' '"' til ,a~') hi I ~1JI1i It llluHUlY whl'" P ,I O.

l\ Hf"rU'H with HtIH.wl'H rtH'fttt'I"uh h".vlII~ fil., jLl~~UWftfif I" '1"1,,, II I'ml' '~I

th41 flu!thcwiH nh'llLcly t'lItlo;lti"rNj Uu, UWI! til r Inw" "'1 J. 1'1 ,.hlt, HIlt ',.,

Ativ.mt.n.gf'H ,f N , .. ut' Iojllttu.hll! foutl TIHh 11I1i~1',f'~ Iii ,f~ t t !l1JlHl"! fH , HI

p<tWHI'I nf.'l .'1 I l.d·t

,.__f, _H

.1 " I"

" ,

wlwrH J 1Ln.t <I nrct ant"'J(t'w (nut r"'W~t1\t) Th, UUhJI'fI.·.,1 ~ • ., ,~, •• " tit<, \ u' culltl!l OCIII4('''!lH "lImb,.r" nrul It lU • yut, ut t Ij,lt I~ kUHW!. I'w' HI '.4~ I,. ,\ ',1 i;f rtIClll1l'tt(i an th .. !, lIu~thud. By "OIHlltlf'lUJo( ",,,,throll tal ri .,f t" au ,t,. ,,1"111 1ft

<t+ t 1)/tl(t I t)w .... , au • t 1)111 t It I • Itt • I q1a I •• It ill (~Vl(<<'nt Umt

It\romo.duuhlu-t'lltry ULhl.·RlVlrag N II,Il." "tth flu IdKI .. "*,,.t/4/f, 'i ,1."" •• ", "unlln.l" ~~l)ltt" J(IVirlg N II. I, f' N 1"'. 'I'" '·"fa ht, rl'.utah' ,'Ht1>;t.llf ,,, rl tn •. , em~t of intt'mlmuwng t !Llnl e I Hhuwn t hnt

N 11.10'11 I ( I)' N I'd. '"

The UXIHPonMiun I" N~hor .IV~~IL ur "drt fUui tttl' IUKh-·.d ""Im ... 11'11 It. m.,1

j + q - p iM .. PC*ltlvl' WI.~tl lnt.t'Ut·r, A",I If /J ,I. q. N .".. 1.

43, 44J

.ExpansIons in Elliptic Motion

43

It IS now only necessary to consider the construction of the table for N -P, 0, q when 'P IS positive But this IS mdicatsd by

(t ~ t-I)2 = ~ N _p 0 q tP = ~ --q__ '- t'l' (- t;-l)q-'r

, , r' (q - r)'

whence p = 2r - q, and

N_p, 0,'1 = (- l)t(q-P) -::----:-:~~----:-:~

The tabulation of Cauchy's numbers, which are all posmve or negative

s = (~- 1 yn = c- e cos E)7n = (- 1e)m (y + y-l)m Then

U = {( - ie)m (y + y-l)m + (- le)m+I (y + y-l)m+l} exp ripe (y - y-l)]

= {(- 1e)m (y + y-l)1n + (-le)m+'1 (y + y-l)m+l} ~ (i pe)2 (y _ y-l)q/q I

q

= (- 16)711. (y + y-1)m ~ (kpe)!l (y _ y-l)'ljq I

q

+ (-le)m+l (y + y-l)m+l ~ (tpe)q-l (y - y-l)'l-l/(q -1) I

and

IS the coefficient of P In U and therefore of e» 1 S

When p = 0 the exponential function disappears and the constant term IS

u = (- 1e)"" (y + y-I r + (- 1e)m+l (y + y-l)m+l and IS therefore the first 01 the second of the forms

(te)"" m.! [(im) 1]-2, (~e)m+l (m + 1) I lei (m + 1)] !}-S according as m IS even or odd

On the other hand,

an th erefore

Hence

44

Expanstons in Ell'q)t't() M ot~on

COB IV

IS the coefhcient of yP-l In V and therefore also the coefficient of zP in S Companson with the preVIous result shows that

mN _p, 1111-1, 1]+1 = pN -:p, tn, q - qN -p, m+l, q-l IS an Identity From this the recurrence formula

(m - p + q + 2) N -1'+2, m, g- 2 (m - q) N -:p, m, q + (m + 'P + q + 2) N -p-9, m, IJ can be easIly deduced

45 The development In terms of M 01 Z of the functions

IS of special importance Here n IS any POSItive or negative mteger, and 1£ rn IS al~o a pOSItIve or nega.tlve mteger It IS only necessary to consIder the

second form This mvolves Htmeen:« coeffic~ents X: fn., whe-re

Now

r (j2 1 ± ~s rr

1 Sdw = hdt = n-1h,dM = ab dM = a2 GOS cp dM

o

and therefore X,' m can be expressed by a definite mtegral mvolvmg y and E, or by one Involvmg /lJ and w, by means of (4), (5). (6), thus

and 2'11" x:,m = I ~ (1 - ,&~(1 + {a~)-n-l xtn-\ (1 + {aa;)-n-t+\ (1 + {l/lJ-l)-n-t-, exp [~t9 cos cp {({a + a;-l)-l - ($ + a;)-l}] dw

The first of these forms shows that (1 + {ai)n+l X:' m 18 the coefficient of '!/,-m. in the expanded product Y1YS, where

Y1 = (1 - {ay)'~+1-m exp (i~ey)

Yg = (1 - {ay-l )n+l+m exp (- iterl)

Similarly the second form shows that (1 + fP)n+l (1 - {l2)--sn-S Xn. '" 18 the

coefficient of a;\-m. in the expanded product X1X21 where "

Xl = (1 + {a.:c)-n-i+'l exp [s cos cp {a/lJ (1 + {aa;)-lJ

Xli = (1 + {ar.rJ)-rl-2-f. exp [- ~ cos cp {a/lJ-l (1 + {lx-I)-I]

•

given by Tisserand (Yeo Gel, I, ch xv) An obvious method consists in expanding the exponential function oocumng In the first of the two Integral forms m a aeries WIth Bessel's coefficIents Thus

= 2'71" (1 + f'B)_-1 :t Jp (se) X~';"

p

yn (~) = (1 _ fjy)n+l-'m (1 _ f3 -1 n+1+711

and therefore equally the coefficient of r+p+m In the expansion of

Now

(~-h+ 1) 1 hI

(J-k+l) leI

('I,-p-k+l) J (p+k)1

(J -k+ 1) Tel

where h = p + k, and If J IS POSItIve the coefficient of yP IS

( _ fj)P ~ (t - p + 1) I (~ - p) (~ - p - k + 1) J

pI k (p + 1) t» + k)

= (- fj)P (~) F( p -~, - J, P + 1, (ai) In the ordinary notation for a hypergeometric series Hence there are two possrble forms for X::'pm

- 'm - n - 1, '/, - P - HI" + 1, fJ2)

of which the first lS available If 'b - P - m > 0 and the second If ~ - P - 'YfL < 0,

oient has a meaning If t - p = 'lit both forms become

When n 18 assumed to be POSItIve, at least one of the first two arguments of the senee IS always negative, and therefore the serrea 18 a polynomial m ~i For In the £rst form WIth ~ - P - If)/, > O. the second argument IS eertamly

46

Expans't0n8 in Elliptw M otum

[OB IV

negative If 111. IS POSItIve, If m IS negative, n + 1 - m > 0 and the binomial ooefficient shows that t - p - m < n + 1 - m, so that the first argument IS negative Similarly when the second form IS valid It also I~ d. terminating series When n IS negative one of the known tranaformations of the hypergeometnc senee may be necessary to gIve a finite form Hence Hs·nsen's coe6iClents axe reduced to the form

where X:,;" represents, WIth a sImple factor. a hypergeometnc polynomIal

ill fP ThlS form was first given by HIll

46 The periodic series In M found above are evidently legmmate Founer expansIons, satl8fymg the necessary eondlnons WIth e < 1. and as

such are convergent The Bessel's coefficients are gIven In explicit form by the Benes (11) whroh also ]S at once seen Lo be absoluLely convergenL for

all values of e But In practical apphcanons the expansions are generally ordered not as Founer senes In M but as power serIes In e Under these

circumstances the question of convergence IS altered and needs a special

InvestIgation Now

E=M+esmE

considered as an equation ill E has one root in the intenor of a gtven contour. and a~y regular functaon of tins root can be expanded by Lagrange's theorem

as a power senes in e, provided that

lesmEI<IE-MI

at all paws of the grven contour- We have then to find a contour WIth the

required property, and to examine Its limits

Weare to regard e and M as gIven real constants The equation E = M + p cos X + trp sin X

where p IS constant, defines cl. circular contour At any point on It

SIn E = sin (M + P cos X) cosh (p sin X) + £ cos (M + p cos X) SInh (p sin X) so that

I am E III = smJ (M + P cos X) coshl (p sm X) + cos" (M + p cos X) smh! (p sin X) = cosh! (p sin X) - COSB (M + p cos X)

whlle

IE-MI=p

• Of Wluttaker's Modem AnaZY8U1, p 106, Whlttaker and Watson, p 188

,

of any regular function of E In powers of e 18 valid for all val uee of M The

The most unfavourable oint on the contour for the re uired condrtion IS

that at which I smE lis greatest And our senes IS to be valid for all real values of M Hence the condition IS always fulfilled If It IS fulfilled when

sin X = ± 1, cos (M + p cos X) = °

or

In which case

X=±,7T, M=±!'1l" I sInE 1= coshp Thus the required condrtion becomes

e < p/cosh p

The greatest value of e IS therefore hmited by the maximum value of cosh which IS ven b

cosh P = p sinh p

greatest when p IS about 120 and that Its value IS then about i With

Let tan ex be the greatest possible value of s, so that

It easily follows that

exp p = cot ta, coth p = sec a whence, by the equation g'lVmg p,

cos a Log cot! a = 1

or, usmg common loganthms and takmg logarithms once more,

• log cos a + log log cot!C% + 036221569 = °

In this form It IS easily venfied that

a = 33° 32' :3" 0, tan a = 0 662 7434.

below this hmit, with the exception of some, but not all, of the periodic

astronomical eXpd.llSlOnS,

48

Ewpan8~ons l,n Ell1pt~c Motwn

..

[OR IV

I t may be convement for reference to gIve the following table

)

)

2 4& ( 4et 4e'

- J,(4e)=- 1--+--

e 3 _ 5 15

)

~ J, (5 ) = 6256' (1 _ 256" 625e4_

6 II - e - 384 ~ ~ + IJ44!

)

2 T (6) 81ell(1 9e2 81e' )

e ., 8 e - 40 7 + 112 -

2J: (26) - 6 (1

26' e' et

3 + 8 90 +

)

15e2 189B'

16 + 640

)

2J,' (5 ) = 62564 (1 _ 85es + 8756' _

II e 384 24 448

)

..

2J.' (6 ) = 81e8 (1 _12~ 1'356' _

" 6 40 7 + 112

)

These can easily be earned further If necessary, but they are often enough for practical purposes

Bessel's eoeffieients occur naturally m several physieal problems discussed by Euler and D Bemoulh from 1732 onwards In 1771 Lagrange * gave the expressron of the eccentric anomaly m terms of the mean anomaly, the result (19) above, and found the erpansions of the coeffleients as power senes, thus c:tD.tlClpatmg Bessel's work (1824) of more than half a. century later

* Oeutn"u, m, p 180 TIns reference. winch leems to have been overlooked, 18 due to Prof Wluttaker

RELATIONS BETWEEN TWO OR MORE POSITIONS IN AN ORBIT .AND THE TIME

C osen 0 sa IS yany ve con 1 Ions I IS evident tha.t when the focus is g'lven, and two pointe on the curve, an mfinrte

num er 0 or I S WI pass roug em e or It ecomes determinate when the length of the transverse aXIS IS given, though In general the solutnon IS not unique For let the pomts be PI' Fg and the focal distances 1'ur9 In the first place we take an elhpmo orbit wrth major aXIS 2a The second focus hes on the circle wrth centre P1 and radius 2a - r], It also hes on the c ole with radius P

pomts provided (0 being the length of the chord P1P9)

2a - 1'1 + 2a - r~ > c

or

...

(1)

If this inequality be saasfied two orbits fulhl the given eonditions , If not, no such orbit esists We notice that the two mnersectaons he on opposite

SIde of the chord, In the other on opposite sides In other words, 10 one

I, W 1 e In the other orbit It does not Only when 4a = r1 + 'l"1I + o the two CIrcles mentioned touch one another In a single point on P1P~ and the two orbits coincide In tlns case th.e chord. passes through the second focus

When the orbit IS the concave branch of an hyperbola. the second focus

1 WIth centre Ps and ramus r2 + 2a These CIrcles always Intersect In two

r1 + 2a, + r, + 2a > o

always There are therefore always two hyperbolas which satISfy tb.e condinons The second fOCI he on opposrte sides of the chord and hence In the one case the cb.ord intersects the axis between the two foci and the difference

50

Relat~ons between two or mot e P08'tt~on8

[oa V

between the true anomalies at the pointe Pl' Pg IS less than 1HO'>, while In the other case the chord mtersecte the a.XIS beyond the attrctctmg fOLUS and the difference between the anomalies IS greater than 1800

Under do repulsive force varymg Inversely as the square of tho distance tho convex branch ofan hyperbola can be described The position of tho second focus IS agam given by the mtersectaon of two circles, the onc with centre PI

and radIUs '11 - 2a and the othel wIth centre P'J and radius 1'2 - 2a, Theso CIrcles mtersect 10 two points provided

or

4a < Tl + r a - o (2)

There are then two hyperbolas and ill the one case the chord mtoraecta tho

aXIS at a pomt between the two foci while In the other It CUtb the axul at a pOInt beyond the second focns

It IS easy to see similarly that It IS always possible to draw four hyp~rbolas such that one branch passes through P1 while the other branch pasijQS

through Ps These have no mteiest from the kinematical pomt of VlOW smce It 1S ImpoSSIble for a partIcle to pass from one branch to the other

The case of parabolic solutions, two of which always exist, can be inferred

from the foregomg by the principle of continuity But It 11:3 othorWIso cl~ar that the directrix touches the circles with centres PlJ Pjland radii '/'11 t!j 'I'hosc

circles, which intersect in the focus, have two real common tclngents either of which may be the directnx The corresponding axes are the perpendiculars

from the focus to these tangents In the case of the nearer tangent It if:! evident that the part of the axis beyond the focus mtersocta tho chord PJPJ

and the difference of the auomahes IS greater than 1800 III tho case of tho Opposlte tangent, on the other hand, It IS the pcl.rt of the axis towards the

directnx which cuts the chord and the difference of the anomalies 18 loss than umo

These Simple geometncal considerations 'Show that, when the transverac ans 18 given, two points on an orbit may be joined In general by four elliptic arcs (of two ellipses), by two concave hyperbolic arcs, by two convex hyperbolic arcs, and 10 particular by two parabolic arcs 'I'his conclusion IS qualmod by the conditione (1) and (2) which of course cannot be satlsfied simultaneously All these different cases must present themselves when we sock the tune occupied in passmg from one gtven POlOt to another, as we shall at once see

49 Let Eh E'J be the eccentric anomalies Cl.t two pomts P1, PJ on an ellipse, and let

Then

2G=E2+ E1, 2g=E2- El

1"1 = a (1- e cos El)' rg= a (1- 6 cos E'J)

48-50J

'lin an Orbit and the Time

51

and

= 2a(1- e cos G COS g) Again, e being the chord PIP",

OS = a" (cos ED - COS ES~ + a" (1 - &) (sin Ell - sin ES"

Hence If we put

then

and

& = 4all sm2 9 (1 - COs2 h) c = 2asmgsmn

or

If further we now put

or

we have

e - 8 = E" - E1, cos t (e + 0) = e cos t (E" + E1) '1'"1 + '1'"2 + c = 2a {I - cos (h + g)} = 4a sinlliE

'1'"1 + '1"" - 0 = 2a {I - cos (h - g)} = 4a sm2 ~8 ..

(3) (4) (5)

the time t of descnbing the arc PlP" IS gIven by nt = Eg - El - e (sm Es - sin EI)

= e - 0 - 2 SIn i ( E - 8) cos i (e + 0)

6

where e and 0 are gIven by (4) and (5) In terms of '1"1 +'1'"21 c and a, and tlns

18

50 It IS evident that (4) and (5) do not gIve E and 8 without ambiguity, and tlns point must be examined. We suppose always that Es - E, < 360Q, 1 e that the arc described IS less than a smgle circuit of the orbit , and we assume. that the eceentnc anomal IS reckoned from the pencentre in the

direction of motion N ow It IS consistent WIth (3) to take t (e + 0) between bet een the same limits Hence E lies

between 0 and '1f' and to lies between - t7T and + t?7" But the equation of the chord p\P2 referred to the centre of the ellipse shows that It cuts the axis of u; In the point

(JJ = a cos t (Es - E1)/cos i (ED + E1), '!J = 0

52

Relatwns between two or more Posit~ons

[OR V

so that, If Q IS this point, A the pencentre and FIF9 the foot,

FIQ flJ - ae cos 1 (Ii: - 0) - COB t (e + 8) 8In IE sm 18

AQ = fJ) - a = cos g (E2- E1) - cos t (Ea+ E1) =- sin !El sm !E~

F':J.Q x+ae cost(e-~)+cOSt(e+o) cos gEcosiB

AQ = flJ - a = cos i (Ea - &1) - COS t (E2 + AI) = sin tEl SIn kE2

ow sin le and cos to are always positive We may also take El less than 2'7T and SIn !EI osrave then 81

t e arc includes or does not Include the peneentre In the first equation the left-hand side IS negatIve when the chord mterset.'bs the aXIS betweere

the pencentre and the hrst (attrd.ctmg) focus, In the second when the mtereecnon falls between the pencentl'€ and the second foeus Otherwlse

both members are posibrve Hence we see that SIn!O IS posltlve If (1) the arc eontams the pencent,re and the chord Intersects FlA., or (2) the arc does

not contain the pericentre and the chord does not intersect F1.4, and that cos e IS OSI er-

sects FgA, or (4) the arc does not contain the pencentre and the chord does not mtersect E,..d. In other 'Words, 19m is IS pOSitIve when the segment

formed by the arc and the chord does not contain the first focus, and cos-le

Let e1 and 81 be the smallest

The other possible values are 2'7l' - el and - 81 If we put

there axe four cases to be dlBtlngu18hed, namely

(a)

(d)

when the segment contama neither focus,

(b)

when the segment conta.ms the attractmg, but not the other focus

(0)

t = 2'7T!n - t:s - ~

when the segment contains the second, but not the attractmg focus,

when the segment contaans both fOCI It IS easy to see from § 48 that when the extreme pomts of the arc alone are gtven these four cases are always presented by the geometncal condrnons and can only be distinguiahed by further knowledge of the circumstances Usually It IS known that the arc IS

Impa.ra.tlvely short and hence that the solution (a) 1S the rIght one

50-62]

in an Orbit and the Time

53

51

The correspondmg theorem for parabohc motion IS easily deduced as

(4) and (5) become

ae2 = rl + '1"9 + 0, a& = '1"1 + r~ - 0

At the same time, If we replace 11. by pi/a,., (~) becomes ,.,.~t = iat (ell - sa)

= i ('1"1 + '1"2 + c)t + ! ('1"1 + rz - o)t

As this applies to the motion of a. comet, and the mass of a comet may be eonsidered neghgible, we may therefore wnte

t

which IS the required equation It was first found by Euler As re ards

the ambiguous sign, the second focus IS at an infinite distance and does not come Into consideration But 8 IS negative or posmve accordm as the

\egment formed by the arc descnbed and the chord contains or does not oontam the focus of the arabola Hence the lower (+) SI IS to be used

when the angle described by the radius vector exceeds 180°, and the upper

- SI IS to be used when this an Ie IS less than 180°, as It almost

always IS in actual problems

52 The solumon of ('1) as an equation In C IS facilitated by a transformation due to Encke We put

c = ('1"1 + '1'2) sin ry, 0 < ry < 90°

and

'I'hen (7) becomes

3"1 = (1 + Sin ry)t 1= (1 - SIn ry)~

= (cos try + am f'Y)' + (cos f'Y - SIn fry)8 Flrst we take the u er SI n, 10 w hich case

(8)

8'1] = 6 sin 1"'1 coaa try + 2 SlUSh'

If we put

e

then

Q

and

Hence

where

c = ('1"1 + '1",,) "1P

P = SIU "'1/"7 = 3 sID f(8) ";(cos J<8»/SlD (8)

(10) (11)

54 Relat'lOns between two or more Poeuums

Smce p. and 'TJ are both functecns of 6, p. can be tabulated with the no] When such a table IS available (cf Ba.uBchmger's Tajeln, No X-"'-l known, C IS immediately given by (10)

In the second place we take the lower SIgn In (8), so that 3"1 = 2 coss try + 6 sm~ i'Y cos try

6 cos trY 4 coss I'Y

If now we put

then

and

3'TJ 2 V2 sin (8), 90° < e < 1350 sm ry 2 .v2 am t ® v( cos i6)

as before Hence (J 0) and (11) apply equally to thIS case, WIth the

that a as given by (12) IS an angle ill the second quadrant IDste first Except for thIS the SolutIon IS formally the same in both

different tables would be necessary The case of angular monora 180°, how9"ter, seldom demands consIderatIOn 10 practIce

to the focus we have (§ 30)

63 For motlOn along the concave branch of an hyperbola under

r1 = a (e cosh El - 1), r~ = a (e cosh E ... -1) and we may suppose E ... > El Hence

where

= 2a {cosh i(e - 0) COshi(e + 0) - I}

e- 8 = Eg - Ell COshi(e+ 8)=ecosh i (E~ + E1) AgalD, the chord c 18 given by

c2 = a'" (cosh E'J, - cosh E1'f + a ... (fP - 1) (sinh E2 - smh El)2 = 4a'" smh! i (E'J. -. E1) sinh" t (E. + E1)

+ 4a2 (ell-I) smh2 i (Ea - 81) coshs i (Ea -I = 4a!J sInh1i (e- 0) {-I + eoshs i (e + 8)}

or

c = 2a sinh t (e - 8) sinh t (e + 8) Hence

r1+r'J,+c=2a(coshe -1)=4asmhllie 1"1 +rl-c= 2a(cosh 0 -1) = 4a smh"'io

52-54

~n an

nt = 6 sinh E2 - H2 - (13 sinh El - E1)

= 213 8mh! (E2 - E1) cosh l (Es + E1) - (E2 - E1) = 2 sinh t (e - 0) cosh -f (e + 0) - (e - 0)

=slOhe-smh8-(e-o) (16)

where e and S are glven by (14) and (15) This IS the form which Lambert's theorem takes III this case

We may take t (e + 8) as defined by (13) POSItIve, and t (e - 8) IS positrve SInce E,j > E, Hence E IS positive N ow the equation of the chord referred

o e cen re 0 e yper

f.C = - a cosh t (E2 - E1)/cosh 1- (Es + E1), Y == 0

or, (- (£6, 0) bemg the attractmg focus withm this branch,

(17)

The left-hand side IS negative or POSitive accordmg as the mtersectton falls

eyon sinh is IS positive when the angular motion about the focus IS less than 180°,

and negatrve when It exceeds 180° Thus the Sign of 8 IS determmed. If

we put

then

smh j e == + ~, smhto = ± m2

or

exp i e = + 1n" + It/'fI"-12+ 1, exp to == ± 'fI'~ + "fms2 + 1 smh e = 2'Jnt ",Im12 + 1, SInh 8 == ± 2ms It/~2 + 1 Hence (16) can be wntten (Log denoting natural logarithm) 'v'm2+1

where the upper or the lower Sign IS to be taken according as t e angu

.. actin focus IS less or eater than 180°

54 The corresponding theorem for motion- along the convex branch of

an hyperbola under a repulsive force In this case 32)

Hence

rl + r2 == 2a {cosh t (e + 8) cosh! (e - S) + I}

56

Relattons between two or more PoS'tttoruJ

[OM V

where

e - 0 = Es - Ell cosh! (e + 8) = e cosh t(E'J.+E1) and as in § 5J

(1 fS)

We have therefore r1 + r!l + c = 2a (cosh e + 1) = 4a cosh' ! e

r1 + r'J. - c == 2a (cosh 0 + 1) = 4a coaha ! 8 Then by ~ 32 (22), If p.' _ 'lb!la3,

e = 2a sinh! ( e - 0) sinh t (e + 0)

(19) (20)

- 2e sinh ! (EJ - E1) cosh! (Es + E1) + Ee El 2 sinh i (e - 0) cosb t (e + 0) + e 0

= sinh e - sinh 8 + e - 8 (21)

where E and 8 are given by (19) and (20) This IS analogous to the other

forms of Lambert's equanon

Puttmg as before

we have of neceesity

cosh i e - + mh cosh t 8 + ~

but there IS agam an ambigurty in the values of e and 8 Now we may take

Eg>E1 and 'lCe S) positrve I and we mcty define l(e+o) as the posrtive value which satlS:6.es (18) Hence e IS ositive and ex

equation 1 ) now corresponds

!C a,e 2" smh ~ e smh V 8 Jeosh i (Ea + E1)

showing that 8 IS positive If the chord mtersects the aXIS at a point on tho

SIde of the focus towards the centre It must be noticed that thiS focus 19, as before, the focus within the branch and not the centre of force Hence

exp t8 > or < 1 according as the angular motion about this focus < or > lHOu It follows that

exp (i-e) = + fn.t + Vm1'J. -1, exp (,8) = + ms ± Vms".:-y

and hence tha.t

nt = 2~ Vm1'ia - 1 =1= 2m.". V'l'nt8 - 1

+ 2 Log (fnt + V fnt~ -1) =+= 2 Log (ms + V riI.a!l- - 1)

where Log denotes naturallogauthm and the upper or the lower sign 18 to be taken according as the motion about the Internal focus (not the centre of force) 18 less or greater than 1800

In all cases, whether the motion IS along a parabola or elther branch of an. hyperbola, when two focal drsteaces are given In posinon and nothmg

54, 55]

in an o-i« and the Time

67

more IS known about the circumstances, the discussion of § 48 shows that the embrgurties in the expresslOns for the time of describing the arc correspond to the dtstmct solutions of the geometncal problem Hence they cannot be decided without further information In practice, however, It rarely happens that the angular motion about a focus exceeds 1800 and this limitation, by which the upper SIgn can be taken, will be generally understood

55 A quantity of great importance In the determination of orbits IS the ratIo, denoted by '9, of the sector to the trlangle The case of ellIptic motlOD.

IS taken first Smee » = kJCLb. where h IS the constant of areas, twice the area of the sector IS, by (6),

ht = ab {e - 0 - (SIn e- sin ~)}

But If (lXI' YI), (Will Y'J) are the axtremitaea of the arc, twice the area. of the tnangle IS

2~ = (WIY2 - (J;",'!/l)

= ob Ism Eg(cos EI- e) - smE1(cos Es - e)}

= ab {SIn (E", - Et) - 26 COS i (E2 + E1) Sin t (E2 - E1)} = ab [sin (e-~) - (sm e= sm~)}

by (3) Hence

e-B -(sm e - sm 8)

y = sm (eo - 8) (sm E sm ~)

(22)

ThIS expressIon contains OJ nnphcitly and this quantIty IS to be ehmmated

Let 21 be the angle between 7"1 and r~ and let g, k have the meanmg asslgned to them 1 n § 49 Then

l6al SlOg iE sm2i8= (r1 + rt+ 0) (rl + 7"1-0)

= (11 + rlj)J - r19 - '1'82 + 2r1'1'"g cos 2/

= 4ryrg cosBf

whence

Also by (4) and (5)

2a (cos 9 - cos h) = 2 cos! v;:;:;; TI + 7",1 = 2a (SlD~ 1e + sm2 t~) _ 2a{l 0089 cos h)

and therefore

'"I + Tt 2 cos/cos g VTI7", = 2a smllg

Agam/ by (22),

nt

'!J sin 2g 2 sin 9 cos It ant

nM

I l It \

II,

HUWt' tI'ltl~ It C h. Itll utt ... 1 It maf

, h 'lUI, (cj

'1

a (~!I "111 :) 'II ,

~111 /' :l, II I , .. I,'

(~ .""If \ ',I

·111 "

•

In tlHl llutnt,lUll (It Un,mlll W~I \\uh'

1'1 I i •

I ~! , 'II I \' I. t'

and tlwu (~:I) lturt (2·') hi'('lIIm'

!l J III I, ( I 4 !Im' t :11

f·f • ~

Tht, vu.ltw tlt.v 1M llu lw fUII ... 1 fly ", .. h IUl{ ,hi. film fit •• p •.• 1 ... ,. m V ,."" 'I , h" l4()lutlUtl lWlng l~lrfi.r .. wd hy .. tmu· 1I"~t,h"d ut .'Pln 1I\ItU,~t lI.tl

) Ii ,,~ tUtu[t f'Ht, fonn N' 4 ,'1I\,'" W, ,1 Jmlllltl,~ 1 • r' fa!jlll 1 h.

Cl111})(,,1 when a IH Inrg(l ILlHl t IUld fllilt. lIu,,1I "iltl f ~'21 ,It. II An u ,

But hy §§ 5111i2

~(ra f 'II). (/'1. rJ)«·fll'.,

.V - :4( r, • f.i r'f~ 'Y

~ i (l .. ~hf·CII)'.

(1'''' ('" + l'/),du')'.

Thl1H 1},likf· 'f1 and "', if4 n fUllrtanu qf'l)' (m .... " .. ,I I*,U. ""1'''''' ,,~~. It be tabuln.t.ocl wlth thel "rgtunNl~ 'J. wlu~w

'IJ" 2kt/(" •.. r,)' ,. tlHUI i'Y(2 f ('t,f'j40fJ (Ot &llschlngt~r'N Tcif,ln; No xxu II.)

whoro

55-57 ]

~n an Orbit and the Time

59

57 In the case of the branch of an hyperbola concave to the focus of

ht = ab [sinh € - sinh 0 - ( e - ~)}

SInce h = v(J.1.p) = nab And, If (XH f!h), (xe, Y2) are the exbremitres of the arc, twice the area of the focal triangle IS

2~ = X2'Y1 - f1J1Ya

by (13) Hence

smh e - smh ~ - (€-~)

'Y = sin h € - sinh 0 - sinh (€ - S)

Now we have by (14) and (15)

(27)

or

2 cosf~rlr2= 2a (cosh h - cosh g)

where 2h = e + S, 2g = is - 8 Also by addrtion of the same equataons (14) and (15)

r1 + r2 = 2a (cosh 9 cosh h - 1)

and therefore

Ill(rl + '1~ - 2 cosfcoshg ~rlr~) = 2f.'t'/(2cosj vr1r,.)' (28)

y = nt/(2 sinh 9 cosh h - sinh 2g) = ant/smhg(2 COSjVTITg)

sinh 2g - 2g

= 2 smhg (cosh II, - cosh g)

a smh 2g-2g

- 2 cosj"'!rlTe SInh 9

(29)

As In the case of the ellipse we wnte 1 + 2l = '191 + r, , 2 cosjVrlrs

Jd~

m/A- -

..... (2 cos!Vr1r ... t

60

Relations between two or mot e Pontions

[OIl V

and thus (28) and (29) become

y~ = m',/(l- smh2tU)

'//- 'll = m~ (sinh 20 - 20) / sinh" 0

(30) (31)

ThIS paJr of equations In y and 9 must be solved by some process of apploxlmatron so th.a.t the value of'Y may be found

58 The case of the branch whIch IS convex to a centre of repulsIve

force at the focus (- ae, 0) needs shght modifications TWIce the area of the

sector IS ~by (21)

ht= ah(smh e>: sinh B+ e - B)

while twice the area of the tnangle IS 2A - ~YJ f&2Yl

= ab [sinh Eg (cosh El + e) - sinh EJ (cosh E2 + e)}

- db {smh (E2 - EI) + 2e sInh t (Eg - EI) co~h i' (EJ. + EI)} = ab [sinh (e - 8) + sinh E - SInh 8}

by (18) Hence the ratio of sector to tnangle IS smh E smh 8+e 8

'!f = sinh (e - 8) + sinh e - sinh B

(32)

In thIS case we have by (19) and (20)

16at cosh' ieCOShlii8 = ('1'1 + '1'2)9 - 09- 4rlTIl cosif

or

and

2 cosjvrlr2 = 2a (cosh h + cosh g)

r1 + r2 = 2a (1 +- cosh It cosh q)

where 2h= e+ B, 2g = e - 8 Hence

1-

2 cosfcosh g "trlr), (rJ + r2) - 2a smh2 9

But (32) may be wntten

y = nt / (sinh 2g + 2 SInh 9 cosh h) = ant/smhg(2cosjvr1r,)

a.n.d therefore yl (2 cos j cosh 9 .j '1'1 '1'2 - '1'1 - rl) = 2p.'tO / (2 cos j"; r 1 r2)2 since ngal= p.' Also by (32)

1 _ _ 8mh (e - 8) - (e - cS)

'!J - sinh (e - ~) + sinh e - sinh S

_ smh 2g - 20

2 sinh 9 (cosh 9 + cosh h)

(33)

a

=---=

2cos!vrtr2

smh2g-2g sIDhg

37-59J

tn an Orbit and the Tvme

61

ence

4

If as before we write

1 + 2l= r1 + rg 2 cos!Vr1r2 '

then (33) and (34) become y'J= m'J/(cosh'J.tg -l)

(35) (36)

y2 _ '!/ = m2 (sinh 2g - 2g) / smh' 9

and these agam, when solved by a method of approximation, grve the value of Y In tlns case when rl. r, and! are known

59 Some useful approximationa can be obtained from

W lC 18 easily prove et e any regu ar ncnon 0 t

powers of t beyond the fourth order we may write

=

Let Xl' XII' X8 be the values of X when t= - Ts, 0 and 7"1 Then we have

three palrs of equations, obtamed y su statuting t ese va ues In e a ove From these SIX equations the coefficients ao, , a" can be eliminated and the result expressed m determinant form IS clearly

Xl 1 - Ts 7l - T83 Ts" = 0

X2 1 0 0 0 0

s, 1 7"1 7"12 T1s T1"

s, 0 0 2 - 67"8 12T32

X2 0 0 2 0 0

X8 0 0 2 6T1 127"111

alculated without difficult , and the result after

(37)

62

Relat~on8 between two or more Poeuume

[011 V

60 Now In the case of the motion of two bodies In a plane we have

a; = - p,m I 13, '!J = - JL'!I / r1

Hence subsututmg a; and '!J successivelj for X In the formula jusb obtained we have, to the foul th order In the Intervals of time,

0= Xl7"1 (1 + p..dl/ rId) - X2'T1I (1 - fl'AJI r ,l) + as 7"s (1 + fl'.A .1 tIl) o tll7"l (1 + }-tAd) /) Y2T2 (1 P.:42/'1'-l) + YJ'IS (1 4- pJ1.s/ ' 'I ')

The solution of these equatIons In the ordmcl.IY form gIves

But the denommators Me respectIVely double the a.rea.s of the trl(l,ngles whQ!'o

SIdes are pans of Tl, rill 18 Hence we have the formulae of GIbbs,

where, according to the customary notation, [131 J] denotes. double the ,trait. of the tna.ngle whose SIdes are r.., rs. and .AI> Au, .AJ ha.ye the 'h*ne~ found

above (37) ThIS expresses the ratio of the tnangles correctly to tho thud ord er of the time mtervals

A second mterestmg example 1S provided if we take X == r2 In this (".1.I..,C'

we have (~ 25 and 26)

dt2r Hence the formula (38) gIves

= - {Tl (7"27"3 -7"12) + T2 (7"1 7"J + Tall) + 7"3 (7"17'2 - 7"l)} !J- /Ga == - (3TI7"t7"3 - "r13 + T~'1 - 7",3) p.16a

= - {3Tl 7"117"3 + 3Tl7"J (7"1 + "rJ)} p./6a

= - P.T1T2Ts/ a

(40)

The form (40) applies to an ellipse and gives the means of calculating an approximate value of a when rll r., ra are known It must be adapt, .... d to the hyperbola by changIng the SIgn of a For the parabola the nght-hund SIde vamshes and we have the relation between the three radn vectoros

?'12Tl - 't·l"r2 + rl'T3 = 2p. (A1Tt/11 + .AtT2/r, + AJ'TI/~ J)

which holds provided we may neglect terms of the fifth order In the time

**60, 61J in. an Orbd. and the Tvme 63
**

61 Returmng to ilie formulae of GIbbs (39), In which the denommators

are correct to the fourth order, we have

T) [117:a = ~+ p.AJ/r:·l_1 + p.As ""AI

TI Lr!l1,] 1 + ,.."A1/rI3 - TS3 - TIs

'C21Tl7'z1 = 1 + ,.."As/ rs3 = 1 + p.As + /1-A9

T3 [Tl7'l] 1 - ,.."Az/rl TS3 r,l

T!;l [7 2TJ] _ 1 + ,.."A1/ rIB _ 1 ,u.A1 /1-Az

7".TT T.l- 1 uAq/r"J - + TiB + rs

~ g

to the third order Bu t to the first order

1 1 aT2

- = ----T

r3 rS r~ 1

8 2 2

, , ::t.,..

rlS 8 + ;Ts

...... T2 rll

.... 1."l.lUI;;

T1 [TIr!!] _ 1 + ,.,. (.As - AI) 3p.TS(A ..d.)

_ --- --- 3Tl+ ITs

TslTJra] TzS Til"

T2[11T2]_1 ,..,,(As+Az) 3}LT2A

_ + - - aTl

7"3 [1' 113] r2:l 7'24

7"2 [? ArB] _ 1 + ,u (AI + Ai) + 3}LTs A

... -r')". ')"~' - T,,3 T9." 1 T8

• L • ~ ...

For the coeffioients we easily find from (37)

12 (A9 + As) = TITs + 7"22 + TIT2 - 7"82 = 2 (T,l- Ta9)

12 (AI + A!l) = 7"1 T3 + 7"22 + TzTs - 7"\' = 2 (7"iS - T{I.)

, '" I , A \ 2\ - (- ... .... 9'\ or.S -I- 7" 8

J.~ \.Ll.S"'1 + .0.1 3) "'1" 1 \I 'l;I I '3, . ~ ." ..

J1.nN t.h- rB.L'· 'R

[r, r ,] To {I" pro (0 O)}

n == - 1 + 63 (7"1~ - T82) - 44 T1 + 7"s

TzTS T1 T2 T2

[r, r,] To {I" ) 1"", ( .) } (41)

[l = - 1 + (f! (T').2 - TS'l. - 4r:i TIT'). - '7'8 7"1 >-

r1'l3 7"2 T2 !;l

- ~ )

lr2r~ 7"1 {I r: ( '). 7"12) + ';' : (T'J.Ta

[ ] - + ----a T!rl 7"{".) 7"1I}

rITa '1"2 6r2 T9

These formulae are correct to the third order and If the terms mvolvmg

r2 be omitted they express the ratios of the manglee in terms of the single

distance Til to the second order Hence then value for the determmation of

orbits 64 Relat'to'J"tS betuees: two or more Poslt~on~

62 WIthout Joss of accuracy the ratios can be expressed III L~:t:"H two distances r1 and 1'"3 instead of T2 and r2 The forms fouQ-Q. 1 ,;

may be derived thus we have to the first order

rl = t , - r2TS, '18 =? ~ + '} .Tl

whence

and therefore

or

In the terms of the third order we have simply

Hence the ratlOs of the triangles to the requu ed order become 4

where, lftH ~, t, are the tnnee corresponding to the dIstances 'lJ r~:. ,. ", , t

'T]: ta - t~, 'rg tJ - q, 7'8 - t~ tl

Equrvalent but rather simpler expressions In terms of the extreme <lit

may be obtamed by obserVlng thdt

1 1 3'2 1 1 3r J

3 ~ + 4 'rJ. J ~ ~ '1"1

whence

By substitutaon In (41) lt 18 easily found that

•• _ ... - 4

[r11'"1l] - '1"3 {1 + f.t ( + 2) 71 + P. TJ' - TsI}

---- -- TITs T2 - --_

[, IrS] TSl 12r1.i T~ 12r.J t,

[rsrs] _ '7"1 (1 + iJ. Tl- T13 + p. ( A) Tl}

--- ~ -- - TT +'1". -

[r1r3] Til ( 1211J T2 12rJ1 1 d .! Til

From the method by which all the expresarons of this kind have been <1 It IS clear that the results apply equally to all undisturbed orbits, ell i I hyperbolic

CHAPTER VI

THE ORBIT IN SPACE

63 HItherto we have considered the relative motion of two boches only as referred to axes In the plane In which the motion takes place It IS now

We take a sphere of arbitrery umt radius with the Sun at Its centre The echpnc for a gIven date 18 a great circle on this sphere That hermsphere which contains the North Pole of the Equator may be called the northern hemisphere On the echpne IS a fixed pomt 'Y which represents the equinocnal point for the gIven date and from which longitudes are

cZ~nat~on and may be denoted by ~ It IS an angle which may he between

Let P be the pomt on the great circle of the orbit wlnch represents the radius vector through the perihelion and Q any other pomt on the same great circle represenung a radius vector WIth the true anomaly 'W, so that PQ = w We may denote the arc np lying between 0° and 3600 by t», so node to

point on the plane of the orbit, IS called the argument of the lat'ttude It IS

to define the element 'Gr, whieh IS called the long~tude of pB'Mhel'tOn, as the sum of the two angles n. + Q) although only one of these IS measured along the eolipnc The angle til' + w or n + 6) + W 18 called the long~tude m the orbt.t We have thus defined the three elements, the longitude of the

The O'rb~t tn Space

[OH VI

idmg node, the mchnanon of the orbit and the longitude of perihelion, ired to fix the posmon of the orbit In space and with these It IS Isary to mention the date of the echpnc and equInox to which they eferred

!t The motion must now be de£mtely related to the bime Let to be ooch arbitranly chosen and T the tune of perIhelion passage Then,

o.g the mean motion, the mean anomaly corresponding to the epoch 18

63-

at (B

tb

Mo- n(to- T)

rther Yo or T might be regarded as an element of the orbit but ill the

p ane ary or 1 1 IS more usual to employ the mean lonfPtude at )ooh, E, which IS defined as the sum 'GT + Mo Thus at an

W IS e ongrtu e In the orbit and E the eccentric anomaly, the on of the planet 18 given by

ta

tani(u-m-)=

E - e sm E = M = n (t - T)

n

1

lean motion and the mean dIstance are connected by the relatnon (~ 24) na~ #' - kN (1 + om)l

m 18 the IDd.SS of the planet (ne h ble In the case

)mp ete e ements can now be enumerated and Illustrated by the case of met Mars

**MalB (m lid 093 500)
**

Epoch to 1900 Jan O,Oh GMT

ean ongitude E 293" 44' 51" 36

Longitude of perihelion 'ZiT 3'34 13 6 88 Equinox

Longitude of node n 48 47 9 36 19000

Inclination ~ 1 51 1 32

Eccentricity e 009330895

Mean motion n 1886'1 51862

Log of mean distance log a o 1828970303 rmber of independent elements IS SIX, correspondmg to the SIX conof integration which enter In to the solutaon of the equations of motion, lemg ill their general form three In number and of the second order

en the orbit IS parabolic the eccentncity IS 1 and the mean distance rte The scale of the orbrb IS mdieated by the penhehon distance q ~ time of penhehon passage T IS given Instead of the mean longitude

63-6~ ]

The Orb'bt in Space

67

a.t Il. chosen epoch Thus prehmmary parabolic elements of Comet a 1906

T

Cd

n

1905 Dec 2229263 GMT

89° 51' 53" 7 }

286 24 22 1 19060

126 26 7 3 1296318

65 If axes 0 (0;1' Yl, Zl) be taken such that OXI passes through the node,

1 , N ole of

the orbit, the coordmates of the planet (or comet) are flh=rcos(ld + w), YI=rsm(Cd+w), Zl=O

when lts true anomaly IS w Let the axes be turned about O»; so that OYl takes the position OY2 In the plane of the echptae and OZ'J IS directed towards

e equi-

N ext, let the axes be turned about OZ'J so that Oa:s passes t roug noctml point and OYs 1S in longitude 9Uo Then

.La == 0)11 cos n - '!J2 sin n, '!Js = '!j2 cos !l + X2 sm .n, Za = Zli Hence the relataons between (xJ' YJ, zs) and (XlJ '!JlJ Zl) are gIven by

cos!}

- cOB~smn

sm s sm!1

o

cos ~

sm s

'rhlS scheme will glve the hehocentnc echptic coordmates of the planet It IS convenient to write

sin a sin A = cos.o., sin a cos A = - cos s sm!1

sm b' sin B' == sin .0., sm b' cos B' = cos ~ cos

a:a=rsmasm +Cd+W

'lis == r sin b' sin (B' + co + w)

Zs =rsm~ sm(Q.) +w)

Hence, If R, L17 Bl are the geocentrIc distance, longitude and latItude (the last alwa s a very small angle) of the Sun, which may be taken from the

Nautioal Almanao, and fl., A., e ge

of the lanet,

A cos 'X. cos/3 = R cosLl cos Bl + r sm a sm(A' + Q.) + w A SID I\. cos /3 = R 81 n L, cos B, + r SIn b SIn (B' + Q.) + w)

A sm/3 = R smBl + r em t sm(oo + w)

whence the geocentnc echptIc coordmates of the planet

68

The Orbit in Space

[oa VI

66 Were the elements given with reference to the equator Instead of the eelrptic, and this IS sometnnee done (though not often), the same formulae would give equatorial coordinates with the substrtutaon of R A and dechnstaon for longitude and latitude To obtain equatonal coordinates from ecliptic elements another transformataon IS necessary Let the last system of axes be turned about Oa,J so that O//J comes mto the plana of tho eq119.tor and the new aXIS 0,4, 18 dIrected towards the N pole of th(' equa.tOl

Then the obliquity of the echpno bemg denoted by Eo,

llI, XS, y, -Ys cos eo £3 sin co, s, 53 COS eo +Ya sm eo

From the above relations between CXal YJI Zs and Xh It follows

Sin a sm d smb smB

SIn a cos..+ SlII b COS B

cos a. cosb

z, sin o sin 0

sin c cos a

cos c

where It 18 easily seen that

sm a sm ..A cos n

sin acos..A = - COSt SID n

cos a SlD~ smn

smb sm B =

COSE'o sm.n

slOb cosB = cosb

COb eo COS ~ cos n - sin eo sin /, cos eo SlD ~ cos n - sm fa cos,

sin C '3In a = smc COS e

sin s, smO

SIU Co COS 1. cos n + COb eo S1 n ~

The hehocentno equatonal coordmates of the planet now become OJ" = r 8m a, 8m (.A + 00 + w)

y, = r SIn b SIn (B + (j) + w)

Z" = r sin C sin (0 + (j) + w)

Thus, for example, the above elements for Comet a 1906 lead to

tli" = r [9 803389] Sin (243° 29' 42" J + w)

'!14 = r [9 9998JO] sm (331 33 15 1 + w)

$4 = r [9 887772] SIn (60 14 19 5 + w)

referred to the equator of 1906 0

Let (w, Y, Z) be the geocentrIc equatorial coordinates of the planet and (X, Y, Z) the correspondmg geoeentnc coordInates of the Sun, WhICh may be taken chrectly from the Nauttcal Almanac or other ephemeris 'I'hus

oIi=X+a,,, y=Y+y~, z=Z+z,

66 67

Orbit ~n

69

ace

But

[1)= cosacos ~ y= Sinacos, Z= sm where A, a, 0 are the geocentnc distance, right aacension and dechnamon of the planet These coordinates can therefore be calculated from the equations

A cos a cos 8 = X + r sm a. sin (A + ro + w)

A sin a cos 8 = Y + r sin b sm (B + ro + w)

A sin 8 = Z + 'I S10 C sin (0 + ro + w)

ThIS form of equations, introduced by Gauss, IS very converuent for the systematic calculation of positaons In an orbit

Let 'YAB represent the first

The new

B=1800_"" 0=",'

N ow the analogies of Delambre may be wntten In the smgle formula, easily remembered,

sm {45° ± (450 _ ibTa)} sm {45° + (450 - ,.B±"""A)}

sin {45° ± (450 _ to)} = cos l45° + (450 _ to)l

where the ambrgutfnes ± =+ must be read consietently but mdependently In two sets of three Hence takmg (1) all lower Signs, (2) all + SIgnS, (3) all _ SIgnS and (4) all upper SIgnS In the above formula, we have

sm 1 (!l' - flo + Cc) - ro') sin ,,,,' = sm 1 (n - 01) am t (", + 'lrl) cos fl' - n + Cd - ro' sin ",' = cos t (!l -~) sin t ('lr _~)

preC6S8'tOn IS glven by

tan!(D/ - III - "I - Aru) = tan! (n - Ill) sIn! ('" + 7rl)/sm !(t - '11"1) tan t (0' _ III - VI + AM) = tan! (n - TIl) COS !(t + 7rl)/COS l (t - '11"1)

70

The Orb~t ~n Space.

[OR VI

where ACd = Cd' - co~ and (by Napier's analogy mvolvmg B + a and.A)

t 1_( ') cos!-(O+!}.' - 2TI1-'h) t 1.

a0'2" ~-~ = cost(O-n' + "h) an"2"?1'1

68 When the mterval t 18 moderately short, however, these rigorous e uataons for the effect of recession are not required and It 18 more COD-

vement to use differentaal formulae We now consider 'YAB as the xe echptic of 1850 ° and '}flAG as a vanable echptIC Since

cos 0 = sin A sin B cos 0 - cos.A cos B

or

dO = - cos b d.A + sin a sm B do

(1)

Also, since

sm o COb b db=smBcosc de -cos Osmb dO

= sm B (cos c - cos a SIn a sin b) de + cos 0 sin b cos b dA

or

SID

SJIDJ]aily, SlDce

smOsm a=smA sm c

sm a cos (J, • da = cos .A sin 0 dA + sm.A oos 0 do - cos 0 sm a dO - (cos .A sm G + cos a sm a. cos b) dA

+ (sin A cos c - sm A cos a sin a sin b) de

= cos a sm b dA + SID A cos a cos b de

or

sin 0 da. = sin b dA + S10 A cos b de

(3)

By a. shght change of notation we now put no, Cdo and 1.0 for the elements a.t T = 1850 0, n, Cd and 1. for the elements at time T + t (instead of n/~ ro' and '1,') and defme the posmon of the ecliptic and equinox at T + t relative to those at T by 01 = II, 1.1 = Tr and no = II + V', so that

a=~-~ b=n-rr-~ o=~-n

.A ='1t', B = 1800 -1.0, O='l,

Hence by eubstrtutaon 10 (1), (2) and (3)

d~= -cos (n - II -'0/') d7T'- sm(Mo - ro)smto dIT SIn 1. d (n - n - ""') = cos 1. sin (n - n - V') d7r - cos (roo - CI») sm s, dII

- SID'l, dm= sm(!! - n -'0/') dTr- cos (0 - n -y.) SIn 7T' dn

in the coeffICIents of dn we ma ut ~ = to, 6) = roo and 'IT' = 0, this bemg

the mutual mclmation of the fixed and movmg echptae Hence we have sun ply

dt / dt = - cos (n - IT - '+') d'TT"/ at

dnl dt = d'l/l'ldt + cot ~ sin (n - IT -..;) d'Tf'ldt dO) fdt = - cosec s sm(n - IT - '0/) dnr/dt

These are to be mtegrated between t = tl and t = ~, and the coefflcients of d'1r / dt are vanable with the trme Provided the interval IS no more than a few years, It IS sufficiently accurate to proceed thus Writmg

- t cos n - II - d'lT' dt

we take II + V' d7rfdt and d",jdt from appropnate tables (eg Bauschmger's

Tafeln, No xxx) with the argument T +! (t: + tl 1 ...: 1 =

a roximate values of nil, '1.2 can be obtained and the caleulatnon IS then

repeated with the corresponding values i n and z

69 It IS impossible to correct the first observataons of a moving body for parallax In the ordinary way because Its distance IS unknown But the line of observation intersects the plane of the ecliptic in a certain point, called by Gauss the locus fictu«, the posraon of which can be calculated If the observation IS then treated as though made from tlus point the effect of parallax IS allowed for and also the latitude of the Sun

Let the observation be made at SIdereal time T at a place whose geocentnc latrtude IS q, Let ex, 8 be the observed R A and declmation, reduced to mean equinox The geocentnc equatonal coordinates of the place of observation are (p cos ¢ COS P, p cos ¢ sin T, p Sin tk), p bemg the Earth's radius at the place, and the correspondmg echptic coordmates (p~, pht, phs), where

~ = cos l cos b = cos cf> cos

= 1 l cos b = cos Sin T cos eo + SIn 4> sin eo

ka = Sin b = sin <p cos eo - COS 4> Sin T sin eo

trtude of

H~ = cos X cos fJ = cos cos a

=~x~ =~S~a~~+~8~~

&=~~ =~8~~-~8~a~~

are the direetaon cosines of the hne of observation, X, ~ being the geocentric longrtude and latitude of the observed object The Na'Utwal .Almanac gives R", L, and Bl the geocentric radius vector, longrtude and latitude of the Sun.

72

The Orbit in Space

[OH VI

Hence in heliccentne ecliptic coordmates the equation of the lme of observanon 18

a; + R,. cos Ll cos Bl - htp '!f + ~ sm Ll "OS Bl - hap

HI = Ht

_ z + Rl sm Bl - hsp == _ A - H8 .Q

where a IS the distance from the place of observanon to the point (a:, y, z) posmvely In the direction away from the object If then tlns line mtersects

the plane of the ecliptIc 10 the point (the locus fictus) a; - R cos L. y = R am L, Z 0

A == (hsp - Rt sin B1) I H;

R cos L - Bq cos Ll cos ~ + p~ - (hsp Nt sm B1) HdDa

- R sin L = - B, SID L, eoe B, + p~ - (haP - ~ sm BI) H'J/Hs

But these exact equatIOns ea.n be slmphfied, regard bemg had to the small

quantities mvolved For B, < }" In general, so that sm B, =B1, cos B, = 1 Also we may put p -pRl where p IS the solar parallax, 8" 80 Hence wntmg

l1 Rt (hap - Bl)/Hs

ece L, ss; +~smLl si; -pRlh,. (haP-Bl)RIHdHa

- sm L, dRl - B, cos L, dLI - pRI~ (haP - Bl) RtH2/ Ha whence

- d~/~ =p(h" cos Z, + hasm LI) - (h,p- B1) (HIcosLI + H2smL1)/Ha dLI p (hJ sm L1 - h.,. cos L1) (Asp - B,.) (HI sm Ll - Da cos LI )/Ha

or agam

dBx! ill P cos b cos (k &) - (p sm b BI) COS (L1 - XI) cot f1

dh =p cos b SID (LI-l)- (p smb - B1) SIn (L1 - A.)cotf3

AI ~ (p 8m b B1)/sm {j

Here both p and Bl are naturally expressed in seconds of arc Thus dLl~ the additrve correetion to the Sun's longitude, 18 appropriatelv expressed 10 the same unit The Naut~cal Almanac gives log R,., to which the sddrnve

con eetion 18

d e; loglo E d R"

d log ~ = R; 206265" = Rt [4 3234 - 10]

Fmally, had the observamon actually been made from the locus flctus It would have been made later ill time by the interval required for hght to travel the distence Il. But the light equation, or the time over the mean distance from the Sun to the Earth, IS 49885 Hence the additive correctJ.on to the trme of observanon 18 (m seconds)

1:1 49885 A

dt = R,. 206265" = It,. [7 '3832 - 10]

The reduction to the locus £ictus IS a refinement rarely employed in practice

Slnce

, ,

CHAPTER VII

CONDITIONS FOR THE DETERMINATION OF AN ELLIPTIC ORBIT

70 There are certain propernes of the apparent motion of a planet or a

true orbit and which can be considered WIth advanta.ge apart £rom the details

o numenca ca cu a IOn W IC are necessary or a prac iea so u IOn. y are closely connected WIth the direct method of solumon devised by Laplace, but they equally contain prmciples which are fundamental to all methods.

Let (~, y,.e) be the hehocentno coordmates of the planet, (X, Y, Z) the heliocentric coordmates of the Earth Then

u; = - p.x/r*,

p. = k' (1 + m), IJ.o :;:: lei (1 + mo)

m and mo being the masses of the planet and the Earth Let (a, b, c) be the correspondmg geocentric direcnon cosmes of the planet, so that

z=Z+o 1

p being the geocentnc distance of the planet The observed POSItIon of the

pianet is grven m ng aseeneion an system of axes be chosen,

a = cos a cos 8, b = SIn a cos B, e == Sin 8

I , I

I

I

g;=X+ap+2ap+ap

JUC/"'" - lI-oX / R' + iLp + 2ap + ap ==

or

X (Plr*- Po/B') + ap + 2ap +a (p+p.p r =0

and smnlsrly

y (p./r' - I-'o/.RJ) + bp + 2bp + b (p + p.p/TA) = 0 Z (P/TA -Il'o/ RI) + cp + 20p + 0 (p + p.p/r') = 0

", I

I 1

74

Oonduum« for the Determ'tnation

[OR vn

These are three equations m P, p and p + fJ-p/".s, the solution of which can be wntten down at once In the form

-p 2e_ p-1r3 - Po / Ra (2)

XI - X -

a a a a a a a

b b Y b b Y b b b

e c Z c o Z o e c

the value of p not bemg required 11 The determmants In (2) can be calculated when the first clnd ~econd denvatrves of the three direction cosmee are known Now

a=-smacoso a-cosasmo 1)

a=-SlDa.cOSO a-cosa.coso (li+2smasm8 a8-cosacos8 &-coslXsmo S

0= coso 8-smo 82

The derIvatIves ll, a, 8, 0 are most SImply calculated from a senes of obseI vt·d

values by Lagrange's mterpolation formulae If the number of observationa IS three, made at th~ tImes ti, t;I, ti, we have accordmg to thIS rule,

It ~) (t t;) (t fa) (t t1) (t~) (t - t;)

2a1 2(l2 2as

(1- (tl-~) (~ - ta) + (~- t,) (tg - ~) + (ts -~) (ta- ~)

or, If we choose t - ft, the tIme of the middle observatwn,

a.=ag

7'17'S7'3 a = - 7'll. (l1 + 7". (7"1 -7',) all + 7",2 txa = 7"111 (as - Cl1) + 7"'111 (a,J - Ql) 7"17'II7"a ex = 27"1 (ll - 27"11 ~ + 27"s as = - 27'1 (ag - a1) + 27"s (CX1- all) where

7"l=ta-tlh 7'2=ta-th 7"8=~-~

These formulae, which apply equally to the deelmanons, mutat1s mutn.n<iu;, are only correct rf the observataone are made at very short mtorvals of time and are Ideally accurate SInce the accuracy of observataons has practical htmtamons, moderately long Intervals must be used and a greater Dum ber of observed places IS necessary for satlsf.a.ctory results Our im mediate concern, however, 1S rather WIth general principles than practical methods of calculation

7<r-73]

oj an Elliptic Orbtt

75

72 It IS now possible to calculate the qnantlty l gIven by

l= a a OJ -fr2 a a X

b b b b b Y

o C Z

o c c

and we then have by (2)

orbit and this IS one relation between p and r In essence It IS fundamental 10 all general methods of Imdmg an approximate orbit A second relation IS available because we know the angle t between R and p, namely

r2 = Ria + pi + 2Rp COs t (4)

If r be ehmmated between (3) and (4) an equation of the eighth degree In p results, and It will be necessary to examme the nature of the possible roots For the moment we suppose that the appropriate value of p has been found Then the corresponding value of p IS gtven by (2) and the components of the

fJJ=X+ap+ap, y=Y+bp+bp, z=Z+op+op (5)

where X, Y, Z must be found from the solar ephemens by mechanical differentiation Thus when p and p are known, (1) and (5) give the three nehocentric coordinates of the planet and the three correspondmg components of velocity at a given tune t From these data the elements of the planet's orbit, assumed for the present purpose to be elhptrc, can be calculated Without

difficulty

73 Since equatonal coordmates have een use 1 e e euipnc

elements of the orbit will also be referred to the equatonal plane If new coordinates (e, "', t) be taken so that the aXIS of E passes through the node and the a.XIS of ~ through the N pole of the orbit, the transformatIon scheme IS (cf § 65)

'1J - SID 11' cos ~ I

cosO' COSt'

sin 1,'

~ smO' sm s'

- cos .{}/ am 1,'

COs~'

76

Oond'tt~ons for the Determinauo»

[OR VII

Hence 10 the plane of the orbit,

~=msm n/sm~' -ycosn'sm~' +ZCOS~/=O ~=a:smn/sm~' -yCOS.n/SID~' +zcos",'=O grvmg for the determination of!},' a.nd ~'

SIn .0' SID ",' cos 0' sm ",'

(6)

'!I.e yz lU lCZ $Y W!J

Also, If u IS the argument of latitude (or rather of deelmation),

and

E = 1 cos U = x cos 0' + Y sin il'

(7)

'7} = -::r; sin D/ cos ",' + Y cos 0' cos ",' + z sm s '

or

r sm'U z cosec t'

(8)

by the above equation for ~ Similarly, If V IS the velocIty and X the angle

between V and the radius vector produced.

f V cos (11 + X) 0; cos n' + Y SIn n I (9)

1'J = V sin ('1.£ + X) = s cosec ",'

(10)

Thus V and X' as well as r and u, are determmed Now If W IS the true anomaly at the pomt, the polar equatIon of the orbit gtves

p = r (1 + e cos w) (11)

pcotx resmw

SInce tan X - rdw/dr But the cons-ant of areas IS

(12)

It = Vr sin X = tJ(p.p) = k 'l/p

(13)

gt~g p and hence 6 and w The mean distance a can be deduced from the known values of p and 6, or directly from the relanon

V2 = 2p./r - ILia (14)

and the mean motaon n from the equation p. = 1c2 = n9aS Also the element 'riJ" 18 gtven by v' = {}' + 'Ur - w Fmally the epoch of penhehon passage IS determmed by the two eqnataons

tan!E=JG ~:) tdIli-W

n(t- T) =E-esmE

E bemg the eccentric anomaly at the pomt of the orbrt observed

74. We now return to the considera.tlon of the solution of equations (3) and (4), following the method of Charher, which gives the clearest VIew of the geometncal condmons of the problem The first of these equataons IS based on the assumption that the point of observataon IS moving under gravity about the Sun 'I'he point which so moves IS In realIty the centre

(15)

even degree), and two positive roots must satIsfy the equataon, namely + 1

,

73-75J

oj an Elliptic Orbit

77

of gravIty of the Earth-Moon system and, stnctly speaking, the observanons

IS pam an no e cen re 0 teart u t t IS IS

a matter of detail which our immediate purpose does not require us to stop and consider SImIlarly we may neglect the mass of the Earth as well as that of the planet and put R = 1 Then the equations become SImply

lp=1-1jr3 (16)

either p or r has been found, and It IS simpler to ehmmate p Thus l21.8 = l2~ + 2lr3 (1" - 1) cos t + (1" - 1)2

or

l2".s - (l2 + 2l COS t + 1)".s + 2 (l cos t + 1) ,.a-l:11: 0 (18)

2 (l cos Y + 1) = {(I - 1jr) (,-2 -1- pi) + 2p9.}J p2

== {(I-ljr) (,-2 -1) + p9. (1 + IJr)}/p2

which ]S obviously POSItIve, whether r IS greater or less than 1 And the coefficient of r6 IS essentaally negative Hence, by Descartes' rule of SIgnS, there are at most three POSItIve roots and one negative root The latter certaml exists because the last term IS ne atrve the e uamon bem of

be a fourth real root, and therefore m all three real and POSItIve roots, ODe real and negative root and four lmagmary roots But the thrrd POSItIve root mayor may not satIsfy the problem

Now by (16) r IS greater or less than 1 according as l IS positive or ne atIve If then the two roots which are ill uestaon he on 0 osite SIdes

of 1, the spunous root can be detected and a umque solutaon of the problem can be found But If the he on the same SIde the cannot be discrnnmated

between in this way, and an ambiguity exists If we divide (18) by (r-l), we obtain

f (r) = l2r6 (r + 1) - (2lr cos t + r3 - 1) (rJ + T + 1) = 0

Thus

1(0)=+1, f(+1)=2l(t-3cQsy.)

So that the roots l

by + 1, and a unIque solutnon exists, 1

75 The geometrical mterpretanon 18 mstrucnve The equation (16) for different values of the parameter l represents a family of curves m bipolar coordinates, the poles being E (the Earth) for p and S (the Sun) for r The planet hes at the mtersection of one of these curves WIth a straaght hne

78

Conditione lor the Determinahon

[OB VII

---------------------------------_._--------------------

Wl1

, , ,

•

I

I ,

I I

I

I

I

I

g1

-' Q)

~\ cw::. C"I

I

I

I

\ \ I \

\

\ ,

,

\

\

,

\

\ , ,

,

or In rectangular coordinates,

II

of an EUiptw Orbit

79

drawn through E In a gIven direction But there may be two mtersecmons,

p'l (l- 3 cos t) = (I-Ijra) {I-lIra + i (1 + pB -r2)}

IS posrtive This expression changes SIgn when we cross the circle r = 1 and agam when we cross the curve

Puttmg pi = 1 + r2 - 2r cos ¢ we get for the polar equation of this curve with

4 - 3r cos cf> = 1/1"

(19)

I II

r(4- 3x)= 1

showing that the curve has an asymptote 3a: = 4 MOVIng the OrIgm to E

, -

curve consiste of a loop crossmg the BE aXIS at the point r = 5604, </J = '1r, and

,

The plane of the figure IS that contammg S, E and P (the planet), It IS only necessary to show the curves on one SIde of the aXIS because this IS one of symmetry

A few curves of the family (16) are also shown In the figure, for values

circle r = 1, called here the I' zero" circle It IS evident that when l IS

IS positive r > 1 and the curve lies entirely outside this CIrcle When l has a large negatrve value, the curve consists of a SImple loop surrounding Band an Isolated conjugate point at E As -l decreases from 00 the loop mereases m SIze until, when l = - 3, the loop extends to E, where there IS a cusp .Afterwards as l a roaches 0 the loop, still passmg through E, approximatea

II

more and more closely to the zero Circle

oop passmg roug s e va ue 0 a s e oop expan 5, en mg

to enfold the zero CIrcle Finally, when l = + 0 2959, It reaches the axis agaIn

must be remarked that l cannot be greater than + 3 For l = (r -1)/"sp = (r-1 + r-2+r-3) (r -l)/p

But r > 1 and r - 1 < P Hence the limit 18 established and we have only to follow the values of l from + 3 to 0 At first the curve consists of a small

and forms a node on the further side 0 As t e va ue 0 a s sta rt er the curve breaks up into two dietmct loops The larger contmues to expand outwards at all pointe and recedes to mfimty. while the Inner, always passmg through E, contracts until finally It becomes the zero circle These features In the development of the family of curves will be evident m the figure

80

Oonditto'JUJ for the Determination

[OH VII

It will now be apparent that the Inmtmg curve and the zero circle divide space Into certam regrons and that the solution of the problem of determmmg an orbit by the method mdicated IS unIque or not according to the regIon In which the planet happens to be Thus we distmguish four cases

(1) If the planet IS within the loop of the limmng curve there are two solumons

(2) In the space between the loop and the zero cIrcle the solutIon IS

UnIque

(3) Outside the zero CIrcle and to the left of the asymptotIc branch of the lmntmg curve there ale agaIn two solutions

(4) If the planet hes to the nght of the asymptone branch of the hmltmg curve only one solutIon IS possIble It happens that newly dlS-

covered minor planets are usually observed near opposrtaon and therefore

thIS 18 the case whIch most commonly occurs

76 There IS another curve whIch has cODstderable Importance m the

problem of determmmg an orbit by a method of approximation and to which Charher has gIven the name of the" smgular" curve We may find It thus

If we ehmmate r between the equations (16) and (17) we have

lp = 1 - (1 + 2p cos". + p~) - I

whlC.h 18 an equatIon gIvmg the values of p for a !me drawn through E 1ft

the cbrectlon" Two of the values become equal and the hne touches the curve (16) If

l = 3 (cos Y. + p) (1 + 2p cos '+' + p2) - i

= '3 (cosY. + p)Jr3

Hence the locus of the pomts of contact of the tangents from E to the famlly

of curves (16) 18

or

(l-l/.,.a)!p = 3 (cos t + p)jr3 2r2(r -1)=3 (p2+ra-l)

or a.gam

(20)

ThIS 18 the equatron of the smgular curve If we change from bipolar coordmates to the polar equation With the ongm at 8j we obtain

3 (1 - 2r cos c/J + ~) = 2r1' - 5r2 + ~

or

r=4 - 3 cos ~jr

(21)

Companson of this form With the equation (19) of the hmltmg curve shows at once that these two curves are the Inverse of one another With respect to