1: \documentclass[11pt]{article}

2: \usepackage{amsmath}

3: \usepackage{amsthm}

4: \input{epsf.tex}

5: 

6: 

7: 

8: 

9: \newtheorem{theorem}{Theorem}[section]

10: \newtheorem{proposition}[theorem]{Proposition}

11: \newtheorem{lemma}[theorem]{Lemma}

12: \newtheorem{corollary}[theorem]{Corollary}

13: \newtheorem{define}[theorem]{Definition}

14: 

15: 

16: 

17: \def\Empty{}

18: 

19: %

20: % Definitions that use @ : 

21: \catcode`\@=11

22: % Redefine section heading to get smaller letters (this

23: % is copied from the original in rep12.sty)

24: 

25: \def\section{\@startsection {section}{1}{\z@}{-3.5ex plus -1ex minus 

26: -.2ex}{2.3ex plus .2ex}{\large\bf}}

27: 

28: 

29: % Our own caption style --

30: 

31: \def\fnum@figure{{\small Figure \thefigure}}

32: \def\fakefigure{\def\@captype{figure}}

33: 

34: \long\def\@makecaption#1#2{

35:     \vskip 10pt 

36:     \def\FCap{#2} \def\NoCap{\ignorespaces}

37:     \ifx \FCap\NoCap

38:        \setbox\@tempboxa\hbox{#1}  % This is to avoid the damn colon.

39:       \else

40:        \setbox\@tempboxa\hbox{#1: \small \it #2}

41:     \fi

42:     \ifdim \wd\@tempboxa >\hsize   % IF longer than one line:

43:         \unhbox\@tempboxa\par      %   THEN set as ordinary paragraph.

44:       \else                        %   ELSE  center.

45:         \hbox to\hsize{\hfil\box\@tempboxa\hfil}  

46:     \fi}

47: 

48: 

49: % Our page heading style:

50: \pagestyle{headings}

51: \oddsidemargin -0.15in

52: \evensidemargin 0.00in

53: \textwidth 6.9in

54: \textheight 9.5in

55: \headsep 12pt

56: \topmargin -28pt

57: %\def\@evenhead{\rm \thepage\hfil \leftmark}%        Left heading.

58: \def\@oddhead{\hbox{}\rightmark \hfil \rm\thepage}% Right heading.

59: %\def\chaptermark#1{\markboth {\sc {\ifnum \c@secnumdepth >\m@ne

60: %      \@chapapp\ \thechapter. \ \fi #1}}{}}%

61: \def\sectionmark#1{\markright {\sc{\ifnum \c@secnumdepth >\z@

62:       \S\thesection.\hskip 1em\relax \fi #1}}}

63: %\def\thebibliography#1{\chapter*{References\@mkboth

64: %{\sc References}{\sc References}}\list

65: % {[\arabic{enumi}]}{\settowidth\labelwidth{[#1]}\leftmargin\labelwidth

66: %    \advance\leftmargin\labelsep

67: %    \usecounter{enumi}}

68: %    \def\newblock{\hskip .11em plus .33em minus -.07em}

69: %    \sloppy

70: %    \sfcode`\.=1000\relax}

71: 

72: \catcode`\@=12

73: %

74: 

75: 

76: \def\oplabel#1{

77:   \def\OpArg{#1} \ifx \OpArg\Empty {} \else

78:   	\label{#1}

79:   \fi}

80: 

81: 

82: 	

83: \def\lref#1{\ref{#1} (#1)}% Prints label as well as number

84: %

85: %

86: \def\defineterm#1{{\it #1}}

87: %

88: % Make a blank figure of given size with label in center.

89: %	\blankfig{label}{# of inches}{caption}

90: %

91: \newlength{\saveu}

92: \def\blankspace#1#2{

93:   \setlength{\saveu}{#2in}

94:   \divide\saveu by 2

95:   \vskip\saveu

96:   \centerline {\small \mbox{}#1}

97:   \vskip\saveu

98: }

99: 

100: \long\def\placefig#1#2#3#4{

101:   \begin{figure}[#4]

102:     \blankspace{#1}{#2}

103:     \caption[#1]{#3}

104:     \oplabel{#1}

105:   \end{figure}}

106: \long\def\blankfig#1#2#3{

107:   \placefig{#1}{#2}{#3}{htbp}}

108: \long\def\herefig#1#2#3{

109:   \par\noindent\vbox{

110:   \blankspace{#1}{#2}

111:   \fakefigure\caption[#1]{#3}

112:   \oplabel{#1}

113:   \vspace{\intextsep}

114:   }}

115: \long\def\botfig#1#2#3{

116:   \placefig{#1}{#2}{#3}{b}}

117: 

118: 

119: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

120: 

121: % Definitions that use @ : 

122: \catcode`\@=11

123: 

124: % page heading on even pages gives chapter, but on same side of page

125: % as for odd pages. 

126: 

127: 

128: \newcommand{\zz}{\mbox{${\bf Z} \oplus {\bf Z}$}}

129: \newcommand{\pa}{\mbox{$<_{\alpha}$}}

130: \newcommand{\pb}{\mbox{$<_{\beta}$}}

131: \newcommand{\pu}{\mbox{$<_u$}}

132: \newcommand{\poa}{\mbox{$>_{\alpha}$}}

133: \newcommand{\pou}{\mbox{$>_u$}}

134: \newcommand{\oh}{\mbox{$\overline h$}}

135: \newcommand{\og}{\mbox{$\overline g$}}

136: \newcommand{\spr}{\mbox{${\bf S}^2 \times {\bf S}^1$}}

137: \newcommand{\cs}{\mbox{${\bf S}^1$}}

138: \newcommand{\ch}{\mbox{$\widehat{\bf C}$}}

139: \newcommand{\pp}{\mbox{$\Phi_{\theta}$}}

140: \newcommand{\ii}{\mbox{${\cal I}$}}

141: \newcommand{\dd}{\mbox{${\pi_1(M)}$}}

142: \newcommand{\cl}{\mbox{${\cal L}$}}

143: \newcommand{\kk}{\mbox{${\cal K}$}}

144: \newcommand{\ca}{\mbox{${\cal A}$}}

145: \newcommand{\is}{\mbox{${\cal I}^s$}}

146: \newcommand{\iu}{\mbox{${\cal I}^u$}}

147: \newcommand{\res}{\mbox{${\cal R}^s$}}

148: \newcommand{\reu}{\mbox{${\cal R}^u$}}

149: \newcommand{\bp}{\mbox{$B^+$}}

150: \newcommand{\bn}{\mbox{$B^-$}}

151: \newcommand{\ba}{\mbox{${\bf A}$}}

152: \newcommand{\aaa}{\mbox{${\cal A}$}}

153: \newcommand{\bb}{\mbox{${\cal B}$}}

154: \newcommand{\ak}{\mbox{${\cal A}_{\kappa}$}}

155: \newcommand{\at}{\mbox{${\cal A}_{\tau}$}}

156: \newcommand{\ap}{\mbox{${\cal A}_{\alpha}$}}

157: \newcommand{\ab}{\mbox{${\cal A}_{\beta}$}}

158: \newcommand{\ag}{\mbox{${\cal A}_g$}}

159: \newcommand{\lat}{\mbox{${\cal L A}_{\tau}$}}

160: \newcommand{\lak}{\mbox{${\cal L A}_{\kappa}$}}

161: \newcommand{\lal}{\mbox{${\cal L A}_{\alpha}$}}

162: \newcommand{\lag}{\mbox{${\cal L A}_{\gamma}$}}

163: \newcommand{\lab}{\mbox{${\cal L A}_{\beta}$}}

164: \newcommand{\la}{\mbox{${\cal L A}$}}

165: \newcommand{\af}{\mbox{${\cal A}_f$}}

166: \newcommand{\uu}{\mbox{${\cal U}$}}

167: \newcommand{\ro}{\mbox{${\cal R}$}}

168: \newcommand{\vv}{\mbox{${\cal V}$}}

169: \newcommand{\we}{\mbox{${\cal W}$}}

170: \newcommand{\re}{\mbox{${\cal R}$}}

171: \newcommand{\wwx}{\mbox{${\cal WX}$}}

172: \newcommand{\et}{\mbox{$\eta _-$}}

173: \newcommand{\tsv}{\mbox{$\sigma ^s_e$}}

174: \newcommand{\hhu}{\mbox{${\cal H} ^u$}}

175: \newcommand{\hhs}{\mbox{${\cal H} ^s$}}

176: \newcommand{\hu}{\mbox{${\cal H}(\widetilde {\cal F} ^u)$}}

177: \newcommand{\hs}{\mbox{${\cal H}(\widetilde {\cal F} ^s)$}}

178: \newcommand{\hp}{\mbox{${\cal H}$}}

179: \newcommand{\tup}{\mbox{$\Theta ^u_+$}}

180: \newcommand{\tun}{\mbox{$\Theta ^u_-$}}

181: \newcommand{\tsp}{\mbox{$\Theta ^s_+$}}

182: \newcommand{\tsn}{\mbox{$\Theta ^s_-$}}

183: \newcommand{\vs}{\mbox{$\varphi_s$}}

184: \newcommand{\vu}{\mbox{$\varphi_u$}}

185: \newcommand{\vg}{\mbox{$\varphi_G$}}

186: \newcommand{\vf}{\mbox{$\varphi_F$}}

187: \newcommand{\vif}{\mbox{$\varphi^{\infty}_F$}}

188: \newcommand{\vig}{\mbox{$\varphi^{\infty}_G$}}

189: \newcommand{\vil}{\mbox{$\varphi^{\infty}_L$}}

190: \newcommand{\vi}{\mbox{$\varphi^{\infty}$}}

191: \newcommand{\vr}{\mbox{$\varphi _s$}}

192: \newcommand{\vvv}{\mbox{${\cal V}$}}

193: \newcommand{\hlf}{\mbox{$X^L_F$}}

194: \newcommand{\hfl}{\mbox{$X^F_L$}}

195: \newcommand{\zf}{\mbox{$\zeta_F$}}

196: \newcommand{\zl}{\mbox{$\zeta_L$}}

197: \newcommand{\pin}{\mbox{$\partial _{\infty}$}}

198: \newcommand{\cc}{\mbox{$\cal C$}}

199: \newcommand{\lsf}{\mbox{${\cal L}^s_F$}}

200: \newcommand{\luf}{\mbox{${\cal L}^u_F$}}

201: \newcommand{\ga}{\mbox{$\gamma $}}

202: \newcommand{\oo}{\mbox{$\cal O$}}

203: \newcommand{\ooo}{\mbox{$\cal J$}}

204: \newcommand{\uo}{\mbox{${\cal J}^u$}}

205: \newcommand{\so}{\mbox{${\cal J}^s$}}

206: \newcommand{\ou}{\mbox{${\cal J}^u_+$}}

207: \newcommand{\rrr}{\mbox{$\times {\bf R}$}}

208: \newcommand{\rrrr}{\mbox{${\bf R}$}}

209: \newcommand{\rp}{\mbox{$[0,+\infty)$}}

210: \newcommand{\rr}{\mbox{$\times \{ 0 \}$}}

211: \newcommand{\pn}{\mbox{$\partial N$}}

212: \newcommand{\mi}{\mbox{$\widetilde M$}}

213: \newcommand{\wl}{\mbox{$\widetilde \lambda$}}

214: \newcommand{\wgs}{\mbox{$\widetilde \Lambda^s$}}

215: \newcommand{\ww}{\mbox{$\widetilde W$}}

216: \newcommand{\ti}{\mbox{$\partial \widetilde T$}}

217: \newcommand{\ttt}{\mbox{$\widetilde T$}}

218: \newcommand{\ppn}{\mbox{$\partial _+ N$}}

219: \newcommand{\lll}{\mbox{$l _{\epsilon}$}}

220: \newcommand{\iT}{\mbox{$i _{\widetilde T}$}}

221: \newcommand{\wwp}{\mbox{$\widetilde \Phi$}}

222: \newcommand{\w}{\mbox{$\widetilde {\partial _+ C}$}}

223: \newcommand{\pw}{\mbox{$\varphi _{\widetilde T}$}}

224: \newcommand{\lmun}{\mbox{$\Gamma ^s \cup \Gamma ^u$}}

225: \newcommand{\gc}{\mbox{${\cal G}$}}

226: \newcommand{\gsgg}{\mbox{${\cal G}^s_*$}}

227: \newcommand{\gl}{\mbox{${\cal G}$}}

228: \newcommand{\gugg}{\mbox{${\cal G}^u_*$}}

229: \newcommand{\gsg}{\mbox{${\cal G}^s$}}

230: \newcommand{\gug}{\mbox{${\cal G}^u$}}

231: \newcommand{\gs}{\mbox{$\Gamma ^s$}}

232: \newcommand{\gu}{\mbox{$\Gamma ^u$}}

233: \newcommand{\ggs}{\mbox{${\cal F}^s_T$}}

234: \newcommand{\ggu}{\mbox{${\cal F}^u_T$}}

235: \newcommand{\gsr}{\mbox{$\Gamma ^s \times {\bf R}$}}

236: \newcommand{\gur}{\mbox{$\Gamma ^u \times {\bf R}$}}

237: \newcommand{\se}{\mbox{$S _{\epsilon}$}}

238: \newcommand{\gus}{\mbox{$\Gamma ^u \downarrow S_{\epsilon}$}}

239: \newcommand{\vt}{\mbox{$\overline \varphi _{\widetilde T}$}}

240: %\newcommand{\vvvv}{\mbox{$\varphi _{\widetilde T}$}}

241: \newcommand{\hh}{\mbox{${\bf H}^2$}}

242: \newcommand{\hhh}{\mbox{${\bf H}^3$}}

243: \newcommand{\lmS}{\mbox{$\Gamma ^s \downarrow S$}}

244: \newcommand{\lmU}{\mbox{$\Gamma ^u \downarrow S$}}

245: \newcommand{\lms}{\mbox{$\Gamma ^s \downarrow S_{\epsilon}$}}

246: \newcommand{\lmu}{\mbox{$\Gamma ^u \downarrow S_{\epsilon}$}}

247: \newcommand{\wall}{\mbox{$L \times \R$}}

248: \newcommand{\wa}{\mbox{$\widetilde {\alpha}$}}

249: \newcommand{\wb}{\mbox{$\widetilde {\beta}$}}

250: \newcommand{\wdd}{\mbox{$\widetilde {\delta}$}}

251: \newcommand{\lap}{\mbox{$\Gamma ^s \cup \Gamma ^u \cup Q$}}

252: \newcommand{\tz}{\mbox{$\widetilde T \times \{ 0 \}$}}

253: \newcommand{\es}{\mbox{${\cal E}^s_F$}}

254: \newcommand{\eu}{\mbox{${\cal E}^u_F$}}

255: \newcommand{\wt}{\mbox{$\widetilde T$}}

256: \newcommand{\wss}{\mbox{$\widetilde S$}}

257: \newcommand{\rs}{\mbox{$\widetilde \rho(\widetilde S)$}}

258: \newcommand{\trs}{\mbox{$\Theta(\widetilde \rho(\widetilde S))$}}

259: \newcommand{\wn}{\mbox{$\widetilde N$}}

260: \newcommand{\si}{\mbox{$S^2_{\infty}$}}

261: \newcommand{\su}{\mbox{$S^1_{\infty}$}}

262: \newcommand{\sss}{\mbox{$S^s_{\infty}$}}

263: \newcommand{\sus}{\mbox{$S^u_{\infty}$}}

264: \newcommand{\s}{\mbox{$S ^1$}}

265: \newcommand{\fol}{\mbox{$\cal F$}}

266: \newcommand{\gal}{\mbox{$\cal G$}}

267: \newcommand{\gp}{\mbox{$\cal G_+$}}

268: \newcommand{\gm}{\mbox{$\cal G_-$}}

269: \newcommand{\gn}{\mbox{$\widetilde {\cal G}$}}

270: \newcommand{\gmn}{\mbox{$\widetilde {\cal G}_-$}}

271: \newcommand{\gpn}{\mbox{$\widetilde {\cal G}_+$}}

272: \newcommand{\fn}{\mbox{$\widetilde {\cal F}$}}

273: \newcommand{\ls}{\mbox{${\cal L} ^s$}}

274: \newcommand{\lss}{\mbox{${\cal L}^s_*$}}

275: \newcommand{\lu}{\mbox{${\cal L} ^u$}}

276: \newcommand{\lsd}{\mbox{${\cal L} ^s_d$}}

277: \newcommand{\lns}{\mbox{$\widetilde {\cal L} ^s$}}

278: \newcommand{\lnu}{\mbox{$\widetilde {\cal L} ^u$}}

279: \newcommand{\fs}{\mbox{${\cal F} ^s$}}

280: \newcommand{\fu}{\mbox{${\cal F} ^u$}}

281: \newcommand{\fss}{\mbox{${\cal F} ^{ss}$}}

282: \newcommand{\fuu}{\mbox{${\cal F} ^{uu}$}}

283: \newcommand{\fuo}{\mbox{${\cal F} ^{uu}_0$}}

284: \newcommand{\fnss}{\mbox{$\widetilde {\cal F} ^{ss}$}}

285: \newcommand{\fnuu}{\mbox{$\widetilde {\cal F} ^{uu}$}}

286: \newcommand{\fns}{\mbox{${\widetilde {\cal F}}^s $}}

287: \newcommand{\fnso}{\mbox{${\widetilde {\cal F}}^s_{\cal O} $}}

288: \newcommand{\fnuo}{\mbox{${\widetilde {\cal F}}^u_{\cal O} $}}

289: \newcommand{\fnsf}{\mbox{${\widetilde {\cal F}}^s_F $}}

290: \newcommand{\fnsg}{\mbox{${\widetilde {\cal F}}^s_G $}}

291: \newcommand{\fnuf}{\mbox{${\widetilde {\cal F}}^u_F $}}

292: \newcommand{\fnu}{\mbox{${\widetilde {\cal F}}^u $}}

293: \newcommand{\wv}{\mbox{${\overline {\varphi}}$}}

294: \newcommand{\oxi}{\mbox{${\overline {\xi}}$}}

295: \newcommand{\wvv}{\mbox{${\varphi}$}}

296: \newcommand{\fsS}{\mbox{${\cal F} ^s_S$}}

297: \newcommand{\fsG}{\mbox{${\cal F} ^s_G$}}

298: \newcommand{\fsE}{\mbox{${\cal F} ^s_E$}}

299: \newcommand{\fuG}{\mbox{${\cal F} ^u_G$}}

300: \newcommand{\fuS}{\mbox{${\cal F} ^u_S$}}

301: \newcommand{\fsT}{\mbox{${\cal F} ^s_T$}}

302: \newcommand{\fuT}{\mbox{${\cal F} ^u_T$}}

303: \newcommand{\fSS}{\mbox{$\widetilde{\cal F} ^s_{\cal S}$}}

304: \newcommand{\fUS}{\mbox{$\widetilde{\cal F} ^u_{\cal S}$}}

305: \newcommand{\fnuS}{\mbox{$\widetilde {{\cal F}}^u_S $}}

306: \newcommand{\fnsS}{\mbox{$\widetilde {{\cal F}}^s_S $}}

307: \newcommand{\ws}{\mbox{$\widetilde  W^s$}}

308: \newcommand{\wu}{\mbox{$\widetilde  W^u$}}

309: \newcommand{\hnt}{homotopically non trivial}

310: \newcommand{\up}{\mbox{${\cal U} _+$}}

311: \newcommand{\un}{\mbox{${\cal U} _-$}}

312: \newcommand{\Dp}{\mbox{$\Delta ^+$}}

313: \newcommand{\Dn}{\mbox{$\Delta ^-$}}

314: \newcommand{\fg}{\mbox{$\phi _g$}}

315: \newcommand{\h}{\mbox{$\bf H$}}

316: 

317:  

318: 

319: \def\@evenhead{\rm \leftmark \hfil \thepage}%        Left heading.

320: \def\chaptermark#1{\markboth {\sc {\ifnum \c@secnumdepth >\m@ne

321:       \@chapapp\ \thechapter. \ \fi #1}}{}}%

322: 

323: \catcode`\@=12

324: 

325: 

326: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

327: 

328: 

329: % Macros by Lee Mosher

330: %

331: %

332: %

333: % math symbols

334: %

335: \def\leftceiling{{\lceil}}

336: \def\rightceiling{{\rceil}}

337: \def\Reals{{\bf R}}

338: \def\reals{\Reals}

339: \def\Integers{{\bf Z}}

340: \def\integers{\Integers}

341: \def\rationals{{\bf Q}}

342: \def\Hyperbolic{{\bf H}}

343: \def\hyperbolic{\Hyperbolic}

344: \def\Euclidean{{\bf E}}

345: \def\euclidean{\Euclidean}

346: \def\Complex{{\bf C}}

347: \def\complex{\Complex}

348: \def\union{\cup}

349: \def\Union{\bigcup}

350: \def\intersect{\cap}

351: \def\intersection{\intersect}

352: \def\Intersection{\bigcap}

353: \def\inverse{{-1}}

354: \def\boundary{\partial}

355: \def\bd{\boundary}

356: \def\bdy{\boundary}

357: \def\suchthat{\bigm|}

358: \def\composed{\circ}

359: \def\comp{\composed}

360: \def\restrictedto{\bigm|}

361: \def\isomorphic{\approx}

362: \def\cross{\times}

363: \def\origin{{\bf O}}

364: \def\half{{1\over2}}

365: \def\inclusion{\hookrightarrow}

366: \def\infinity{\infty}

367: \def\concat{\ast}

368: \def\Sum{\sum}

369: \def\superset{\supset}

370: \def\from{\colon}

371: \def\homeo{\approx}

372: \def\disjunion{\coprod}

373: \def\tensor{\otimes}

374: 

375: %fix the next one

376: \def\directsum{{\bf +}}

377: 

378: %

379: %

380: % Math tokens of more than one letter

381: %

382: \def\func#1{\operatorname{#1}}

383: \def\Core{\func{Core}}

384: \def\Stable{\func{Stable}}

385: \def\Unstable{\func{Unstable}}

386: \def\cl{\func{cl}}

387: \def\closure{\cl}

388: \def\image{\func{image}}

389: \def\diam{\func{diam}}

390: \def\Isom{\func{Isom}}

391: \def\interior{\func{int}}

392: \def\Area{\func{Area}}

393: \def\area{\Area}

394: \def\Volume{\func{Vol}}

395: \def\volume{\Volume}

396: \def\Vol{\Volume}

397: \def\Length{\func{Length}}

398: \def\domain{\func{Domain}}

399: \def\range{\func{Range}}

400: \def\supp{\func{Supp}}

401: \def\source{\func{Srce}}

402: \def\target{\func{Targ}}

403: 

404: 

405: 

406: 

407: 

408: 

409: % Names with funny marks over the letters

410: \def\Mobius{M\"obius}

411: \def\Poincare{Poincar\'e}

412: \def\Teichmuller{Teichm\"uller}

413: 

414: 

415: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

416: 

417: 

418: 

419: 

420: 

421: %MosherMacros.sty}

422: 

423: \def\centeredepsfbox#1{\centerline{\epsfbox{#1}}}

424: 

425: \begin{document}

426: 

427: \title{Laminar free hyperbolic $3$-manifolds}

428: \author{S\'{e}rgio R. Fenley

429: \thanks{Reseach partially supported by NSF grant 

430: DMS 0296139.}

431: %\footnote{Version 1.0 Comments, suggestions, corrections

432: %are appreciated.}

433: }

434: \maketitle

435: 

436: \vskip .2in

437: 

438: \section{Introduction}

439: 

440: We analyse the existence question for essential

441: laminations in $3$-manifolds.

442: The purpose of the article is to prove that there

443: are infinitely many closed hyperbolic $3$-manifolds which

444: do not admit essential laminations. This

445: answers in the negative a fundamental question posed

446: by Gabai and Oertel when they introduced essential

447: laminations in \cite{Ga-Oe}, see also \cite{Ga4, Ga5}. 

448: The proof is obtained

449: by analysing certain group actions on trees and showing

450: that certain $3$-manifold groups only have trivial

451: actions on trees. There are corollaries concerning

452: the existence question for Reebless foliations and

453: pseudo-Anosov flows.

454: 

455: This article deals with the topological structure of

456: $3$-manifolds. Two dimensional manifolds are extremely

457: well behaved in the sense that the universal cover

458: is always either the plane or the sphere (for closed

459: manifolds), the fundamental group determines the manifold

460: and many other important properties. Similarly for

461: a $3$-manifold one asks: When is the universal cover

462: $\rrrr^3$? When does the fundamental group determine

463: the manifold?

464: Are homotopic homeomorphisms always isotopic? 

465: An obvious necessary condition is that the manifold

466: be {\em irreducible}, that is, every embedded sphere bounds

467: a ball. As for $2$-manifolds, the existence of a compact

468: codimension one object which is topologically good 

469: is extremely

470: useful. 

471: A properly embedded $2$-sided surface 

472: not ${\bf S}^2, {\bf D}^2$ is {\em incompressible}

473: if it injects in the fundamental group level \cite{He}.

474: A compact, irreducible  manifold with an incompressible

475: surface is called {\em Haken}. Fundamental work of Haken

476: \cite{Hak1,Hak2} and Waldhausen \cite{Wa} shows that

477: Haken manifolds have

478: fantastic properties, answering in the positive the

479: 3 questions above.

480: 

481: But how common are Haken $3$-manifolds, that is how

482: common are incompressible surfaces amongst irreducible

483: $3$-manifolds? In some sense they are not very common.

484: Recall that {\em Dehn surgery} along an orientation

485: preserving  simple closed curve $\delta$ is the process

486: of removing a tubular neighborhood $N(\delta)$ (a solid torus)

487: and glueing back by a homeomorphism of the boundary - which

488: is a two dimensional torus $T_1$ \cite{Rol,Bu-Zi}. 

489: The surgered manifold is completely determined

490: by which simple closed curve in $T_1$ becomes the new

491: {\em meridian}, that is, which curve of $T_1$ is glued to

492: the null homotopic curve in the boundary of $N(\delta)$.

493: Hence this is parametrized by a pair of relatively

494: prime integers $(q,p)$, corresponding to the description

495: of simple closed curves in $T_1$.  

496: When viewed this way, 

497: this set of relatively prime $(q,p)$ is the

498: {\em Dehn surgery space} $-$ a subset of

499: ${\bf Z}^2 \subset {\rrrr}^2$.

500: The same can be done iterating the process doing Dehn

501: surgery on links \cite{He,Rol,Bu-Zi}.

502: Notice that all closed, orientable $3$-manifolds can be obtained

503: from ${\bf S}^3$ by some Dehn surgery on an appropriate link

504: in ${\bf S}^3$ \cite{Rol}. So one can interpret

505: how common a property is by

506: verifying how many of the Dehn surgered manifolds have

507: that property. Along these lines some of the many

508: results on incompressible surfaces are:

509: If $K$ is a two bridge knot in ${\bf S}^3$ then almost all

510: Dehn surgeries on $K$ yield manifolds without incompressible

511: surfaces \cite{Ha-Th}. The same is true for any knot $K$ in

512: a manifold $M$ so that $M - K$ does not have any

513: closed incompressible surfaces \cite{Hat1}.

514: Notice that there are also results on the other direction:

515: for example Oertel \cite{Oe} proved that for many star links

516: in ${\bf S}^3$, 

517: then any non trivial Dehn surgery

518: yields a manifold with incompressible surfaces.

519: There are similar results for Montesinos knots \cite{Ha-Oe}.

520: Basically a lot of it depends on whether the complement

521: has closed incompressible surfaces or not.

522: In many cases the complement does not have

523: such surfaces, yielding the non existence results

524: for most Dehn surgered manifolds.

525: 

526: This amongst other reasons led to the concept of an

527: essential lamination as introduced by Gabai and Oertel

528: in the seminal paper \cite{Ga-Oe} of the late 80's. 

529: A {\em lamination} is a foliation of a closed subset

530: of the manifold. Roughly a lamination in a closed

531: $3$-manifold is {\em essential} if it has no sphere

532: leaves or tori leaves bounding solid tori, the complement

533: of the lamination is irreducible and the leaves in

534: the boundary of the complement are incompressible

535: and end incompressible in their respective complementary

536: components \cite{Ga-Oe}.

537: Gabai and Oertel proved the fundamental result that

538: essential laminations have deep consequences: 

539: the manifold $M$ is irreducible, its universal cover

540: is $\rrrr^3$, leaves of the lamination inject

541: in the fundamental group level, efficient closed 

542: transversals are not null homotopic; amongst

543: other consequences \cite{Ga-Ka3}. In addition

544: manifolds with {\em genuine} essential laminations

545: satify the weak hyperbolization conjecture \cite{Ga-Ka4}: either

546: there is a ${\bf Z} \oplus {\bf Z}$ subgroup of 

547: the fundamental group or the fundamental group

548: is Gromov hyperbolic \cite{Gr,Gh-Ha}.

549: Genuine means that not all complementary

550: regions are $I$-bundles, or equivalently it is not

551: just a blow up of a foliation.

552: Brittenham also proved properties concerning homotopy

553: equivalences for manifolds with essential laminations

554: \cite{Br2}.

555: 

556: In addition essential laminations are extremely common:

557: For example if $K$ is a non trivial knot in ${\bf S}^3$

558: then off of at most two lines and a couple of points

559: in Dehn surgery space,

560: the surgered manifold contains

561: an essential lamination.

562: This is obtained as follows: first Gabai

563: constructed a Reebless foliation $\fol$

564: in $({\bf S}^3 - N(K))$ which is transverse to the boundary

565: \cite{Ga1,Ga2,Ga3}.

566: Reebless means it does not have a Reeb component:

567: a foliation of the solid torus with the boundary being

568: a leaf, all other leaves are planes spiralling to the

569: boundary \cite{Re,No}.

570: Then results of Mosher, Gabai \cite{Mo2}

571: show that either there is

572: an incompressible torus transverse to $\fol$ or there is

573: an essential lamination in ${\bf S}^3 - N(K)$ with

574: solid torus complementary regions. This lamination remains

575: essential off of at most two lines in Dehn surgery space

576: \cite{Mo2} - see more on solid torus complementary regions 

577: later.

578: Also Brittenham produced examples of essential laminations

579: which remain essential after all non trivial Dehn surgeries

580: \cite{Br3,Br4}.  Roberts has also obtained many important

581: existence results concerning alternating knots in the

582: sphere 

583: \cite{Ro1,De-Ro} (partly jointly with

584: Delman) and punctured surface bundles \cite{Ro2,Ro3}.

585: 

586: 

587: So succesful was the search for essential laminations that

588: at first one might wonder whether all manifolds that

589: can (irreducible, with infinite fundamental group),

590: in fact do admit essential laminations. 

591: Given that an incompressible torus is an essential lamination,

592: the Geometrization conjecture \cite{Th2} suggests that one should

593: only have to analyse Seifert fibered spaces and hyperbolic

594: manifolds \cite{Sc,Th2}. 

595: 

596: The situation for Seifert fibered spaces has

597: been resolved: Brittenham produced examples of Seifert

598: fibered spaces  which are irreducible,

599: have infinite fundamental group, universal cover $\rrrr^3$,

600: but which do not have essential laminations \cite{Br1}.

601: Naimi \cite{Na}, using work of Bieri, Neumann and Strebel

602: \cite{BNS},

603: completely determined which Seifert fibered manifolds

604: admit essential laminations.

605: 

606: 

607: For hyperbolic $3$-manifolds there were two fundamental 

608: open questions: 1) (Thurston) Does every closed hyperbolic

609: $3$-manifold admit a Reebless foliation? 2) (Gabai-Oertel \cite{Ga-Oe},

610: see also \cite{Ga4,Ga5})

611: Does every closed hyperbolic $3$-manifold admit

612: an essential lamination?

613: In the past year question 1) was answered in the negative

614: by Roberts, Shareshian and Stein \cite{RSS} who produced

615: infinitely many counterexamples. The goal of this article

616: is to answer question 2) in the negative. We now proceed

617: to describe the examples.

618: 

619: Basically one starts with a torus bundle $M$ over the circle and

620: do Dehn surgery on a particular closed curve. Let

621: $\phi$ be the monodromy of the fibration associated to

622: a $2$ by $2$ integer matrix $A$, so that $A$ is hyperbolic.

623: Let $R$ be a fiber which is a torus.

624: There are two foliations in $R$ which are invariant 

625: under the monodromy $\phi$, the stable and unstable foliations.

626: The suspension flow in $M$ induces two foliations in $M$

627: with leaves being planes, annuli and M\"{o}ebius bands.

628: Suppose there is a M\"{o}ebius band leaf. 

629: Blow up that leaf, producing a lamination with

630: a solid torus complementary component with closure a solid

631: torus with core $\delta$ 

632: and with some curves $\eta$ removed from the boundary.

633: The curves $\eta$ are called the {\em degeneracy locus} of

634: the complementary region of the lamination \cite{Ga-Ka1}.

635: One can think of $\eta$ as lying in the boundary of $N(\delta)$,

636: which is a two dimensional torus.

637: Let $(1,0)$ be the curve in $\partial N(\delta)$ which bounds

638: the fiber in $M - N(\delta)$. Under an appropriate choice for

639: the curve $(0,1)$ of $\partial N(\delta)$ then $\eta$ is

640: represented by $(1,2)$. Do Dehn surgery along $\delta$.

641: If $\xi$ is the new meridian (the Dehn surgery slope),

642: then results of essential laminations \cite{Ga-Oe,Ga-Ka1}

643: show that $\lambda$ remains essential in the

644: Dehn surgery manifold $M_{\xi}$

645: if the intersection number of $\xi$ and $\eta$ is at least

646: $2$ in absolute value. If $\xi$ is described as $(q,p)$ then

647: this is equivalent to $|p - 2q| \geq 2$.

648: Therefore the open cases for essential laminations are

649: $|p - 2q| \leq 1$.

650: 

651: For simplicity of notation we omit the explicit dependence

652: of $M$ on $\phi$. It is always understood that $M$ depends

653: on the particular $\phi$.

654: 

655: In a beautiful and fundamental result, Hatcher \cite{Hat2}, showed

656: that if $p < q$ then then Dehn surgery

657: manifold $M_{\xi} = M_{p/q}$ has a Reebless foliation. This

658: is done via an explicit construction involving train tracks

659: and branched surfaces. In the last year Roberts, Shareshian

660: and Stein considered a particular type of monodromy,

661: namely generated by  the matrix

662: 

663: 

664: $$A \ \ = \ \

665: \left[

666: \begin{array}{rr}

667:  m & -1 \\ 1 & 0 \\ 

668: \end{array}

669: \right] \ \ \ \ m \leq -3$$

670: 

671: 

672: The eigenvalues of $A$ are negative. Consider the point

673: $(0,0)$ in $\rrrr^2$ and its projection $O$ to 

674: the fibering torus $R$. Let $\delta$ be the closed

675: orbit of the suspension flow through $O$. Because

676: the eigenvalues are negative, the leaf of 

677: the stable foliation through $O$ is a M\"{o}ebius

678: band. When it is blown open into an annulus the

679: degeneracy locus is $(1,2)$ as described above.

680: In a groundbreaking work, Roberts, Shareshian and

681: Stein \cite{RSS} considered Dehn surgery on these

682: manifolds and proved a wonderful result: 

683: if $p$ is odd, $m$ is odd and $p \geq q$ then

684: $M_{p/q}$ does not admit Reebless foliations.

685: In this article we consider a subclass of these

686: manifolds and show they do not admit essential 

687: laminations:

688: 

689: \vskip .2in

690: \noindent

691: {\bf Main Theorem:} \ Let $M$ be a torus bundle over the

692: circle with monodromy induced by the matrix $A$ above.

693: Let $\delta$ be the orbit of the suspension flow coming

694: from the origin and $M_{(q,p)} = M_{p/q}$ be the manifold obtained

695: by $(q,p)$ Dehn surgery on $\delta$. Here $(1,0)$ 

696: bounds the fiber in \ $M - N(\delta)$ \ and $(1,2)$ is

697: the degeneracy locus. 

698: Then if $m \leq -4$ and $|p - 2q| = 1$, the

699: manifold $M_{p/q}$ does not admit essential laminations.

700: \vskip .18in

701: 

702: The manifold $M - \delta$ is atoroidal \cite{Th4,Bl-Ca}

703: and fibers over the circle

704: with fiber a punctured torus.

705: By Thurston's hyperbolization

706: theorem in the fibering case $M - \delta$ has

707: a complete hyperbolic structure of finite volume \cite{Th3}.

708: By Thurston's Dehn surgery theorem $M_{p/q}$

709: is hyperbolic for almost all $p/q$ \cite{Th1}. Therefore:

710: 

711: \vskip .15in

712: \noindent

713: {\bf Corollary:} \ There are infinitely many closed, hyperbolic

714: $3$-manifolds which do not admit essential laminations.

715: \vskip .13in

716: 

717: We mention that Calegari and Dunfield \cite{Ca-Du}

718: approached the existence problem from a different point

719: of view. Following ideas

720: and results of Thurston \cite{Th5,Th6} concerning

721: the universal circle for foliations they showed

722: that 

723: a wide class of essential laminations also possess

724: a universal circle. 

725: One consequence is that tight essential laminations

726: with torus guts (see \cite{Ca-Du} for definitions)

727: have universal circles.

728: Hence the fundamental groups

729: act on the circle. Under certain conditions and if

730: the fundamental group is orderable then the 

731: action lifts to a non trivial action in $\rrrr$ and

732: they obtain nonexistence results for these types of laminations.

733: See more below.

734: 

735: 

736: Another immediate corollary is:

737: 

738: \vskip .15in

739: \noindent

740: {\bf Corollary:} \ If $m \leq -4$ and $|p-2q| = 1$, then

741: the manifolds $M_{p/q}$ above do not admit Reebless foliations.

742: \vskip .13in

743: 

744: About half of this result has already been established by

745: Roberts, Stein and Shareshian \cite{RSS}, namely the situation when

746: $m$ is odd. See more on $m$ odd below.

747: Another consequence is:

748: 

749: \vskip .2in

750: \noindent

751: {\bf Corollary:} \ If $m \leq -4$ and $|p-2q| = 1$ then

752: $M_{p/q}$ does not admit pseudo-Anosov flows.

753: \vskip .2in

754: 

755: For basic definitions and properties of pseudo-Anosov flows

756: consult \cite{Mo1,Mo2}. This result provides infinitely

757: many hyperbolic manifolds without pseudo-Anosov flows.

758: We stress that 

759: Calegari and Dunfield \cite{Ca-Du}

760: previously obtained conditions implying manifolds

761: do not admit pseudo-Anosov flows and showed for example that

762: the Weeks manifold does not admit pseudo-Anosov flows.

763: 

764: 

765: We remark that Dehn surgery on torus bundles over the circle

766: has been widely studied, for example: a) Which surgered 

767: manifolds have incompressible surfaces \cite{Fl-Ha,CJR},

768: b) Virtual homology \cite{Bk1,Bk2}, c) geometrization

769: \cite{Jo,Th1,Th2,Th3,Th4}.

770: 

771: 

772: \vskip .2in

773: We now describe the key ideas of the proof of the

774: main theorem.

775: The proof is done by looking at group actions on trees.

776: For simplicity first consider the case of a

777: Reebless foliation $\fol$. 

778: Novikov proved that leaves are incompressible

779: and transversals are never null homotopic \cite{No}.

780: Hence the lift to the universal

781: cover $\fn$ is a foliation by planes and its

782: leaf space is a simply connected $1$-dimensional

783: manifold, which may not be Hausdorff.

784: This $1$-manifold can be collapsed to a tree.

785: The fundamental group acts on this tree.

786: Roberts et al analysed group actions 

787: on simply connected non Hausdorff $1$-manifolds and also on trees

788: $-$ under the conditions

789: $p \geq q$ and $p, m$ odd, they ruled out the existence

790: of Reebless foliations \cite{RSS}.

791: Notice that the leaf space of the lifted foliation

792: $\fn$ is an orientable 

793: object and it makes sense to talk about orientation

794: preserving homeomorphisms. In order to stay in

795: the orientation preserving world they restricted

796: to $p, m$ odd.

797: 

798: 

799: 

800: Now consider an essential lamination $\lambda$.

801: The results of Gabai and Oertel \cite{Ga-Oe} 

802: imply that the lift to the universal cover

803: $\wl$ is a lamination by planes

804: in $\mi$.

805: To get the leaf space

806: blow down closures of complementary regions

807: to points and also non isolated leaves (on both

808: sides) to points. This produces an order 

809: tree as defined by Gabai-Kazez \cite{Ga-Ka2}

810: also called a non Hausdorff tree in this

811: situation \cite{Fe}.

812: A further appropriate collapsing of the (possible)

813: non Hausdorff points yields an actual tree

814: where the group acts non trivially.

815: The strategy is to show there are no nontrivial

816: actions of the group on trees.

817: An action is {\em trivial} if it has a global

818: fixed point.

819: A crucial difference from the case of foliations

820: is that in the case of laminations the

821: tree does not have an orientation in general.

822: Hence orientation dependent arguments cannot be used.

823: This was very important and widely used in \cite{RSS}.

824: Since we do not have an orientation here, the condition

825: $m$ odd does not play a role, which allows us

826: to consider $m$ even as well. 

827: Notice that the

828: condition $|p - 2q| =1$ obviously implies that

829: $p$ is odd. On the other hand there are many examples

830: with $p$ even so that $M_{p/q}$ has a Reebless foliation

831: - for example $p = 4, q = 1$ or

832:  $p = 8, q = 3$ (this has $p > q$!).

833: So to rule out Reebless foliations,

834: some further condition on $p, q$ should be necessary

835: when $p$ is even.

836: Except for ruling out trivial actions,

837: the proof here is done entirely in the tree $-$

838: we never go back to the original non Hausdorff tree.

839: For the sake of completeness we state this result

840: from which the main theorem is an easy corollary:

841: 

842: \vskip .2in

843: \noindent

844: {\bf {Theorem:}}  \ Let $M_{p/q}$ be the manifold

845: described above. If $m \leq -4$ and $|p - 2q| = 1$,

846: then every action of $\pi_1(M_{p/q})$ on a tree

847: is trivial.

848: \vskip .2in

849: 

850: 

851: The fundamental group of $M_{p/q}$ denoted by

852: $\gl$ can be generated by two elements

853: $\alpha$ and $\tau$. 

854: Actions of a homeomorphism on a tree are easy

855: to understand: either there is a fixed point

856: or in the free case there is an invariant axis.

857: An axis is a properly embedded copy of the reals

858: where the homeomorphism acts by translation.

859: The proof breaks down as to whether the generators

860: above act freely or not yielding 3 main cases

861: to consider (when $\tau$ acts freely it does

862: not matter the behavior of $\alpha$).

863: The proof subdivides into various subcases.

864: Invariably the analysis goes like this:

865: apply a certain relation in the group to a well

866: chosen point. One side of the relation

867: implies the image of the point is in a certain

868: region of the tree while the other side of the

869: relation implies it is in a different region - contradiction!

870: An important idea is that of a {\em local axis},

871: which has all the properties of axis except perhaps

872: being properly embedded. Homeomorphisms with fixed points

873: may have local axes. This is extremely useful

874: in a variety of cases.

875: 

876: 

877: 

878: We note that ${\bf Z}$ actions on non Hausdorff  trees

879: had been previously analysed in \cite{Fe} and \cite{Ro-St1,Ro-St2},

880: with consequences for pseudo-Anosov flows \cite{Fe}

881: and Seifert fibered spaces \cite{Ro-St1,Ro-St2}. 

882: 

883: 

884: There is a large literature of group actions

885: on trees which were brought to the

886: forefront by Serre's fundamental monograph 

887: \cite{Se}. The analyis usually involve a metric

888: which is  invariant under the actions

889: \cite{Mo-Sh1,Mo-Sh2,Mo-Sh3} or actions on simplicial

890: trees \cite{Se}.

891: We stress that the tree involved in here is 

892: not simplicial and it is not presented in

893: general with a group invariant metric $-$ unless

894: there is a holonomy invariant transverse measure

895: of full support in the lamination, e.g when

896: there is an incompressible surface.

897: So the proof is entirely topological and

898: in that sense elementary.

899: The topology of the manifold, particularly

900: the condition $|p-2q| = 1$ plays a crucial role.

901: Notice that in the foliations case there

902: is a pseudo-metric lying in the background which is

903: used from time to time in the proof by Roberts

904: et al \cite{RSS}. The pseudometric distance

905: between two points measures

906: how many jumps between non separated points

907: are necessary to go from one point to the other.

908: This pseudometric was analysed and used previously by

909: Barbot in \cite{Ba1,Ba2} with consequences for

910: foliations.

911: 

912: 

913: The results of this article mean that the search for

914: structures more general than essential laminations,

915: but still useful takes an added relevance.

916: One idea previously proposed by Gabai \cite{Ga5} is

917: that of a {\em loosesse lamination}. We will have more comments

918: on that in the final remarks section.

919: 

920: We are very thankful to Rachel Roberts who introduced

921: the idea of considering group actions in

922: the foliations case and other ideas.

923: 

924: 

925: 

926: 

927: 

928: 

929: 

930: 

931: 

932: \section{The group}

933: \label{grou}

934: 

935: Here we compute the fundamental group of $M_{p/q}$.

936: Start with $M$ the torus bundle over the circle

937: with monodromy induced by 

938: 

939: $$A \ \ = \ \ \left[ \begin{array}{rr} 

940: m & -1 \\ 1 & 0 \end{array} \right] \ \ \ \ \ 

941: {\rm where} \ \ m \leq -3$$

942: 

943: 

944: For notational simplicity the dependence of $M$ on $A$ is omitted.

945: 

946: 

947: 

948: 

949: The eigenvalues of $A$ are 

950: 

951: $$\frac{m \pm \sqrt{m^2 - 4}}{4}$$

952: 

953: \noindent

954: which are both negative  and the matrix is

955: hyperbolic. The eigenvector directions produce two

956: linear 

957: foliations in $\rrrr^2$ with irrational slope

958: and invariant under $A$. They induce two foliations

959: in the torus $T^2$. Since $A$ is integral it induces

960: a homeomorphism $\phi$ of $T^2$, which leaves

961: the foliations invariant. Let $O$ in $T^2$ be the

962: image of the origin. Let $M$ be the suspension

963: of $\phi$ and let $\fol$ be (say) the suspension

964: of the stable foliation of $T^2$. Then $\fol$ has

965: leaves which are planes, annuli and M\"{o}ebius bands.

966: Identify $T^2$ with a fiber in $M$ and let $\delta$

967: be the orbit through $O$, which is a closed orbit

968: intersecting $T^2$ once.

969: Since the eigenvalues of $A$ are negative, the stable

970: leaf containing $\delta$ is a M\"{o}ebius band. 

971: We do Dehn surgery on $\delta$. 

972: We first determine the fundamental group of $M - N(\delta)$.

973: To do that 

974: let 

975: 

976: $$D \ = \ N(\delta) \cap T^2 \ \ {\rm (a \ disk)}, \ \ \ 

977: V \ = \ T^2 - D \ \ {\rm (a \ punctured \ torus)}.$$

978: 

979: 

980: \noindent

981: Choose a basis for the homology of $\partial N(\delta) 

982: = T_1$, a torus. 

983: Let $(1,0)$ be the curve in $T_1$ bounding the fiber $V$ of

984: $M - N(\delta)$.

985: Blow up the leaf of $\fol$ through $\delta$. It blows

986: to a single annulus and the complementary region

987: is a solid torus with core $\delta$. 

988: The closure of the complementary region 

989: is a solid torus with a  closed curve in the boundary removed.

990: The removed curve is the degeneracy locus of the

991: complementary component \cite{Ga-Ka1}.

992: Since the leaf of $\fol$ was a M\"{o}ebius band,

993: the degeneracy locus intersects the curve

994: $(1,0)$ twice. Choose the curve $(0,1)$ so that

995: the degeneracy locus is the curve $(1,2)$ in this basis.

996: Let $M_{p/q}$ be the manifold obtained from

997: $M$ by doing $(q,p)$ Dehn surgery on $\delta$.

998: By results about essential laminations, the lamination

999: $\lambda$ remains essential in $M_{p/q}$ if

1000: $|p - 2q| \geq 2$.

1001: Let $\gamma$ be the curve $(0,1)$ in $T_1$ and

1002: $\tau$ be the curve $(1,0)$.

1003: The degeneracy locus is the curve $\gamma \tau^2$.

1004: Notice there are two tori here: one

1005: which is a fiber of the original fibration (here denoted

1006: by $T^2$), another which

1007: is the boundary of $N(\delta)$ (here denoted by $T_1$).

1008: The Dehn surgery coefficients refer to $T_1$.

1009: 

1010: %We first compute the fundamental group of $M - N(\delta)$.

1011: Suppose the disk $D$ above

1012: is a round disk of radius $\epsilon$ sufficiently

1013: small.

1014: The universal abelian cover of $T^2 - D$ is the plane

1015: with disks of radius $\epsilon$ around integer lattice

1016: points removed. Let $E$ be the one around the origin.

1017: We pick $4$ points in $\partial E$:

1018: $a = (-\epsilon, 0),

1019: b = (0, - \epsilon),

1020: c = (\epsilon, 0)$ and $d = (0, \epsilon)$,

1021: see fig. \ref{01}, a.

1022: Let $a'$ be the image of $a$ under $A$, etc.., see fig.

1023: \ref{01}, b.

1024: 

1025: The image of $\partial E$ under $A$ is an ellipse

1026: which can be deformed back to $\partial E$, see fig.

1027: \ref{01}, b.

1028: Notice $b', d'$ are in the $x$ axis

1029: and $d' = a$.

1030: 

1031: Let the image of $a$ in $T^2 - D$ be the basepoint

1032: of the fundamental group of $M - N(\delta)$

1033: for simplicity still denoted by $a$ and likewise

1034: for $b, c, d$.

1035: Let $l$ be an arc along the image of $\partial E$

1036: under $A$, going counterclockwise from $d'$ to $a'$.

1037: 

1038: %\blankfig{01}{1.}{The fundamental group}

1039: 

1040: 

1041: 

1042: \begin{figure}

1043: \centeredepsfbox{lam1.eps}

1044: \caption{

1045: Computing the fundamental group of $M - N(\delta)$.}

1046: \label{01}

1047: \end{figure}

1048: 

1049: 

1050: 

1051: 

1052: 

1053: 

1054: 

1055: 

1056: 

1057: We pick a basis for $\pi_1(T^2 - D)$:

1058: Let \ $\alpha  \ =  \ \overline{ac} * l_1$ \ (see fig. \ref{01}, c)

1059: where the arc $\overline{ac} \subset \partial E$

1060: is traversed in the counteclockwise direction

1061: and $l_1$ is parametrized as \ $\{ (t,0) \ | \

1062: \epsilon \leq t \leq 1 - \epsilon\}$.

1063: Here $*$ denotes concatenation of arcs.

1064: Let also 

1065: 

1066: $$\beta \ = \ \overline{ad}_{clo} * l_2 * \overline{ba}_{clo},$$

1067: 

1068: \noindent

1069: where $l_2$ is parametrized as \ $\{ (0,t) \ | \

1070: \epsilon \leq t \leq 1 - \epsilon \}$,

1071: and the ``clo" subscript means the arcs are traversed

1072: clockwise in $\partial E$.

1073: We identify $\alpha$ and $\beta$ with their images

1074: in $T_2 - D$, so they generate the fundamental group

1075: of $T_2 - D$.

1076: It is easy to see that 

1077: the curve $\gamma = [\alpha,\beta]$ is just 

1078: an counterclockwise turn around $\partial E$.

1079: Then 

1080: 

1081: $$ \tau^{-1} \alpha \tau \ \ = \ \ l * \overline{a'c'} * l'_1 

1082: * l^{-1}.$$

1083: 

1084: \noindent

1085: The composition $l * \overline{a'c'}$ is roughly one 

1086: counterclockwise turn around

1087: $\partial E$ so it is the curve $\gamma$.

1088: The straight arc $l'_1$ goes from

1089: $c' = (m \epsilon, \epsilon)$ to

1090: $(m(1-\epsilon), 1 - \epsilon)$ - roughly going 

1091: one step up and $|m|$ steps to the left.

1092: This together with $l^{-1}$ can be isotoped to

1093: $\beta \alpha^m$  \ (where we are identifying $\alpha,

1094: \beta$ with the appropriate covering translates).

1095: We conclude that $\tau^{-1} \alpha \tau = \gamma \beta \alpha^m$.

1096: Similarly 

1097: 

1098: $$\tau^{-1} \beta \tau \ \ = \ \ l *  \overline{a'd'}_{clo} *  l'_2 

1099: *  \overline{b'a'}_{clo} *  l^{-1}$$

1100: 

1101: \noindent

1102: So in the same way it is easy to see that 

1103: $\tau^{-1} \beta \tau = \alpha^{-1}$.

1104: Notice that $\alpha, \tau$ generate $\pi_1(M - N(\delta))$.

1105: Hence

1106: 

1107: 

1108: $$\pi(M - N(\delta)) \ \ = \ \ 

1109: \{ \alpha, \tau \ \ | \ \ \tau^{-1} \alpha \tau = \gamma \beta \alpha^m,

1110: \ \ \tau^{-1} \beta \tau = \alpha^{-1}, \ \gamma = [\alpha, \beta] \ \}$$

1111: 

1112: \noindent

1113: After $(q,p)$ Dehn surgery on $\delta$ we obtain

1114: $q \gamma + p \tau$ is the new meridian or

1115: $\tau^p \gamma^q = 1$. Hence

1116: 

1117: 

1118: $$\gl \ \ = \ \ \pi_1(M_{p/q}) \ \ =   \ \ 

1119: \{ \alpha, \tau \ | \ \tau^{-1} \alpha \tau = \gamma \beta \alpha^m,

1120: \ \ \tau^{-1} \beta \tau = \alpha^{-1}, \ \gamma = [\alpha, \beta],

1121: \ \tau^p \gamma^q = 1 \}$$

1122: 

1123: 

1124: 

1125: 

1126: In the proof we will use these and

1127: the following variations of these

1128: relations extensively:

1129: 

1130: $$\tau^{-1} \beta \tau = \alpha^{-1}, \ \ \tau \alpha \tau^{-1} = \beta,$$

1131: 

1132: $$ \tau^{-1} \alpha \tau \ = \ 

1133: \gamma \beta \alpha^m = \alpha \beta \alpha^{m-1}

1134: \ = \ \alpha \tau \alpha^{-1} \tau^{-1} \alpha^{m-1}$$

1135: 

1136: 

1137: $$

1138: \alpha \tau  = \tau \gamma \beta \alpha^m = \tau \alpha \beta \alpha^{m-1}$$

1139: 

1140: $$\alpha \beta = \gamma \beta \alpha, 

1141: \ \ {\rm or} \ \ \alpha \tau \alpha^{-1} \tau^{-1}= \tau \gamma 

1142: \alpha^{-1} \tau^{-1} \alpha$$

1143: 

1144: \noindent

1145: A little manipulation with the relations also yields

1146: 

1147: $$ \tau \beta \tau^{-1} \ = \ \beta \alpha \beta^{m-1}

1148: \ = \ \gamma^{-1} \alpha \beta^m$$

1149: 

1150: These and circular variations of these will be used throughtout

1151: the article.

1152: 

1153: Since $q, p$ are relatively prime there are $e,f$ in ${\bf Z}$ with

1154: $ep + fq = 1$. Let $\kappa = \tau ^f \gamma^{-e}$.

1155: Then $\kappa$ is a generator of the ${\bf Z}$ subgroup

1156: of $\gl$ generated by $\tau, \gamma$ and

1157: $\tau = \kappa^q, \gamma = \kappa^{-p}$.

1158: 

1159: 

1160: \vskip .1in

1161: \noindent

1162: {\bf {NOTATION:}} $-$ In the arguments group elements act

1163: on the {\underline {right}}.

1164: \vskip .1in

1165: 

1166: \section{Outline of the proof}

1167: 

1168: Given the presentation of $\gl$ above the proof

1169: of the main theorem

1170: is broken into 4 cases:

1171: 

1172: \begin{itemize}

1173: 

1174: \item 

1175: Case R - $\rrrr$-covered case

1176: 

1177: \item 

1178: Case A - $\tau$ acts freely

1179: 

1180: \item 

1181: Case B - $\alpha$ acts freely $\tau$ has a fixed point.

1182: 

1183: \item 

1184: Case C - $\alpha$ and $\tau$ have fixed points.

1185: 

1186: \end{itemize}

1187: 

1188: If $\mu$ acts freely on a tree, let $\aaa_{\mu}$ be

1189: its axis.

1190: If $\mu$ has a local axis, we denote it by ${\cal L A} _{\mu}$.

1191: Unlike a true axis, a homeomorphism may have more than

1192: one local axis.

1193: The context will make it clear which one is being considered.

1194: 

1195: 

1196: \vskip .15in

1197: \noindent

1198: {\underline {Case R}}  $-$ $\rrrr$ covered case.

1199: 

1200: The $\rrrr$-covered case is simple.

1201: Given that $p$ is odd, this implies that

1202: $\tau$ is orientation preserving in $\rrrr$.

1203: The case $\alpha$ orientation preserving 

1204: is simple. The other case (which implies $m$ is even)

1205: leads to $p > 3q$ which for our purposes is

1206: enough. It also leads us to move away from

1207: orientation preserving arguments, which is more

1208: like the laminations case.

1209: We note that there is an easy linear non trivial

1210: action on $\rrrr$ when $p = 4, q = 1$. Notice that

1211: in this case $p$ is even.

1212: 

1213: \vskip .15in

1214: \noindent

1215: {\underline {Case A}} $-$ $\tau$ acts freely.

1216: 

1217: This implies that $\kappa$ also acts freely

1218: and $\aaa_{\kappa} = \aaa_{\tau}$.

1219: We analyse how $\aaa_{\kappa}$ intersects

1220: $\ak \alpha$ and other translates (here $\ak \alpha$

1221: is the image of $\ak$ under $\alpha$).

1222: Let $u = \alpha \beta$. One uses the

1223: relation $\alpha \beta = \gamma \beta \alpha$ to

1224: analyse how $\ak$ intersects $\ak u$ which

1225: breaks down into various cases as to whether

1226: this intersection is empty, a single point

1227: or a segment.

1228: One particularly tricky case needs the 

1229: condition $m \not = -3$.

1230: 

1231: \vskip .15in

1232: \noindent

1233: {\underline {Case B}} $-$ $\alpha$ acts freely, $\tau$ has

1234: a fixed point.

1235: 

1236: Let $z$ be a fixed point of $\tau$.

1237: First suppose that $z$ is not in the axis 

1238: $\ap$ 

1239: of $\alpha$.

1240: Suppose there is no fixed point of

1241: $\tau$ between $z$ and $\ap$.

1242: Here let $\uu$ be the component

1243: of $T - \{ z \}$ containing $\ap$.

1244: The case $\uu \tau \not = \uu$ is easy to

1245: deal with.

1246: It follows that $\uu \tau = \uu$

1247: producing a local axis $\lat$ of $\tau$ which

1248: is contained in $\uu$ and has one limit point

1249: in $z$. The proof breaks down as to whether

1250: $\lat$ intersects $\ap$ or not.

1251: Empty intersections are easy to deal with,

1252: the other case being trickier.

1253: 

1254: Then suppose $z$ is in $\ap$.

1255: We remark this is a crucial case, because this

1256: is likely what happens for the essential laminations

1257: we know to exist when $|p - 2q| \geq 2$.

1258: These come from the original stable lamination

1259: on the fibering manifold. In that manifold,

1260: $\alpha$ acted freely and $\tau$ had a fixed

1261: point in $\ap$. After the surgery $\alpha$ would

1262: still have at least a local axis, which contains

1263: a fixed point of $\tau$. So one knows the exact

1264: condition $|p - 2q| =  1$ will have to be used here!

1265: 

1266: In this case consider $\uu_1$ be the component

1267: of $T - \{ z \}$ containing $z \alpha$ and

1268: $\uu_2$ the one containing $z \alpha^{-1}$.

1269: It is easy to show that $\uu_1 \tau$ is not

1270: $\uu_1$ and that $\uu_1 \tau$ is in fact

1271: equal to $\uu_2$.

1272: When $\uu_1 \tau^{-1} = \uu_2$ then one

1273: produces a contradiction just using that

1274: $p$ is odd.

1275: The case $\uu_1 \tau^{-1} \not = \uu_2$ or

1276: $\uu_2 \tau \not = \uu_1$ is much more

1277: interesting. Here the  exact condition

1278: $|p-2q| = 1$ is used to show it would imply

1279: $\uu_1 \tau = \uu_1$ which was disallowed

1280: at the beginning.

1281: This actually has connections with the topology

1282: of the situation, see detailed explanation

1283: in section \ref{caseb}.

1284: This is a crucial part of the proof.

1285: One very tricky issue is that a priori

1286: $z$ is only a fixed point of $\tau$ and not

1287: of $\gamma$ $-$ part of the proof is ruling this out.

1288: 

1289: 

1290: \vskip .15in

1291: \noindent

1292: {\underline {Case C}} \ 

1293: 

1294: %This is by far the most technical case of the proof.

1295: Generally an axis is good because it gives

1296: information about where points go. The

1297: case of fixed points is trickier

1298: and one many times searches for local axis.

1299: 

1300: Given two points $a, b$ in a tree let

1301: $[a,b]$ be the unique embedded segment

1302: connecting them. Let $(a,b) = [a,b] - \{ a, b \}$.

1303: Notice $(a,b)$ is exactly the set of points in the

1304: tree separating $a$ from $b$.

1305: 

1306: Here let $s$ be a fixed point of $\kappa$ and

1307: $w$ a fixed point of $\alpha$ so that there

1308: is no fixed point of either in $(s,w)$.

1309: Notice there may be fixed points of

1310: $\tau$ in $(s,w)$!

1311: Let $\we$ be the component of $T - \{ s \}$ containing

1312: $w$ and $\vv$ the component of $T - \{ w \}$ containing

1313: $s$.

1314: The first part of the proof shows that $\we \tau = \we$ and

1315: $\vv \alpha = \vv$.

1316: These are moderately involved cases.

1317: This immediately produces a local axis 

1318: $\lal$ of $\alpha$ contained in $\vv$ and with

1319: one ideal point $w$.

1320: One does not have yet a local axis for $\tau$ 

1321: because we do not know a priori that $\tau$ has

1322: no fixed points in $(s,w)$.

1323: Some technical complications ensue.

1324: 

1325: One then shows that 

1326: $s \alpha, s \alpha^{-1}$ are in $\we$.

1327: Let $z$ be the fixed point of $\tau$ in $[s,w)$ which

1328: is closest to $w$ $-$ $z$ could be $s$.

1329: Using the previous results show that

1330: the component $\uu$ of $T - \{ z \}$ containing

1331: $w$ is invariant under $\tau$.

1332: Now this produces a local axis $\lat$ of $\tau$ 

1333: in $\uu$ with ideal point $z$ and some further

1334: properties.

1335: One then shows that $w$ is not 

1336: in $\lat$ and $z$ not in $\lal$.

1337: 

1338: We are now in familiar ground.

1339: If $\lal \cap \lat$ has at most one

1340: point, then it is easy.

1341: When

1342: $\lal \cap \lat$ has more than one point

1343: we use arguments done in case B $-$ 

1344: this part of the arguments in case B is done

1345: in more generality using local axis (rather than

1346: axis as needed in case B) and can be used

1347: in case C as well. This finishes the proof of case C.

1348: %ping pong lemma ideas

1349: %yield the proof easily.

1350: %There are directions in $\lal, \lat$ 

1351: %corresponding to moving away from $w, s$

1352: %respectively.

1353: %The intersection

1354: %$\lal \cap \lat$ is a segment

1355: %$[w,u]$, $w$ closest to $w$. Consider

1356: %whether the directions

1357: %of $\lat, \lal$ in the intersection

1358: %agree or not.

1359: %In each of these two cases the proof breaks down

1360: %into 4 subcases depending on which

1361: %of $w \alpha, w  \alpha^{-1}$

1362: %is in $[w,w]$ and which of

1363: %$s \tau, s \tau^{-1}$ is in $[s,w]$.

1364: %In each case two of the 4 subcases 

1365: %are very simple to discard, the other two

1366: %very complicated - not the same subcases in the

1367: %two cases.

1368: This finally yields the proof of the main theorem.

1369: 

1370: 

1371: The arguments in this article are very involved.

1372: One possibility to read the article and get

1373: a quick grasp of the proof is to first analyse

1374: the $\rrrr$-covered proof. Then go to the proof

1375: of case B.2 - $\alpha$ acts freely and $\tau$ has

1376:  a fixed point in the axis of $\alpha$ $-$ this

1377: case admits essential laminations if $|p - 2q| \geq 2$

1378: and the topology can be detected. Then read the

1379: proof of $\tau$ acts freely and the other proofs.

1380: 

1381: \section{Preliminaries}

1382: 

1383: 

1384: 

1385: Let $\lambda$ be an essential lamination on

1386: a $3$-manifold $N$. We'll modify $\lambda$ if

1387: necessary to eventually obtain a group action

1388: on a tree which is essentially the leaf space

1389: of the lifted lamination $\widetilde \lambda$

1390: to the universal cover $\widetilde N$.

1391: First if there are any leaves

1392: of $\lambda$ which are isolated on both sides,

1393: then blow each of them into an $I$-bundle of

1394: leaves $-$ needs to be done at most countably

1395: many times. Now 

1396: $\widetilde \lambda$ is a lamination by planes

1397: with no leaves isolated on both sides \cite{Ga-Oe}.

1398: 

1399: Suppose $L$ is a leaf of $\wl$

1400: which is non separated from another leaf $F$

1401: $-$ that is, there are $L_i$ leaves of $\wl$

1402: with $L_i$ converging to both $L$ and $F$.

1403: We do not want that $L$ is not separated from

1404: some other leaf in the other side (the one

1405: not containing $F$). If that happens, blow up

1406: $L$ into an $I$-bundle of leaves.

1407: This can also be achieved by a blow up in $\lambda$.

1408: Since there are at most countably many leaves

1409: non separated from some other leaf we can get

1410: rid of leaves non separated from leaves on

1411: both sides. If needed use blow ups so that

1412: non separated

1413: leaves of $\wl$ are not boundary leaves

1414: of a complementary region of $\wl$ (on the opposite side).

1415: After all these possible modifications assume

1416: this is the original lamination $\lambda$.

1417: 

1418: Now define a set $T_*$ whose elements are:

1419: closures of complementary components of $\wl$ and

1420: also leaves of $\wl$ which are non isolated on

1421: both sides. Then $T_*$ is an {\em order tree}

1422: \cite{Ga-Ka2,Ro-St2}, also called non

1423: Hausdorff tree \cite{Fe}.

1424: The fundamental group $\pi_1(N)$ naturally acts

1425: on $T_*$. If $e$ is any point of $T_*$ which is

1426: non separated another point $e'$, collapse all

1427: points non separated from $e$ together with

1428: $e$. This is OK since no such $e$ is

1429: non separated on more than one side and $e$ also

1430: does not come from a complementary region of $\wl$.

1431: The collapsed object is now an actual tree $T$ and

1432: the action of $\pi_1(N)$ on $T_*$ induces

1433: a natural action of $\pi_1(N)$ on $T$.

1434: In our proof $N = M_{p/q}$ and we will analyse

1435: group actions of $\gl = \pi_1(M_{p/q})$ on the tree $T$.

1436: 

1437: 

1438: \begin{define}{}

1439: A group action on a tree $T$ is nontrivial if no

1440: point of $T$ is fixed by all elements of the group.

1441: \end{define}

1442: 

1443: A lot of results on group actions on trees are

1444: to rule out non trivial group actions \cite{Cu-Vo}.

1445: 

1446: Given point $a, b$ on a tree $T$ let

1447: 

1448: $$(a,b) \ = \ \{ c \in T \ \ | \ \ c \ \ {\rm separates} \ \

1449: a \ \ {\rm from} \ \ b \}.$$

1450: 

1451: \noindent

1452: If $a = b$, then $(a,b)$ is empty, otherwise it is

1453: an open segment.

1454: Let $[a,b]$ be the union of $(a,b)$ and $\{ a, b \}$.

1455: Then $[a,b]$ is always a closed segment.

1456: 

1457: One fundamental concept here is the following:

1458: 

1459: \begin{define}{(bridge)}

1460: If $x$ is a point of a tree $T$ not contained in a 

1461: connected set $B$, then there is a unique 

1462: embedded path $[x,y]$ from $x$ to $B$.

1463: This path has $(x,y) \cap B = \emptyset$ and

1464: either  $y$ is in $B$ or $y$ is an accumulation

1465: point of $B$. We say that $[x,y]$ is the bridge

1466: from $x$ to $B$ and if $y$ is in $B$ we say that

1467: $x$ bridges to $B$ in $y$ or that $x$ bridges

1468: to $y$ in $B$.

1469: \end{define}

1470: 

1471: For example if $T$ is the reals and $B = (0,1)$, $x = 2$, then

1472: the bridge from $x$ to $B$ is $[2,1]$. 

1473: One common use of bridges will be: if $x$ is not 

1474: in a properly embedded line $l$ (as an axis defined below)

1475: let $[x,y]$ be the bridge

1476: from $x$ to $l$.

1477: The crucial property of the bridge is that given $x$ and

1478: $B$, the bridge is {\underline {unique}}.

1479: In various situations this will force some useful 

1480: equalities of points.

1481: Another fundamental concept is:

1482: 

1483: \begin{define}{(axis)}

1484: Suppose that $g$ is a homeomorphism acting freely

1485: on a tree $T$. Then $g$ has an axis $\aaa_g$, a properly

1486: embedded line in $T$, invariant under $g$ and $g$ acts

1487: by translations on $\aaa_g$.

1488: \end{define}

1489: 

1490: This is classical. Here $y$ is in $\aaa_g$

1491: if and only if $yg$ is in $(y,yg^2)$, that is

1492: $yg$ separates $y$ from $yg^2$.

1493: Then it is easy to see that the axis must be

1494: the union of $[y g^i, y g^{i+1}]$ where $i \in {\bf Z}$

1495: \cite{Ba1,Fe}.

1496: To obtain an element in $\aaa_g$ consider any $x \in T$.

1497: If $x g \in (x, x g^2)$ done. Else 

1498: there is a unique

1499: 

1500: $$y \ \in \ \ [x,xg] \cap [x, x g^2] \cap [x g, x g^2].$$

1501: 

1502: \noindent

1503: $y$ is the basis of the tripod with corners $x, xg, xg^2$

1504: \cite{Gr,Gh-Ha}.

1505: A simple analysis of cases using free action yields

1506: $y$ is in the axis.

1507: 

1508: Another simple but fundamental concept for us is:

1509: 

1510: \begin{define}{(local axis)}

1511: Suppose $l$ is a line in  a tree $T$ where a homeomorphism

1512: $g$ acts by translation. Then $l$ is a local axis for $g$

1513: and is denoted by $\la_g$. The local axis may not be unique,

1514: the context specifies which one we refer to.

1515: \end{define}

1516: 

1517: For example if $g$ acts in $\rrrr$ by 

1518: $x g = 2x$, then $\rrrr_+, \rrrr_-$ are

1519: both local axes of $g$ with accumulation

1520: point $x = 0$.

1521: Another characterization of local axis: $x$ is in a 

1522: local axis of $g$ if and only if $x g$ separates

1523: $x$ from $x g^2$ (same definition as for axis

1524: except requiring that $g$ acts freely).

1525: Another characterization: suppose $x g$ is not $x$ 

1526: and let $\uu$ be  the component of

1527: $T - \{ x \}$ containing $x g$. Then $x$

1528: is in a local axis of $g$ if and only

1529: if $\uu g \subset \uu$.

1530: %For these equivalencies see \cite{Fe2}.

1531: 

1532: 

1533: 

1534: 

1535: Let $x$ be a point in a tree $T$. A {\em prong} at

1536: $x$ is a non degenerate segment $I$ of $T$ so that $x$ is

1537: one of the endpoints of $I$.

1538: Two prongs at $x$ are equivalent if they

1539: share a subprong at $x$. Associated to a subprong $I$

1540: at $x$ there is a unique component $\uu$ of $T - \{ x \}$

1541: containing $I - \{ x \}$.

1542: 

1543: \vskip .1in

1544: \noindent

1545: {\bf {Notation}} $-$ If $x, y, z$ are elements

1546: in  a tree we will write $x \prec y \prec z$ if

1547: $y$ separates $x$ from $z$,

1548: or $y$ is in $(x,z)$.

1549: We say that $x, y, z$ (in this order)

1550: are {\em aligned}.

1551: Also $x \prec y \preceq z$ if one also allows

1552: $y = z$ and so on.

1553: Notice that this

1554: is invariant under homeomorphims of the tree.

1555: 

1556: 

1557: The following simple results will be very useful:

1558: 

1559: \begin{lemma}{}{}

1560: Let $x$ be a point in a tree $T$. Then two

1561: prongs $I_1, I_2$ at $x$ are equivalent

1562: if and only if the associated complementary

1563: components $\uu_1, \uu_2$ are the same.

1564: \label{inva}

1565: \end{lemma}

1566: 

1567: \begin{proof}{}

1568: If $I_1, I_2$ are equivalent, there is $y$ in $I_1 - \{ x \}$

1569: also in $I_2$. Then clearly $y \in \uu_1$ and $y \in \uu_2$,

1570: so $\uu_1 = \uu_2$.

1571: Conversely suppose $\uu_1 = \uu_2$.

1572: If $I_1$ is not equivalent to $I_2$, then $I_1 \cap I_2 = \{ x \}$

1573: because $T$ is a tree and it also follows that $x$ separates

1574: $I_1$ from $I_2$. This would imply $\uu_1$, $\uu_2$ disjoint,

1575: contradiction.

1576: \end{proof}

1577: 

1578: \begin{lemma}{}{}

1579: Let $T$ be a tree and $\eta$ a homeomorphism so that

1580: there are two points $x, y$ of $T$ so that

1581: $x \prec x \eta \prec y \prec y \eta$ \ \ 

1582: or \ \ $x \prec y \prec x \eta \prec y \eta$.

1583: Then $x$ and $y$ are in a local axis of $\eta$.

1584: \label{chax}

1585: \end{lemma}

1586: 

1587: \begin{proof}{}

1588: We do the proof for the first situation, the other

1589: being very similar. Let $\uu$ be the component of $T - \{ 

1590: x \}$ containing $x \eta$. Using $x \prec x \eta \prec y$

1591: this is also the component of $T - \{ x \}$ containing

1592: $y$. Apply $\eta$, then $\uu$

1593: is taken to the component

1594: of $T - \{ x \eta \}$ containing $y \eta$. Then

1595: $\uu \eta$ is contained in $\uu$ and $x$ is in

1596: a local axis. Apply $\eta^{-1}$ to $y$ to get 

1597: $y$ is in a local axis as well.

1598: We stress the two local axes produced in

1599: this way a priori may not be the same: there may

1600: be a fixed point of $\eta$ in $(x,y)$.

1601: \end{proof}

1602: 

1603: 

1604: 

1605: 

1606: 

1607: 

1608: \vskip .1in

1609: \noindent

1610: {\bf Global fixed points}

1611: 

1612: Here we consider the case that an essential

1613: lamination $\lambda$ on $N$ would produce

1614: a trivial group action on a tree $T$. 

1615: 

1616: Recall the notion of {\em efficient} transversal

1617: to a lamination: let $\eta$ be a transversal

1618: to a lamination $\lambda$.

1619: Then $\eta$ is efficient \cite{Ga-Oe} if

1620: for any subarc $\eta_0$ with both endpoints 

1621: in leaves of $\lambda$ and interior disjoint

1622: from $\lambda$, then $\eta_0$ is not homotopic

1623: rel endpoints into a leaf of $\lambda$.

1624: Gabai and Oertel showed  that if $\lambda$ is

1625: essential then any efficient transversal

1626: cannot be homotoped rel endpoints into a leaf

1627: of $\lambda$. Also closed efficient transversals

1628: are not null homotopic.

1629: 

1630: 

1631: \begin{lemma}{}{}

1632: If $\lambda$ is an essential lamination in

1633: $N$ then the associated group action of

1634: $\pi_1(N)$ on a tree $T$ as described above

1635: has no global fixed point.

1636: \end{lemma}

1637: 

1638: \begin{proof}{}

1639: Suppose on the contrary that a point $x$ of

1640: $T$ is left invariant by the whole group.

1641: Look at the preimage of $x$ in the possibly

1642: non Hausdorff tree $T_*$.

1643: There are 3 options:

1644: 

1645: 

1646: $1 -$ $x$ comes from a non singular, Hausdorff leaf

1647: $E$ of $\wl$. Then $E$ is left invariant by the whole

1648: group $\pi_1(N)$,

1649: 

1650: $2 -$   $x$ comes

1651: from the closure $R$ of a complementary region 

1652: of $\wl$ in the universal

1653: cover. Then $R$ is left invariant by the whole group.

1654: In this case let $E$ be a boundary leaf of $R$.

1655: 

1656: $3 - $ Finally $x$ may come from a non Hausdorff 

1657: leaf $E$.  

1658: Then the orbit of $E$ under $\pi_1(N)$

1659: consists only of the non separated leaves from $E$.

1660: 

1661: By construction of the tree $T$ above

1662: these 3 cases are mutually 

1663: exclusive.

1664: It follows that in any of the 3 options there is

1665: at least one component $B$ of $\widetilde N - E$ which does

1666: not contain any translate of $E$. In option 1)

1667: any component will do, in option 2) choose the

1668: component not containing $R - E$ and in option

1669: 3) choose the component not containing leaves

1670: non separated from $E$. 

1671: 

1672: Let $A = \pi(E)$ where $\pi: \widetilde N \rightarrow N$ is

1673: the universal covering map. 

1674: Suppose first that $A$ is not compact.

1675: Then it limits on some leaves of $\lambda$ and

1676: there is a laminated box where $A$ intersects it

1677: in at least $3$ leaves and the box intersects

1678: an efficient transversal to $\lambda$.

1679: Lifting to $\widetilde N$ so that the middle leaf

1680: is $E$ then the other 2 leaves are not $E$ (efficient

1681: transversal) and one of them is contained in $B$

1682: producing a covering translate of $E$ in $B$, contradiction.

1683: The same is of course true if $A$ intersects

1684: an efficient closed transversal.

1685: 

1686: Now $A$ is compact. If $A$ is non separating, then

1687: it intersects a closed transversal associated to $g$ in

1688: $\pi_1(N)$ only once. Same proof yields either

1689: $Eg$ or $Eg^{-1}$ in $B$, done.

1690: 

1691: 

1692: Finally suppose that $A$ is separating. Then $C = \pi(B \cup E)$

1693: is a compact submanifold of $N$ which has $A$ as its

1694: unique boundary component.

1695: For any $g$ in $\pi_1(C)$ then $Eg$ is contained in $B \cup E$,

1696: so by hypothesis must be $E$, therefore $\pi_1(A)$ surjects

1697: in $\pi_1(C)$. As $\lambda$ is essential then $\pi_1(A)$

1698: also injects \cite{Ga-Oe}, so $\pi_1(A)$ is isomorphic

1699: to $\pi_1(C)$. As $C$ is irreducible \cite{Ga-Oe}, then

1700: theorem 10.5 of Hempel \cite{He} implies that $C$ is

1701: homemorphic to $A \times I$ with $A$ corresponding

1702: to $A \times \{ 0 \}$. This contradicts the

1703: fact that $A$ is the only boundary component

1704: of $C$. 

1705: This finishes the proof of the lemma.

1706: \end{proof}

1707: 

1708: 

1709: 

1710: \vskip .1in

1711: \noindent

1712: {\bf {Remark:}} $-$ Notice that leaves of essential

1713: laminations may not intersect a closed transversal.

1714: For example this occurs for separating incompressible

1715: surfaces. It also occurs for leaves

1716: of Reebless foliations which have a separating

1717: leaf (which necessarily must be a torus or Klein bottle)

1718: $-$ there are many examples of these.

1719: So Reebless foliations which are also essential

1720: laminations need not be taut foliations!

1721: 

1722: 

1723: 

1724: 

1725: \section{Case R $-$ the $\rrrr$-covered case}

1726: 

1727: 

1728: For the remainder of the article we consider

1729: the manifold $M_{p/q}$ as described in

1730: section \ref{grou} with fundamental group

1731: $\gl$. The goal is to show it does not

1732: admit an essential lamination. Suppose then on

1733: the contrary that there is an essential

1734: lamination $\lambda$ on $M_{p/q}$.

1735: Let $T$ be the associated tree with non trivial

1736: action of $\gl$ on it.

1737: Notice that since $\alpha, \tau$ generate $\gl$ then

1738: no point of $T$ is fixed by both $\alpha$ and $\tau$.

1739: 

1740: 

1741: 

1742: The conditions on the parameters are $|p - 2q| = 1$ and

1743: $m \leq -4$. They will not be used in full force

1744: for all the arguments. Many times all we need

1745: is $p \geq q$ or $p$ odd or $m$ negative or none

1746: of these. 

1747: The proof is done by subdiving into subcases and

1748: showing each subcase is impossible leading to 

1749: various contradictions.

1750: 

1751: 

1752: In this section we assume that $T$ is homeomorphic to

1753: the real numbers and study non trivial actions of $\gl$

1754: in $\rrrr$.

1755: Notice that $\gamma$ being a commutator is an orientation

1756: preserving homeomorphism of $\rrrr$. 

1757: Since $\tau^p \gamma^q = id$, then $\tau^p$ is also

1758: orientation preserving. 

1759: %Here we use the hypothesis that

1760: %$p$ is odd which implies that $\tau$ is now orientation

1761: %preserving.

1762: 

1763: We use the relations from the group presentation

1764: of $\gl$ or variations thereof.

1765: 

1766: 

1767: 

1768: Suppose first the action is orientation preserving on $\rrrr$:

1769: 

1770: \vskip .1in

1771: \noindent

1772: {\bf {Case R.1}} $-$ $\alpha$, $\tau$ are orientation

1773: preserving.

1774: 

1775: As $\beta = \tau \alpha \tau^{-1}$ then $\beta$ also is

1776: orientation preserving and so is the whole group $\gl$.

1777: We subdivide into subcases:

1778: 

1779: \vskip .1in

1780: \noindent

1781: {\bf {Case R.1.1}} $-$ $\tau$ has a fixed point $x$.

1782: 

1783: 

1784: Then $x \alpha$ is not $x$. Orient $\rrrr$ so that 

1785: $x \alpha > x$.

1786: As $\gamma$ is orientation preserving then $x \gamma = x$.

1787: Then  applying $\gamma \tau \beta \alpha^m = \alpha \tau$

1788: to $x$:

1789:  

1790: $$x \gamma \tau \beta \alpha^m = x \alpha \tau > x \tau = x$$

1791: 

1792: \noindent

1793: which uses $\tau$ orientation preserving. Hence $x \beta \alpha^m > x$ \

1794: or \ $x \beta > x \alpha^{-m} > x$ \ (as $-m > 0$).

1795: \ Hence $x \beta^{-1} < x$.

1796: But also 

1797: 

1798: $$x \beta^{-1} \ = \ x \tau \alpha \tau^{-1} \ = \ 

1799: x \alpha \tau^{-1} \ > \ x \tau^{-1} \ = \ x.$$

1800: 

1801: \noindent

1802: This is a contradiction, ruling out this case.

1803: 

1804: 

1805: \vskip .1in

1806: \noindent

1807: {\bf {Case R.1.2}} $-$ $\tau$ acts freely, $\alpha$ has

1808: a fixed point $x$.

1809: 

1810: Assume $\tau$ is increasing in $\rrrr$.

1811: As $\tau = \kappa^q$ and $q$ is positive

1812: then $\kappa$ is increasing.

1813: Here use $x \alpha \tau = x \tau = x \gamma \tau \beta \alpha^m$.

1814: Hence $x \tau \alpha^{-m} = x \gamma \tau \beta$.

1815: As $x \tau > x$ then $x \tau \alpha^{-m} > x$.

1816: Hence $x \gamma \tau > x \beta^{-1}$.

1817: Here $\gamma = \kappa^{-p}$ and $\gamma \tau = \kappa^{q - p}$.

1818: As $q \leq p$ then $q - p \leq 0$ and $\gamma \tau$ is

1819: monotone decreasing or constant.

1820: Hence

1821: 

1822: $$x \beta^{-1} \ < x \gamma \tau \ \leq \ x.$$

1823: 

1824: \noindent

1825: One fact that will be used in a lot of arguments is that

1826: under the condition $p \geq q$ when $\gamma, \tau$ act

1827: freely and $x \tau > x$ then

1828: $x \gamma \leq x \tau^{-1}$.

1829: Notice that $x \tau^{-1} \beta = x \alpha \tau^{-1} = x \tau^{-1}$.

1830: On the other hand

1831: 

1832: $$x \beta \ = \ x \alpha \beta \ = \ 

1833: x \gamma \beta \alpha \ \leq \ x \tau^{-1} \beta \alpha

1834: \ = \ x \tau^{-1} \alpha \ < \ x \alpha \ = \ x.$$

1835: 

1836: \noindent

1837: leading to the contradiction that both $x \beta$ and

1838: $x \beta^{-1}$ are $< x$.

1839: 

1840: Notice a lot of these arguments are using orientation

1841: preserving homeomorphims.

1842: 

1843: 

1844: \vskip .1in

1845: \noindent

1846: {\bf {Case R.1.3}} $-$ $\tau$ acts freely increasing in $\rrrr$

1847: and $\alpha$ acts freely, also increasing in $\rrrr$.

1848: 

1849: Take any $x$ in $\rrrr$. Then $x \alpha \tau > x$

1850: so $x \gamma \tau \beta \alpha^m > x$.

1851: So $x \gamma \tau \beta > x \alpha^{-m} > x$.

1852: Since $x \gamma \tau \leq x$ this implies 

1853: $x \beta > x$.

1854: On the other hand,

1855: 

1856: $$x \beta \ = \ x \tau \alpha^{-1} \tau^{-1} \ <

1857: \ x \tau \tau^{-1} \ =  \ x,$$

1858: 

1859: \noindent

1860: contradiction.

1861: 

1862: \vskip .1in

1863: \noindent

1864: {\bf {Case R.1.4}} $-$ $\tau$ acts freely and increasing

1865: in $\rrrr$, $\alpha$ acts freely and decreasing in $\rrrr$.

1866: 

1867: This implies $z \alpha^{-1} > z$ for all $z$ in $\rrrr$.

1868: For any $x$ in $\rrrr$,

1869: $x \beta = x \tau \alpha^{-1} \tau^{-1} > 

1870: x \tau \tau^{-1} = x$.

1871: Also $x \tau^{-1} \alpha \tau < x$ for all $x$. 

1872: Hence

1873: 

1874: $$x \alpha \beta \alpha^{m-1} \ = \ x \tau^{-1} \alpha \tau 

1875: \ < \ x,$$

1876: 

1877: \noindent

1878: for all $x$. Hence $x \alpha \beta < x \alpha \alpha^{-m}

1879: < x \alpha$ for all $x$ \ ($-m > 0$). But this contradicts

1880: $(x \alpha) \beta > x \alpha$ because $\beta$ is increasing

1881: everywhere as proved above.

1882: 

1883: This finishes the analysis of $T$ homeomorphic to

1884: $\rrrr$ and orientation preserving action.

1885: 

1886: 

1887: 

1888: 

1889: 

1890: 

1891: 

1892: %Suppose first that $\alpha$ acts as orientation

1893: %reversing.

1894: %

1895: %\vskip .1in

1896: %\noindent 

1897: %{\bf Claim} - $\tau$ acts as orientation preserving.

1898: %

1899: %Then $\alpha$ has a unique fixed point $x$. If

1900: %$x \tau = x$ then $x$ is a global fixed point,

1901: %contradiction. So choose orientation on $\rrrr$

1902: %so that $x \tau^{-1} < x$.

1903: %

1904: %Use the relations

1905: %

1906: %$$\tau^{-1} \alpha \tau = \gamma \beta \alpha^m$$

1907: %

1908: %$$\tau^{-1} \beta \tau = \alpha^{-1}$$

1909: %

1910: %Since $\gamma = [\alpha, \beta]$ then $\gamma$ is

1911: %orientation preserving. Since $\tau^p \gamma^q = 1$, then

1912: %$p$ is even and the only fixed point

1913: %$y$ of $\tau$ is in $(x \tau^{-1}, x)$. 

1914: %Then $x \tau^{-1}$ is the only fixed point of $\beta$.

1915: %Then $y \alpha > x$ and $y \alpha \beta < x \tau^{-1}$.

1916: %As $m$ is even then $y \alpha \beta \alpha^{m-1} > x$.

1917: %On the other hand $y \tau^{-1} = y$ so

1918: %$y \tau^{-1} \alpha > x$ and $y \tau^{-1} \alpha \tau < x \tau^{-1} < x$,

1919: %contradiction.

1920: %

1921: %Conclusion: if $\alpha$ is orientation reversing then

1922: %$\tau$ is orientation preserving.

1923: 

1924: 

1925: \vskip .2in

1926: We now deal with orientation reversing cases.

1927: The general case of $\tau$ orientation reversing is

1928: hard, so we use one of the hypothesis to discard it

1929: as follows: $\tau^p = \gamma^{-q}$ is orientation

1930: preserving as $\gamma$ always is.

1931: We are mainly interested in $|p-2q| = 1$, which

1932: implies $p$ odd and if $p$ is odd and $\tau^p$ orientation

1933: preserving then $\tau$ is also orientation preserving.

1934: We now deal with the case $\alpha$ orientation reversing.

1935: 

1936: 

1937: \vskip .1in

1938: \noindent

1939: {\bf {Case R.2}} $-$ $\alpha$ orientation reversing,

1940: $\tau$ orientation preserving.

1941: 

1942: Let $x$ be the unique fixed point of $\alpha$. As

1943: $x \tau \not = x$, assume $x \tau > x$.

1944: As $\tau = \kappa^q$ and $q > 0$, this implies $k$ is

1945: increasing in $x$.

1946: Notice that $x \tau^{-1}$ is the unique fixed point

1947: of $\beta$.

1948: The subcases depend on the relative position

1949: of $x \tau \alpha$ and $x \tau^{-1}$. Notice that $x \tau > x$,

1950: so $x \tau \alpha < x \alpha = x$.

1951: 

1952: 

1953: 

1954: \vskip .1in

1955: \noindent

1956: {\bf Case R.2.1} $-$ $x \tau \alpha < x \tau^{-1}$

1957: 

1958: Then $x \tau \alpha \tau^{-1} = x \beta^{-1} < x \tau^{-2}$.

1959: Notice 

1960: 

1961: $$x \tau \gamma \beta \alpha^m = x \alpha \tau = x \tau > x$$

1962: 

1963: \noindent

1964: so $x \tau \gamma \beta > x \alpha^{-m} = x$ and so

1965: 

1966: $$x \tau \gamma < x \beta^{-1} < x \tau^{-2}$$

1967: 

1968: \noindent

1969: or $x \tau^3 \gamma < x$. 

1970: As $\tau^3 = \kappa^{3q}$ and $\gamma = \kappa^{-q}$, then

1971: $x \kappa^{3q - p} < x$. As $\kappa$ is increasing

1972: in $x$ then $3q - p < 0$ or $p > 3q$.

1973: Arguments such as this will be used in various

1974: parts of the proof.

1975: Since in the end we want $p = 2q \pm 1$ we can discard

1976: this case.

1977: 

1978: \vskip .1in

1979: \noindent

1980: {\bf {Remark}} $-$ What we really wanted was to rule out

1981: this case without using $p = 2q \pm 1$, but we were

1982: unable to do that. Our partial results (without

1983: using $p = 2q \pm 1$) show that

1984: $x \tau \alpha^3 > x \tau \alpha$

1985: so $x < x \tau \alpha^2 < x \tau$.

1986: Also there is a fixed point of $\alpha^2$ between

1987: $x \tau$ and $x \tau^2$ and 

1988: $\alpha^2$ acts expandingly (away from $x$) in some point.

1989: %See pages Surgery 633-640.

1990: Something similar is also true in the following case.

1991: 

1992: 

1993: 

1994: 

1995: 

1996: \vskip .2in

1997: \noindent

1998: {\bf Case R.2.2} $-$ $x \tau \alpha > x \tau^{-1}$

1999: 

2000: %\vskip .1in

2001: %\noindent

2002: %{\bf Case R.2.2.1} $-$ $x \tau \alpha^2 \leq x \tau$.

2003: %

2004: %Here use $\tau^{-1} \alpha^{-1} \tau = \alpha^{1-m} \beta^{-1} \alpha^{-1}$.

2005: %Then 

2006: %

2007: %$$x \tau = (x \tau) \tau^{-1} \alpha^{-1} \tau \ = \

2008: %(x \tau) \alpha^{1-m} \beta^{-1} \alpha^{-1}$$

2009: %

2010: %

2011: %\noindent

2012: %so 

2013: %

2014: %$$x \tau \alpha \beta = (x \tau \alpha) \alpha^{-m}$$

2015: %

2016: %\noindent

2017: %But $x \tau \alpha > x \tau^{-1}$, hence

2018: %$x \tau \alpha \beta < x \tau^{-1}$,

2019: %because $\beta$ is orientation reversing with

2020: %fixed point $x \tau^{-1}$.

2021: %It follows that 

2022: %

2023: %$$ (x \tau \alpha) \alpha^{-m} < x \tau^{-1} < x \tau \alpha$$

2024: %

2025: %\noindent

2026: %Hence $\alpha^{-m}$ expands $x \tau \alpha$ away from $x$.

2027: %Notice $-m$ is even and positive. Hence $\alpha^2$ expands

2028: %$x \tau \alpha$ away from $x$ and consequently

2029: %$\alpha^2$ expands $x \tau$ away from $x$, contradiction

2030: %to this case. 

2031: %

2032: %We conclude that in case B we have $x \tau \alpha^2 > x \tau$.

2033: %

2034: %

2035: %

2036: %\vskip .4in

2037: %\noindent

2038: %{\bf Case B.2} - Suppose that $x \tau \alpha > x \tau^{-1}$

2039: %and $x \tau \alpha^2 > x \tau$.

2040: 

2041: First notice that $x \beta^{-1} < x \tau^{-1}$.

2042: Use

2043: 

2044: $$(x \tau) \tau \gamma \beta \alpha^m = (x \tau) \alpha \tau > 

2045: x \tau^{-1} \tau = x$$

2046: 

2047: \noindent

2048: so $x \tau^2 \gamma \beta > x \alpha^{-m} = x$ and 

2049: 

2050: $$ x \tau^2 \gamma < x \beta^{-1} < x \tau^{-1}.$$

2051: 

2052: \noindent

2053: We conclude as in the previous case that

2054: $x \tau^3 \gamma < x$ or $p > 3q$, also disallowed.

2055: 

2056: 

2057: %\noindent

2058: %{\underline {ASIDE}} In this case we can prove that $\alpha^2$

2059: %expands in $x \tau$ and there is a fixed

2060: %point of $\alpha^2$. See also pages Surgery 620-633.

2061: The reader may think we just got lucky to get $p > 3q$ as 

2062: we have the hypothesis 

2063: $p = 2q \pm 1$. The remaining case explains why this

2064: has happened.

2065: 

2066: 

2067: 

2068: 

2069: \vskip .2in

2070: \noindent

2071: {\bf {Case R.2.3}} $-$ $x \tau \alpha = x \tau^{-1}$.

2072: 

2073: This case is much more interesting.

2074: First

2075: 

2076: $$x \alpha \tau = x \tau \alpha \beta \alpha^{m-1}$$

2077: 

2078: 

2079: \noindent

2080: Since $x \tau \alpha = x \tau^{-1}$ this is left

2081: invariant by $\beta$, so the right side

2082: is $x \tau \alpha \alpha^{m-1} = x \tau \alpha^m$ 

2083: equal to $x \tau$.

2084: Since $m$ is even,  $\alpha^m$ preserves orientation,

2085: therefore $x \tau \alpha^2 = x \tau$.

2086: Also $x \tau \alpha = x \tau \alpha^{-1} = x \tau^{-1}$.

2087: Now notice that 

2088: 

2089: $$x \tau \gamma \beta \alpha^m = x \alpha \tau

2090: = x \tau, \ \ \ {\rm so} \ \ \ 

2091: x \tau \gamma = x \tau \alpha^{-m} \beta^{-1},$$

2092: 

2093: \noindent

2094: or $x \tau \gamma = x \tau \beta^{-1}$.

2095: Now we show that $x \tau^2 \alpha = x \tau^{-2}$.

2096: To show this use $x \beta^{-1} \tau = x \tau \alpha

2097: = x \tau^{-1}$, hence $x \beta^{-1} = x \tau^{-2}$.

2098: Use 

2099: 

2100: $$ \tau^{-2} \beta \tau^2 \ = \ \tau^{-1} \alpha^{-1} \tau

2101: \ = \ \alpha^{1-m} \beta^{-1} \alpha^{-1}$$

2102: 

2103: \noindent

2104: applied to $x$:

2105: 

2106: $$x \tau^{-2} \beta \tau^2 = x \alpha^{1-m} \beta^{-1} \alpha^{-1}$$

2107: 

2108: \noindent

2109: or 

2110: $x \beta^{-1} \beta \tau^2 = x \beta^{-1} \alpha^{-1}$ so

2111: 

2112: 

2113: $$x \tau^2 = x \tau^{-2} \alpha^{-1}.$$

2114: 

2115: \noindent

2116: Then

2117: 

2118: $$x \tau^{-2} = x \tau^2 \alpha =  (x \tau) \tau \alpha

2119: = x \tau \beta^{-1} \tau

2120: = x \tau \gamma \tau$$

2121: 

2122: \noindent

2123: or 

2124: 

2125: $$x  \gamma \tau^4 = x$$

2126: 

2127: \noindent

2128: As seen before this implies $p = 4q$ or $p = 4$,

2129: $q = 1$. 

2130: This is disallowed by $p$ being odd.

2131: 

2132: 

2133: We remark that in this case the group in fact acts non trivially

2134: in $\rrrr$. For instance let

2135: 

2136: $$x \alpha = -x, \ \ \ x \tau = x + 1$$

2137: 

2138: \noindent

2139: It is easy to check they satisfy the equations if $m$ is

2140: even!

2141: 

2142: It may be true that this is the only possibility and when

2143: $x \tau \alpha \not = x \tau^{-1}$ we get a perturbation

2144: of this, namely that $p$ is close to $4q$ and in fact

2145: $p > 3q$.

2146: 

2147: 

2148: 

2149: \section{Case A $-$ $\tau$ acts freely}

2150: 

2151: In this section we consider the case that $\tau$ acts

2152: freely in $T$.

2153: This implies that $\kappa^q$ acts freely in the tree,

2154: and therefore $\kappa$ itself acts freely.

2155: In addition the axes are the same $\aaa_{\kappa} = \at$.

2156: Here we will use the relation $\alpha \beta = \gamma \beta \alpha$

2157: in the following form, defining an element

2158: $u$ of $\gl$:

2159: 

2160: $$u \ = \ \alpha \tau \alpha^{-1} \tau^{-1}

2161: \ = \ \gamma \tau \alpha^{-1} \tau \alpha$$

2162: 

2163: \noindent

2164: We will consider the intersections $\ak \cap \ak \alpha$ and

2165: $\ak \cap \ak u$.

2166: The axis $\ak$ is homeomorphic to the real numbers.

2167: Put an order $<$ in $\ak$ so that 

2168: $x <  x \tau$ for any $x$ in $\ak$.

2169: This induces an order $\pa$ in $\ak \alpha$

2170: so that $x < y$ in $\ak$ if and only if \

2171: $x  \alpha \ \pa \ y \alpha$ \ in $\ak \alpha$ and similarly

2172: put order $\pu$  in $\ak u$ so that $x < y$

2173: in $\ak$ if and only if \ $x u \ \pu \ y u$ \ in $\ak u$.

2174: 

2175: \vskip .15in

2176: \noindent

2177: {\bf {Case A.1}} $-$ $\ak \alpha \cap \ak$ has at

2178: most one point.

2179: 

2180: If the intersection is a single point $x$,

2181: let $y = x$ as well.

2182: 

2183: If they are disjoint, there is a single point

2184: $x$ in $\ak$ bridging to $\ak \alpha$. 

2185: For intance $x$ is the unique point so that

2186: there is a path from $x$ to $\ak \alpha$ intersecting

2187: $\ak$ only in $x$. Another way to characterize $x$,

2188: it is the only point so that $x$ separates the

2189: rest of $\ak$ from $\ak \alpha$. In other words

2190: the components of $T - \{ x \}$ containing $\ak

2191: \alpha$ and the rest of $\ak$ are all disjoint.

2192: In the same way there is a single $y$ in $\ak \alpha$ which

2193: is the closest to $\ak$. Then $[x,y]$ is a path

2194: from $\ak$ to $\ak \alpha$ so that $(x,y)$ does 

2195: not intersect either $\ak$ or $\ak \alpha$ $-$

2196: this is an equivalent way to get the segment $[x,y]$.

2197: This path $[x,y]$ is called the {\em bridge}

2198: from $\ak$ to $\ak \alpha$. This extended notion

2199: of bridges will also be used in the article.

2200: It is invariant by homemorphisms of the tree.

2201: The bridge between connected sets is also unique.

2202: 

2203: We now use the relation above. 

2204: The proof is very similar to ping pong lemma

2205: arguments. Since $\ak$ is invariant

2206: under $\gamma$ and $\tau$, the right side

2207: says that $\ak u = \ak \alpha^{-1} \tau^{-1} \alpha$.

2208: 

2209: 

2210: %\blankfig{02}{1.}{

2211: %The case $\ak \cap \ak \alpha = \emptyset$.

2212: %The same arguments can be used for intersection

2213: %a single point. a. Using $\ak u = \ak \alpha^{-1} \tau^{-1} \alpha$,

2214: %b. Using $\ak u = \ak \alpha \tau \alpha^{-1} \tau^{-1}$.}

2215: 

2216: 

2217: 

2218: \begin{figure}

2219: \centeredepsfbox{lam2.eps}

2220: \caption{

2221: The case $\ak \cap \ak \alpha = \emptyset$.

2222: The same arguments can be used for intersection

2223: a single point. a. Using $\ak u = \ak \alpha^{-1} \tau^{-1} \alpha$,

2224: b. Using $\ak u = \ak \alpha \tau \alpha^{-1} \tau^{-1}$.}

2225: \label{02}

2226: \end{figure}

2227: 

2228: 

2229: 

2230: 

2231: 

2232: The bridge from $\ak$ to $\ak \alpha$ is $[x,y]$ -

2233: degenerate $[x,x]$ when they intersect in a point.

2234: Therefore the bridge from $\ak \alpha^{-1}$ to

2235: $\ak$ \ is $[x \alpha^{-1}, y \alpha^{-1}]$, see fig. 

2236: \ref{02}, a.

2237: Then the bridge $\ak \alpha^{-1} \tau^{-1}$ to

2238: $\ak$ is $[x \alpha^{-1} \tau^{-1},

2239: y \alpha^{-1} \tau^{-1}]$.

2240: This implies that 

2241: the bridge from

2242: 

2243: $$\ak \alpha^{-1} \tau^{-1} \alpha \ \ \ {\rm  to} \ \ \

2244: \ak \alpha \ \ \ {\rm  is } \ \ \

2245: [x \alpha^{-1} \tau^{-1} \alpha, \  y \alpha^{-1} \tau^{-1} 

2246: \alpha].$$

2247: 

2248: \noindent

2249: Notice that $y \alpha^{-1} \tau^{-1}$ is not $y \alpha^{-1}$.

2250: Therefore $y \alpha^{-1} \tau^{-1} \alpha$ is not $y$. 

2251: It now follows that

2252: 

2253: $${\rm the \ bridge \ from} \ \ \ 

2254: \ak u = \ak \alpha^{-1} \tau^{-1} \alpha \ \ \ {\rm  to}

2255: \ \ \ \ak \ \ \ 

2256: {\rm is} 

2257: \ \ \ [x \alpha^{-1} \tau^{-1} \alpha, x].$$

2258: 

2259: 

2260: \noindent

2261: 

2262: On the other hand use that $\ak u = \ak \alpha \tau

2263: \alpha^{-1} \tau^{-1}$.

2264: The bridge from $\ak \alpha \tau$ to $\ak$ is

2265: $[y \tau, x \tau]$, see fig. \ref{02}, b.

2266: The bridge from $\ak \alpha \tau \alpha^{-1}$

2267: to $\ak \alpha^{-1}$ is 

2268: $[y \tau \alpha^{-1}, x \tau \alpha^{-1}]$ and

2269: the bridge from $\ak \alpha^{-1}$ to $\ak$ is

2270: $[x \alpha^{-1}, y \alpha^{-1}]$.

2271: Since $x \alpha^{-1}$ is not equal 

2272: $x \tau \alpha^{-1}$ then the bridge from

2273: $\ak \alpha \tau \alpha^{-1}$ to $\ak$ is

2274: $[y \tau \alpha^{-1}, y \alpha^{-1}]$. 

2275: Finally 

2276: 

2277: $${\rm the \ bridge \ from} \ \ \ \ak u

2278: \ \ {\rm  to} \  \ \ak  \ \ {\rm is} \ \ \

2279: [y \tau \alpha^{-1} \tau^{-1}, \ y \alpha^{-1} \tau^{-1}].$$

2280: 

2281: \noindent

2282: Since the bridge from $\ak u$ to $\ak$ is uniquely defined

2283: this implies

2284: 

2285: $$ y \alpha^{-1} \tau^{-1} \ = \ x, 

2286: \ \ \ \ y \tau \alpha^{-1} \tau^{-1} \ = \ 

2287:  x \alpha^{-1} \tau^{-1} \alpha.$$

2288: 

2289: 

2290: \noindent

2291: So  $y = x \tau \alpha$ and

2292: 

2293: $$x \alpha^{-1} \tau^{-1} \alpha \ = \ 

2294: x \tau \alpha \tau \alpha^{-1} \tau^{-1},

2295: \ \ \ \ {\rm or} \ \ \ \

2296: x \alpha^{-1} \tau^{-1} \alpha \tau \alpha \ = \ 

2297: x \tau \alpha \tau.$$

2298: 

2299: 

2300: \noindent

2301: Use $\tau^{-1} \alpha \tau = \alpha \beta \alpha^{m-1}$,

2302: so 

2303: 

2304: $$\alpha^{-1} \tau^{-1} \alpha \tau \alpha \ = \

2305: \alpha^{-1} \alpha \beta \alpha^{m-1} \alpha \ = \ 

2306: \beta \alpha^m \ = \ 

2307: \gamma^{-1} \tau^{-1} \alpha \tau,$$

2308: 

2309: \noindent

2310: so $x \gamma^{-1} \tau^{-1} \alpha \tau = x \tau \alpha \tau$,

2311: or $x \gamma^{-1} \tau^{-1} = x \tau$. This implies

2312: $x \gamma \tau^2 = x$ and as seen before implies

2313: $p = 2q$. This is disallowed by $p$ odd.

2314: 

2315: 

2316: \vskip .1in

2317: We now consider intersections with more than one point.

2318: 

2319: \vskip .15in

2320: \noindent

2321: {\bf {Case A.2}} $-$ $\ak \cap \ak \alpha \ = \ [x,y]$.

2322: 

2323: Here $x$ is not equal to $y$ and $x < y$ in $\ak$.

2324: We include some ideal point cases: $x$ could $-\infty$

2325: and $y$ could be $+\infty$, in which case

2326: the intersection is a ray in $\ak$.

2327: On the other hand we can never have $\ak = \ak \alpha$.

2328: Otherwise $\alpha, \tau$ leave $\ak$ invariant, so

2329: the whole group does. But $\ak$ is homemorphic to

2330: $\rrrr$ $-$ this was disallowed by no actions

2331: on $\rrrr$.

2332: 

2333: Since the intersection is a non trivial interval one

2334: considers separately whether the orders $<$, $\pa$ agree

2335: on the intersection.

2336: 

2337: \vskip .15in

2338: \noindent

2339: {\bf {Case A.2.1}} $-$ The orders $<$ and $\pa$ agree on

2340: $\ak \cap \ak \alpha$.

2341: 

2342: It is easy to check that this is equivalent to

2343: \ $x \alpha^{-1} < \ y \alpha^{-1}$ \ in $\ak$,

2344: by applying $\alpha$ to the pair $x \alpha^{-1},

2345: y \alpha^{-1}$ both of which are in $\ak$.

2346: 

2347: We now consider $\ak u$. We first use $\ak u = \ak \alpha^{-1}

2348: \tau^{-1} \alpha$. Notice that

2349: 

2350: $$\ak \cap \ak \alpha^{-1} \ = \ 

2351: [x \alpha^{-1}, y \alpha^{-1}] \ \ \ \ \

2352: {\rm so}

2353: \ \ \ \ \ 

2354: \ak \alpha^{-1} \tau^{-1} \cap \ak \ = \

2355: [x \alpha^{-1} \tau^{-1}, \ y \alpha^{-1} \tau^{-1}],$$

2356: 

2357: \noindent

2358: in the correct order. Hence

2359: 

2360: $$\ak u \cap \ak \ = \

2361: [x \alpha^{-1} \tau^{-1} \alpha, \

2362: y \alpha^{-1} \tau^{-1} \alpha].$$

2363: 

2364: \noindent

2365: In addition \

2366: $x \alpha^{-1} \tau^{-1} \alpha \ \pa \ 

2367: y \alpha^{-1} \tau^{-1} \alpha$.

2368: 

2369: 

2370: %\blankfig{03}{1.}{

2371: %Evaluating $\ak u \cap \ak$,

2372: %using $\ak u = \ak \alpha^{-1} \tau^{-1} \alpha$,

2373: %\ a. $y \alpha^{-1} \tau^{-1} \alpha \pa x$,

2374: %\ b. $y \alpha^{-1} \tau^{-1} \alpha >_{\alpha}  x$,

2375: %\ c. $y \alpha^{-1} \tau^{-1} \alpha =  x$.}

2376: 

2377: 

2378: 

2379: \begin{figure}

2380: \centeredepsfbox{lam3.eps}

2381: \caption{

2382: Evaluating $\ak u \cap \ak$,

2383: using $\ak u = \ak \alpha^{-1} \tau^{-1} \alpha$,

2384: \ a. $y \alpha^{-1} \tau^{-1} \alpha \pa x$,

2385: \ b. $y \alpha^{-1} \tau^{-1} \alpha >_{\alpha}  x$,

2386: \ c. $y \alpha^{-1} \tau^{-1} \alpha =  x$.}

2387: \label{03}

2388: \end{figure}

2389: 

2390: 

2391: 

2392: 

2393: 

2394: Notice that $x \alpha^{-1} \tau^{-1} < x \alpha^{-1}$

2395: in $\ak$, hence \ $x \alpha^{-1} \tau^{-1} \alpha \

2396: \pa \ x$ \ in $\ak \alpha$.

2397: Also \ $y \alpha^{-1} \tau^{-1} \alpha \ \pa \ y$ 

2398: \ in $\ak \alpha$.

2399: Given this there are 3 options:

2400: 

2401: %\begin{itemize}

2402: 

2403: %\item

2404: 1) \ If \ $y \alpha^{-1} \tau^{-1} \alpha \ \pa \ x$ \ in

2405: $\ak \alpha$ then 

2406: $\ak u \cap \ak = \emptyset$ and the

2407: bridge from $\ak$ to $\ak u$ is $[x, y \alpha^{-1}

2408: \tau^{-1} \alpha]$, fig. \ref{03}, a.

2409: 

2410: %\item

2411: 2) \ If \ $y \alpha^{-1} \tau^{-1} \alpha \ >_{\alpha} \ x$ \ in

2412: $\ak \alpha$ then $y \alpha^{-1} \tau^{-1} \alpha$

2413: is in $(x,y)$ and $\ak u \cap \ak = [x, y \alpha^{-1}

2414: \tau^{-1} \alpha]$. In addition the orders

2415: $<$ and $\pu$ agree on $\ak \cap \ak \alpha$,

2416: see fig. \ref{03}, b.

2417: 

2418: %\item

2419: 3) \ If $y \alpha^{-1} \tau^{-1} \alpha = x$, then

2420: $\ak \alpha  \cap \ak = [z,x]$. In addition if $z$ is not

2421: $x$ then 

2422: the orders $<$ and $\pu$ disagree

2423: on $\ak \cap \ak u$, see fig. \ref{03}, c. In this case both

2424: $x$ and $y$ are finite.

2425: The last option can occur because $\ak u$ can enter

2426: $\ak$ in $x$ but rather than going up, going in

2427: the opposite direction $-$ the one containing

2428: $x \tau^{-1}$.

2429: 

2430: %\end{itemize}

2431: 

2432: \vskip .1in

2433: 

2434: Notice that the 3 options are mutually exclusive.

2435: We now consider $\ak u = \ak \alpha \tau \alpha^{-1} \tau^{-1}$.

2436: Use 

2437: 

2438: $$\ak u \cap \ak \ = \ 

2439: (\ak \alpha \tau \cap \ak \alpha) \alpha^{-1}

2440: \tau^{-1}.$$

2441: 

2442: \noindent

2443: Here $\ak \alpha \tau \cap \ak = [x \tau, y \tau]$.

2444: So whether $\ak \alpha \tau \alpha^{-1}$

2445: and $\ak$ intersect, depends on the relative positions

2446: of $x \tau$ and $y$.

2447: Notice that $x \tau > x$ in $\ak$.

2448: 

2449: %\blankfig{04}{1.}{

2450: %Using $\ak u = \ak \alpha \tau \alpha^{-1}

2451: %\tau^{-1}$,

2452: %\ a. $x \tau > y$, \ b. $x \tau < y$, \ c. $x \tau = y$.}

2453: 

2454: 

2455: 

2456: 

2457: 

2458: \begin{figure}

2459: \centeredepsfbox{lam4.eps}

2460: \caption{

2461: Using $\ak u = \ak \alpha \tau \alpha^{-1}

2462: \tau^{-1}$,

2463: \ a. $x \tau > y$, \ b. $x \tau < y$, \ c. $x \tau = y$.}

2464: \label{04}

2465: \end{figure}

2466: 

2467: 

2468: %\begin{itemize}

2469: 

2470: %\item

2471: 1')  \ If $x \tau > y$ in $\ak$ then \ $\ak \alpha \tau \cap 

2472: \ak  \alpha = \emptyset$, so

2473: \ $\ak \alpha \tau \alpha^{-1} \cap \ak = \emptyset$.

2474: Therefore $\ak u \cap \ak = \emptyset$ and

2475: the bridge from $\ak$ to $\ak u$ is

2476: \ $[y \alpha^{-1} \tau^{-1}, \  x \tau \alpha^{-1} \tau^{-1}]$,

2477: see fig. \ref{04}, a.

2478: Here $x, y$ finite.

2479: 

2480: %\item

2481: 2') \  If $x \tau < y$ in $\ak$ then 

2482: $\ak \alpha \tau \cap \ak \alpha =

2483: [x \tau, y]$, then $\ak \cap \ak u$ is

2484: \ $[x \tau \alpha^{-1} \tau^{-1},

2485: y \alpha^{-1} \tau^{-1}]$

2486: (the first term smaller in $\ak$),

2487: and the orders $<$ and $\pu$ agree on

2488: $\ak \cap \ak u$, see fig. \ref{04}, b.

2489: 

2490: %\item 

2491: 3') \ If $x \tau = y$, then  $\ak \alpha \tau \cap \ak \alpha

2492: = [y, v]$.

2493: Notice we may have $v \not = y$. So 

2494: $\ak u \cap \ak = [y \alpha^{-1} \tau^{-1}, w]$,

2495: where $w = v \alpha^{-1} \tau^{-1}$.

2496: Here $x$ and $y$ are finite and if

2497: $w$ is not equal to $x \tau \alpha^{-1} \tau^{-1}$, then

2498: the orders $<$ and $\pu$ disagree on $\ak \cap \ak u$.

2499: Notice that order in $\ak \alpha \tau$ goes from

2500: $v$ to $y$, so the increasing order

2501: $\pu$ in $\ak u$

2502: from $w = v \alpha^{-1} \tau^{-1}$

2503: to $y \alpha^{-1} \tau^{-1}$, see fig. \ref{04}, c.

2504: 

2505: %\end{itemize}

2506: 

2507: \vskip .1in

2508: Notice that again all 3 cases are mutually exclusive.

2509: Therefore we can match the 2 pairs of 3 possibilities to

2510: get 3 mutually exclusive cases:

2511: 

2512: \vskip .1in

2513: %\begin{itemize}

2514: 

2515: %\item 

2516: {\bf I} $-$ $y \alpha^{-1} \tau^{-1} \alpha \ \pa \ x$ \ in

2517: $\ak \alpha$ or $x \tau > y$ in $\ak$ and

2518: $\ak \cap \ak u = \emptyset$. In this case

2519: 

2520: $$[x, \  y \alpha^{-1} \tau^{-1} \alpha] \ = \

2521: [y \alpha^{-1} \tau^{-1}, \ x \tau \alpha^{-1} \tau^{-1}]$$

2522: 

2523: %\item

2524: {\bf II} $-$ $y \alpha^{-1} \tau^{-1} \alpha \ \poa \ x$ \ in

2525: $\ak \alpha$ or $x \tau < y$ in $\ak$ and

2526: 

2527: 

2528: 

2529: $$\ak \cap \ak u \ = \ 

2530: [x, \ y \alpha^{-1} \tau^{-1} \alpha] \ = \ 

2531: [x \tau \alpha^{-1} \tau^{-1}, \ y \alpha^{-1} \tau^{-1}]$$

2532: 

2533: %\item

2534: {\bf III} $-$ $y \alpha^{-1} \tau^{-1} \alpha = x$

2535: or $x \tau = y$. 

2536: Then 

2537: 

2538: $$\ak \cap \ak u \ = \ 

2539: [z, x] \ = \ [y \alpha^{-1} \tau^{-1}, w]$$

2540: 

2541: \noindent

2542: If $z$ is not $x$ then the orders

2543: $<$ and $\pu$ disagree on $\ak \cap \ak u$.

2544: 

2545: %\end{itemize}

2546: \vskip .1in

2547: 

2548: We now deal with each situation separately.

2549: 

2550: 

2551: \vskip .1in

2552: \noindent

2553: {\bf {Situation II}} $-$ 

2554: 

2555: Here $x \tau \alpha = x \tau$ and 

2556: $x \tau$ is in $(x,y)$.

2557: Let  $\uu_1$, (respectively $\uu_2$)

2558: be the component of $T - \{ x \tau \}$ 

2559: containing $y$ (respectively $x$).

2560: Here $[x,y] = \ak \cap \ak \alpha$,

2561: $x \tau$ is in the interior of $[x,y]$ 

2562: and then the orders

2563: $<$, $\pa$ agree on $[x,y]$. It

2564: follows that the prongs $[x \tau, y]$,

2565: $[x \tau, y \alpha]$ are equivalent. 

2566: By lemma \ref{inva}, $\uu_1 \alpha = \uu_1$. 

2567: In the same way

2568: $\uu_2 \alpha = \uu_2$. 

2569: This situation is disallowed by the following lemma.

2570: 

2571: \begin{lemma}{}{}

2572: Suppose that ${\cal L}$ is a local axis for $\tau$

2573: and $r$ is a point in ${\cal L}$ with

2574: $r \alpha = r$.

2575: Suppose that $\uu_1$ ($\uu_2$ respectively) is

2576: the component of $T - \{ r \}$ containing $r \tau$

2577: ($r \tau^{-1}$ respectively). Then at least one

2578: of $\uu_1$ or $\uu_2$ is not invariant under $\alpha$.

2579: \end{lemma}

2580: 

2581: \begin{proof}{}

2582: On the contrary suppose that $\uu_i \alpha = \uu_i$ for

2583: $i = 1,2$.

2584: We will arrive at a contradiction.

2585: Let $\vv_i = \uu_i \tau^{-1}$. Then the conjugation of

2586: $\beta$ with $\alpha^{-1}$ by $\tau$ implies that

2587: $\vv_i \beta = \vv_i$, $i = 1,2$.

2588: Use

2589: 

2590: 

2591: 

2592: $$r \tau^{-1} \alpha \tau \ = \ r \gamma \beta \alpha^m$$

2593: 

2594: \noindent

2595: Since $p \geq q$, then $r \gamma \leq r \tau^{-1}$ in

2596: ${\cal L}$ (with $\tau$ increasing in ${\cal L}$

2597: and so $r \gamma \beta$ is in $\vv_2 \cup \{ r \tau^{-1} \}$

2598: contained  in $\uu_2$. Therefore $r \gamma \beta \alpha^m$

2599: is in $\uu_2$.

2600: Consequently $r \tau^{-1} \alpha \tau$ is in $\uu_2$ and

2601: $r \tau^{-1} \alpha$ is in 

2602: $\uu_2 \tau^{-1} = \vv_2$ \ \ \ (*).

2603: 

2604: On the other hand $r \gamma \in \vv_2 \cup \{ r \tau^{-1} \}$,

2605: so 

2606: 

2607: $$r \beta \alpha^{-1} \ = \ 

2608: r \gamma \beta \ \in \ \vv_2 \cup \{ r \tau^{-1} \},$$

2609: 

2610: \noindent

2611: so $r \tau^{-1}$ is in $[r \beta \alpha^{-1},r)$. Apply 

2612: $\alpha$ to obtain 

2613: 

2614: $$r \tau^{-1} \alpha \  \in \

2615: [r \beta, r) \ \ \ \ \ (**).$$

2616: 

2617: 

2618: \noindent

2619: Now

2620: 

2621: $$r \beta \ = \ r \tau \alpha^{-1} \tau^{-1}

2622: \ \ \ {\rm and} \ \ \ 

2623: r \tau \in \uu_1 \ \Rightarrow  \  r \tau \alpha^{-1} \in \uu_1

2624: \ \Rightarrow \ r \beta = r \tau \alpha^{-1} \tau^{-1}

2625: \in \vv_1.$$

2626: 

2627: \noindent

2628: As $r$ is also in $\vv_1$, it follows

2629: from (**)

2630: that $r \tau^{-1} \alpha$ is also in $\vv_1$.

2631: This contradicts (*) above and finishes the proof.

2632: \end{proof}

2633: 

2634: 

2635: 

2636: \vskip .15in

2637: \noindent

2638: {\bf {Situation III}} $-$  Here 

2639: $\ak u \cap \ak = [z, x]$ with $z \leq x$ in $\ak$.

2640: Then 

2641: 

2642: $$\ak u \tau \cap \ak \ = \ 

2643: \ak \alpha \tau \alpha^{-1} \cap \ak \ = \ 

2644: [z \tau, x \tau] \ = \ [z \tau, y].$$

2645: 

2646: 

2647: %\blankfig{05}{.5}{

2648: %Situation III leading to a contradiction.}

2649: 

2650: 

2651: 

2652: \begin{figure}

2653: \centeredepsfbox{lam5.eps}

2654: \caption{

2655: Situation III leading to a contradiction.}

2656: \label{05}

2657: \end{figure}

2658: 

2659: 

2660: \noindent

2661: Hence $\ak \alpha \tau \cap \ak \alpha 

2662: = [z \tau \alpha, y \alpha]$

2663: and \ $z \tau \alpha  = y \ \leq_{\alpha} \ y \alpha$

2664: in $\ak \alpha$ $-$ this is the crucial fact.

2665: %But as $x \tau = y$

2666: %then $\ak \alpha \tau \cap \ak \alpha = [y, w]$ for

2667: %some $w$ with $y \leq \alpha w$.

2668: %It follows that $y \alpha = w$ and

2669: %$y \leq_{\alpha} y \alpha$.

2670: Now

2671: 

2672: $$x \gamma^{-1} \alpha \ = \ 

2673: x \beta \alpha \beta^{-1} \ = \ 

2674: x \tau \alpha^{-1} \tau^{-1} \alpha \beta^{-1}

2675: \ = \ y \alpha^{-1} \tau^{-1} \alpha \beta^{-1} $$

2676: 

2677: $$ \ = \ x \beta^{-1} \ = \ x \tau \alpha \tau^{-1}

2678: \ = \ y \alpha \tau^{-1}.$$

2679: 

2680: 

2681: \noindent

2682: Here the bridge of $y \alpha$ to $\ak$ is \

2683: $[y \alpha, y]$ \ (which a priori could be the single point

2684: $y$). So the bridge from 

2685: $y \alpha \tau^{-1}$ to

2686: $\ak$ is $[y \alpha \tau^{-1}, \ y \tau^{-1}]

2687: \ = \ [y \alpha \tau^{-1}, x]$.

2688: On the other hand 

2689: $y \leq  x \gamma^{-1}$ in $\ak$ so

2690: $y \alpha \leq_{\alpha} x \gamma^{-1} \alpha$

2691: in $\ak \alpha$.

2692: It follows that the bridge

2693: from 

2694: $x \gamma^{-1} \alpha$ to $\ak$ 

2695: is $[x \gamma^{-1} \alpha, y]$.

2696: This would imply $x = y$, contradiction.

2697: 

2698: 

2699: \vskip .2in

2700: \noindent

2701: {\bf {Situation I}} 

2702: $-$ Surprisingly this is the most difficult

2703: case.

2704: Here 

2705: 

2706: $$y \alpha^{-1} \tau^{-1} \alpha \pa x

2707: \ {\rm in} \

2708: \ak \alpha, \  

2709: \ \ x \tau > y \ {\rm in } \ \ak,

2710: \ 

2711: \ \ x = y \alpha^{-1} \tau^{-1}, \ 

2712: \ \ y \alpha^{-1} \tau^{-1} \alpha = 

2713: x \tau \alpha^{-1} \tau^{-1}.$$

2714: 

2715: 

2716: 

2717: 

2718: As \ $y \alpha^{-1} \tau^{-1} \alpha \ \pa \ x$ in

2719: $\ak \alpha$ then $y \alpha^{-1} \tau^{-1} \alpha$ is 

2720: not in $\ak$. Also 

2721: 

2722: $$x \alpha \ = \ 

2723: (y \alpha^{-1} \tau^{-1}) \alpha \ = \ 

2724: x \tau \alpha^{-1} \tau^{-1} \ = \ x \beta,$$

2725: 

2726: 

2727: \noindent

2728: so $x \alpha = x \beta$ $-$ this is a crucial

2729: fact in this proof. 

2730: The bridge from $x \alpha$ to $\ak$ is

2731: $[x \alpha, x]$. Notice also  that

2732: 

2733: $$x \alpha^{-1} \tau^{-1} \alpha \ \ \pa  \ \ 

2734: y \alpha^{-1} \tau^{-1} \alpha \ \ \pa \ \ x

2735: \ \ \ {\rm in} \ \ \ \ak \alpha,$$

2736: 

2737: \noindent

2738: so the bridge

2739: from $x \alpha^{-1} \tau^{-1} \alpha$ to

2740: $\ak$ is

2741: $[x \alpha^{-1} \tau^{-1} \alpha, \ x]$.

2742: It follows that 

2743: 

2744: $${\rm the \ bridge \ from} \ \ 

2745: x \alpha \tau^{-1} \alpha \tau

2746: \ {\rm to} \ \ak \ \ {\rm is }

2747: \ \ [x \alpha^{-1} \tau^{-1} \alpha \tau, \ x \tau] \ = \ 

2748: [x \alpha^{-1} \tau^{-1} \alpha \tau, \ y \alpha^{-1}].$$

2749: 

2750: \noindent

2751: Now

2752: 

2753: $$x \alpha^{-1} \tau^{-1} \alpha \tau \ = \ 

2754: (x \alpha^{-1}) \alpha \beta \alpha^{m-1} \ = \ 

2755: x \beta \alpha^{m-1} \ = \ x \alpha \alpha^{m-1}

2756: \ = \ x \alpha^{m}$$

2757: 

2758: \noindent

2759: Here \ $x \alpha   \ \prec \ x \ \prec \ y \

2760: \prec \ y \alpha^{-1}$ $-$ they are aligned.

2761: It follows from lemma \ref{chax} that

2762: $x, x \alpha$ are in a local axis $\lal$ for

2763: $\alpha$, similarly $y$ is also in a local axis.

2764: Since $y$ is in $[x \alpha^m, x]$, then

2765: also $y, y \alpha^{-1}$ are in $\lal$.

2766: In the same

2767: way 

2768: $(\lal) \tau^{-1} = \lab$ is a local axis

2769: for $\beta$ and 

2770: $x \beta, x, x \tau^{-1}$ are in $\lab$.

2771: Now 

2772: 

2773: $$x \beta \ = \ x  \tau \alpha^{-1} \tau^{-1} \ = \ 

2774: x \alpha, \ \ \ \ {\rm so} \ \ 

2775: \ \ x \alpha \tau \ = \ x \tau \alpha^{-1} \ = \

2776: y \alpha^{-2}$$

2777: 

2778: \noindent

2779: Apply $\alpha \beta \alpha^{m-1}

2780: = \tau^{-1} \alpha \tau$ to $y \alpha^{-1}$:

2781: 

2782: 

2783: $$(y \alpha^{-1}) \alpha \beta \alpha^{m-1} \ = \

2784: y \beta \alpha^{m-1} \ =  \

2785: (y \alpha^{-1}) \tau^{-1} \alpha \tau \ = \ 

2786: (x \tau) \tau^{-1} \alpha \tau

2787: \ = \ x \alpha \tau \ = \ y \alpha^{-2}.$$

2788: 

2789: %\blankfig{06}{1.}{

2790: %Situation I, the hard case.}

2791: 

2792: 

2793: 

2794: \begin{figure}

2795: \centeredepsfbox{lam6.eps}

2796: \caption{

2797: Situation I, the hard case.}

2798: \label{06}

2799: \end{figure}

2800: 

2801: 

2802: 

2803: \noindent

2804: The conclusion is \ \ $y \beta = y \alpha^{-m -1}$ \ \ and

2805: it is in $\lal$.

2806: Now $y$ is not in $\lab$ and the bridge from

2807: $y$ to $\lab$ is $[y,x]$, so the

2808: bridge from $y \beta$ to $\lab$

2809: is $[y \beta, x \beta] \ = \ [y \alpha^{-m-1}, 

2810: \ x \alpha]$.

2811: Therefore $\lal$ and $\lab$ split away from each

2812: other in $x \alpha = x \beta$,

2813: or

2814: 

2815: 

2816: $$ \lal \cap \lab \ = \ [x, x \alpha] \ = \ [x, x \beta].$$

2817: 

2818: \noindent

2819: The homeomorphism $\tau$ conjugates 

2820: the action of $\alpha^{-1}$ in $\lal$ to

2821: the action of $\beta$ in $\lab$ (see fig. \ref{06}).

2822: Now apply $\alpha \tau \alpha^{-m} = \tau \gamma \beta$ to

2823: $x$:

2824: 

2825: $$ (x \alpha \tau) \alpha^{-m} \ = \

2826: (y \alpha^{-2}) \alpha^{-m} \ = \

2827: y \alpha^{-2-m} \ = \ x \tau \gamma \beta.$$

2828: 

2829: 

2830: \noindent

2831: As $x \alpha$ is in $\lab$, then

2832: $x \alpha \tau$ is in $\lal$ and it follows that

2833: $x \tau \gamma \beta$ is in $\lal$.

2834: If $x \tau \gamma \leq x \tau^{-1}$ in $\ak$,

2835: then the bridge from $x \tau \gamma$ to

2836: $\lab$ is $[x \tau \gamma, x \tau^{-1}]$

2837: and so the bridge from

2838: $x \tau \gamma \beta$ to $\lab$ is

2839: $[x \tau \gamma \beta, x \tau^{-1} \beta]$.

2840: But $x \tau^{-1} \beta = x \alpha^{-1} \tau^{-1}$

2841: and 

2842: 

2843: $$x \alpha^{-1} \tau^{-1} \ <  \

2844: y \alpha^{-1} \tau^{-1}  \ < \ x \ \ \ 

2845: {\rm in } \ \ \ \ak.$$

2846: 

2847: \noindent

2848: This would imply $x \tau \gamma \beta$ is

2849: not in $\lal$, contradiction.

2850: Notice 

2851: 

2852: $$x \beta^{-1} \ = \ x \tau \alpha \tau^{-1}

2853: \ = \ y \tau^{-1} \ \ \in \ (x \tau^{-1}, x).$$

2854: 

2855: \noindent

2856: If $x \tau \gamma$ is in

2857: $[x \tau^{-1}, \ x  \beta^{-1})$

2858: then $x \tau \gamma \beta$ is

2859: in $[x \tau^{-1} \beta, x)$ and

2860: not in $\lal$ either, contradiction again.

2861: Therefore $x \tau \gamma$ is

2862: in $[x \beta^{-1}, x]$.

2863: The case $x \tau \gamma = x$ can only occur

2864: when $p = q = 1$. This case can also be ruled out by

2865: a further argument, but as we are mainly interested

2866: in $|p-2q| = 1$ we assume here that $p > q$.

2867: Then $x \tau \gamma$ is in $[x \beta^{-1}, x)$ and

2868: $x \tau \gamma \beta$ is in $[x, x \beta)$.

2869: We conclude that

2870: 

2871: $$y \alpha^{-2-m} \ \in \ [x, x \alpha).$$

2872: 

2873: \vskip .1in

2874: \noindent

2875: {\underline {Claim}} $-$ $y \tau \gamma \beta$ is

2876: in $\lal$.

2877: 

2878: If $y \tau \gamma \geq x$ in $\ak$, then

2879: $x \ \leq \ y \tau \gamma \ \leq \ y$

2880: in $\ak$.

2881: So $y \tau \gamma \beta$ is in $[x \beta, y \beta]$

2882: or

2883: 

2884: $$ y \tau \gamma \beta \ \in \ [x \alpha, y \alpha^{-m-1}]

2885: \  \subset \ \lal.$$

2886: 

2887: 

2888: \noindent

2889: %and in this case $y \tau \gamma \beta \in \lal$.

2890: Notice $x \tau \gamma \beta \in \lal$.

2891: If on the other hand

2892: $y \tau \gamma < x$ in $\ak$, then

2893: $x \tau \gamma < y \tau \gamma < x$ in $\ak$,

2894: and

2895: 

2896: $$y \tau \gamma \beta \ \in \

2897: (x \tau \gamma \beta, x \beta) \ = \ (x \tau \gamma \beta, 

2898: x \alpha) \ \subset \lal$$ 

2899: 

2900: \noindent

2901: and again $y \tau \gamma \beta$

2902: is in $\lal$. 

2903: 

2904: Therefore the claim is proved.

2905: 

2906: \vskip .1in

2907: It now follows that

2908: $y \tau \gamma \beta \alpha^m \ = \ y \alpha \tau$ is

2909: in $\lal$.

2910: If $y \alpha \ \poa \ x$ in 

2911: $\lal$, then $y \alpha > x$ in $\ak$ as well.

2912: Then

2913: $y \alpha \tau > x \tau = y \alpha^{-1}$ in

2914: $\ak$ and $y \alpha \tau$ is not in $\lal$

2915: contradiction.

2916: 

2917: Therefore $y \alpha \ \leq_{\alpha} \ x$ in

2918: $\lal$ and so

2919: $y \alpha$ is in $[x, x \alpha)$.

2920: But $y \alpha^{-2-m} \in [x, x \alpha)$.

2921: Since $y$ is in a local axis for $\alpha$ it follows that

2922: 

2923: $$y \alpha \ = \ y \alpha^{-2-m},

2924: \ \ \ {\rm or} \ \ \ m = -3.$$

2925: 

2926: \noindent

2927: Since we are assuming $m < -3$ this rules out

2928: this case as well.

2929: 

2930: This finishes the analysis of situation I and

2931: completes the analysis of the situation 

2932: orders $<$ and $\pa$ agree on $\ak \cap \ak \alpha$.

2933: 

2934: %\blankfig{07}{1.}{

2935: %The orientation reversing situation,

2936: %\ a. $x \tau > y$, \ b. $x \tau < y$, \ c. $x \tau = y$.}

2937: 

2938: 

2939: 

2940: \begin{figure}

2941: \centeredepsfbox{lam7.eps}

2942: \caption{

2943: The orientation reversing situation,

2944: \ a. $x \tau > y$, \ b. $x \tau < y$, \ c. $x \tau = y$.}

2945: \label{07}

2946: \end{figure}

2947: 

2948: 

2949: \vskip .2in

2950: \noindent

2951: {\bf {Case A.2.2}} $-$ The orders $<$ and $\pa$ 

2952: disagree on $\ak \cap \ak \alpha$.

2953: 

2954: Notice this is equivalent to $y \alpha^{-1} < 

2955: x \alpha^{-1}$ in $\ak$.

2956: Again use $u = \alpha \tau \alpha^{-1} \tau^{-1} 

2957: = \gamma \tau \alpha^{-1} \tau^{-1} \alpha$.

2958: Then 

2959: 

2960: $$\ak u \cap \ak \ = \ (\ak \alpha \tau \alpha^{-1}

2961: \cap \ak) \tau^{-1} \ = \

2962: (\ak \alpha \tau \cap \ak \alpha) \alpha^{-1} \tau^{-1}$$

2963: 

2964: 

2965: There are the following possibilities:

2966: 

2967: %\begin{itemize}

2968: 

2969: %\item

2970: 1) \ If $x \tau > y$ in $\ak$, then

2971: $\ak \alpha \tau \cap \ak \alpha$ is empty

2972: and the bridge from $\ak \alpha \tau$ to

2973: $\ak \alpha$ is $[x \tau, y]$.

2974: Therefore $\ak u \cap \ak = \emptyset$ and

2975: the bridge from $\ak u$ to $\ak$ is

2976: $[x \tau \alpha^{-1} \tau^{-1}, \ y \alpha^{-1} \tau^{-1}]$,

2977: see fig. \ref{07}, a.

2978: 

2979: %\item

2980: 2) \ If $x \tau < y$ in $\ak$, then

2981: $\ak \alpha \tau \cap \ak = [x \tau, y]$.

2982: Hence

2983: $\ak \alpha \tau \alpha^{-1} \cap \ak 

2984: = [y \alpha^{-1}, \ x \tau \alpha^{-1}]$,

2985: where the first endpoint is smaller than

2986: the second in $\ak$. Finally

2987: 

2988: $$\ak u \cap \ak \ = \ [y \alpha^{-1} \tau^{-1}, 

2989: \ x \tau \alpha^{-1} \tau^{-1}]$$

2990: 

2991: \noindent

2992: and the orders $<$, $\pu$ agree on 

2993: $\ak u \cap \ak$, see fig. \ref{07}, b $-$ because

2994: $y \alpha^{-1} < x \alpha^{-1}$ in

2995: $\ak$ and their images under $u$ satisfy

2996: \ $y \tau \alpha^{-1} \tau^{-1} \ \pu \

2997: x \tau \alpha^{-1} \tau^{-1}$ \ in $\ak u$.

2998: 

2999: 

3000: 

3001: %\item

3002: 3) \ Finally if $x \tau = y$, then

3003: $\ak \alpha \tau \cap \ak \alpha

3004: = [y, \ v]$, where \

3005: $v \ \leq_{\alpha} \ y$ \ in $\ak \alpha$.

3006: It follows that the intersection

3007: $\ak \alpha \tau \alpha^{-1} \cap \ak 

3008: = [v \alpha^{-1}, y \alpha^{-1}]$,

3009: the first point precedes in $\ak$.

3010: And then

3011: 

3012: $$ \ak u \cap \ak \ = \ 

3013: [ v \alpha^{-1} \tau^{-1}, \ y \alpha^{-1} \tau^{-1}]

3014: \ = \ 

3015: [t, \ y \alpha^{-1} \tau^{-1}].$$

3016: 

3017: \noindent

3018: Here if $t$ is not $y \alpha^{-1} \tau^{-1}$ then

3019: $<$ and $\pu$ disagree on $\ak u \cap \ak$

3020: $-$ because $y \tau^{-1} \alpha^{-1} \leq

3021: v \tau^{-1} \alpha^{-1}$ in $\ak$.

3022: 

3023: %\end{itemize}

3024: 

3025: \vskip .1in

3026: 

3027: Now use $\ak u \cap \ak

3028: \ = \ (\ak \alpha^{-1} \tau^{-1} \cap \ak \alpha^{-1})

3029: \alpha$.

3030: Here $\ak \alpha^{-1} \cap \ak = 

3031: [y \alpha^{-1}, \ x \alpha^{-1}]$ the first term

3032: precedes in $\ak$.

3033: Again there are 3 possibilities

3034: 

3035: %\blankfig{08}{1.}{

3036: %Using 

3037: %$\ak u = \ak \alpha \tau \alpha^{-1} \tau^{-1}$:

3038: %\ a. $x \alpha^{-1} \tau^{-1} < y \alpha^{-1}$,

3039: %\ b. $x \alpha^{-1} \tau^{-1} > y \alpha^{-1}$,

3040: %\ a. $x \alpha^{-1} \tau^{-1} = y \alpha^{-1}$.}

3041: 

3042: 

3043: 

3044: \begin{figure}

3045: \centeredepsfbox{lam8.eps}

3046: \caption{

3047: Using 

3048: $\ak u = \ak \alpha \tau \alpha^{-1} \tau^{-1}$:

3049: \ a. $x \alpha^{-1} \tau^{-1} < y \alpha^{-1}$,

3050: \ b. $x \alpha^{-1} \tau^{-1} > y \alpha^{-1}$,

3051: \ a. $x \alpha^{-1} \tau^{-1} = y \alpha^{-1}$.}

3052: \label{08}

3053: \end{figure}

3054: 

3055: 

3056: 

3057: 

3058: %\begin{itemize}

3059: 

3060: %\item

3061: 1') \ If $x \alpha^{-1} \tau^{-1} < y \alpha^{-1}$ in $\ak$

3062: then

3063: $\ak \alpha^{-1} \tau^{-1} \cap \ak \alpha^{-1} =

3064: \emptyset$ and the bridge from

3065: $\ak \alpha^{-1} \tau^{-1}$ to $\ak \alpha^{-1}$ 

3066: is $[x \alpha^{-1} \tau^{-1}, \ y \alpha^{-1}]$.

3067: Hence $\ak u \cap \ak = \emptyset$ and the 

3068: bridge from $\ak u$ to $\ak$ is

3069: $[x \alpha^{-1} \tau^{-1} \alpha, y]$, see fig.

3070: \ref{08}, a.

3071: 

3072: %\item

3073: 2') \ If $x \alpha^{-1} \tau^{-1} > y \alpha^{-1}$ in

3074: $\ak$, then

3075: $\ak \alpha^{-1} \tau^{-1} \cap \ak \alpha^{-1}

3076: = [y \alpha^{-1}, \ x \alpha^{-1} \tau^{-1}]$

3077: and hence 

3078: $\ak u \cap \ak = [x \alpha^{-1} \tau^{-1} \alpha, y]$

3079: and the orders $<$ and $\pu$ agree on $\ak \cap \ak u$,

3080: see fig. \ref{08}, b $-$ because

3081: $x < y$ in $\ak$ and

3082: \ $x \alpha^{-1} \tau^{-1} \alpha  \   \pu \

3083: y \alpha^{-1} \tau^{-1} \alpha$ \ in $\ak u$.

3084: 

3085: %\item

3086: 3') \ If $x \alpha^{-1} \tau^{-1} = y$, then

3087: $\ak \alpha^{-1} \tau^{-1} \cap \ak \alpha^{-1}

3088: = [c, \ y \alpha^{-1}]$ and

3089: $\ak u \cap \ak \ = \ [y, z]$ where $z = c \alpha^{-1}$.

3090: If $z$ is not equal to $y$, then the

3091: orders $<$ and $\pu$ disagree on $\ak u \cap \ak$.

3092: 

3093: %\end{itemize}

3094: 

3095: \vskip .1in

3096: Notice both pairs of 3 alternatives are all

3097: mutually exclusive. We match them and obtain 3 possible

3098: situations:

3099: 

3100: 

3101: %\begin{itemize}

3102: 

3103: %\item

3104: {\bf I} $-$ 

3105: $x \tau > y$ in $\ak$,

3106: $x \alpha^{-1} \tau^{-1} < y \alpha^{-1}$ in $\ak$

3107: and

3108: 

3109: $$\ak u \cap \ak  \ = \ \emptyset, \ \ \ \

3110: [y \alpha^{-1} \tau^{-1}, \ x \tau \alpha^{-1} \tau^{-1}] \ = \ 

3111: [y, \ x \alpha^{-1} \tau^{-1} \alpha].$$

3112: 

3113: %\item

3114: {\bf II} $-$ 

3115: $x \tau < y$ in $\ak$, $x \alpha^{-1} \tau^{-1} > y \alpha^{-1}$

3116: in $\ak$,

3117: 

3118: $$ \ak u \cap \ak \ = \ 

3119: [y \alpha^{-1} \tau^{-1}, \ x \tau \alpha^{-1} \tau^{-1}]

3120: \ = \ [x \alpha^{-1} \tau^{-1} \alpha, \ y]$$

3121: 

3122: \noindent

3123: and the orders $<$, $\pu$ agree on $\ak u \cap \ak$.

3124: 

3125: %\item

3126: {\bf III} $-$

3127: $x \tau = y$, \ $x \alpha^{-1} \tau^{-1} = y \alpha^{-1}$

3128: and

3129: 

3130: $$ \ak u \cap \ak \ = \

3131: [y, z] \ = \ [t, y \alpha^{-1} \tau^{-1}].$$

3132: 

3133: \noindent

3134: If $z$ is not $y$ then the orders

3135: $<$, $\pu$ disagree on $\ak u \cap \ak$.

3136: 

3137: %\end{itemize}

3138: 

3139: \vskip .1in

3140: We analyse each case in turn:

3141: 

3142: \vskip .1in

3143: \noindent

3144: {\bf {Situation II}} $-$

3145: 

3146: 

3147: Here $x \tau < y$, \ $x \alpha^{-1} \tau^{-1} > y \alpha^{-1}$

3148: and 

3149: 

3150: $$ y \ = \ x \tau \alpha^{-1} \tau^{-1}, \ \ \ \ 

3151: y \alpha^{-1} \tau^{-1} \ = \ x \alpha^{-1} \tau^{-1} \alpha.$$

3152: 

3153: \noindent

3154: Suppose first that $[y \alpha^{-1}, x \alpha^{-1}]

3155: \cap [x, y] = \emptyset$.

3156: %, see fig. \ref{09}, a.

3157: Since $y \tau  = x \tau \alpha^{-1}$ then

3158: $[y \alpha^{-1}, x \alpha^{-1}]$ is contained

3159: in the set of points $> y$ in $\ak$.

3160: 

3161: In addition

3162: $y \alpha$ is in $\ak \alpha - \ak$ and

3163: $y \pa y \alpha$.

3164: Hence $y$ is in $(y \alpha^{-1}, y \alpha)$, 

3165: producing a local axis $\lal$ of $\alpha$ which

3166: contains $y$.

3167: Now use $\tau^{-1} \alpha \tau = \alpha \tau \alpha^{-1}

3168: \tau^{-1} \alpha^{m-1}$ applied to $x \alpha^{-1}$:

3169: 

3170: $$x \alpha^{-1} \tau^{-1} \alpha \tau

3171: \ = \ x \alpha^{-1} \alpha \tau \alpha^{-1}

3172: \tau^{-1} \alpha^{m-1} \ = \

3173: x \tau \alpha^{-1} \tau^{-1} \alpha^{m-1}$$

3174: 

3175: \noindent

3176: Substitute 

3177: $x \tau \alpha^{-1} \tau^{-1} = y$ in the last

3178: term and $x \alpha^{-1} \tau^{-1} \alpha = y

3179: \alpha^{-1} \tau^{-1}$ in the first term to get

3180: 

3181: $$(y \alpha^{-1} \tau^{-1}) \tau \ = \ y \alpha^{-1} 

3182: \ = \  y \alpha^{m-1}$$

3183: 

3184: \noindent

3185: or $y = y \alpha^{m}$. This is impossible because

3186: $y$ is in a local axis of $\alpha$ and $m$ is not zero.

3187: 

3188: 

3189: \vskip .1in

3190: From now on in situation II suppose that

3191: $[y \alpha^{-1}, x \alpha^{-1}] \cap [x, y]$ is not

3192: empty.

3193: Since $x \tau \alpha^{-1} = y \tau > y$ in $\ak$,

3194: then $x \alpha^{-1} > y$ in $\ak$.

3195: It follows that $y \alpha^{-1} \leq y$ in $\ak$.

3196: 

3197: %\blankfig{09}{1.}{a. Situation II, case 

3198: %$\alpha$ has local axis, b. Case $\alpha$ has a fixed

3199: %point in the axis of $\tau$.}

3200: 

3201: Suppose first that $y \alpha^{-1} < y$ in $\ak$.

3202: There is $r$ in $[y \alpha^{-1}, y]$ which is

3203: fixed by $\alpha$.

3204: Either $r$ is equal to $y$ or $r < y$ in $\ak$.

3205: Let $\uu_1$ (respectively $\uu_2$)

3206: be the component of $T - \{ r \}$ containing

3207: $r \tau$  (respectively $r \tau^{-1}$).

3208: Since

3209: 

3210: $$x \alpha^{-1} \in \uu_1, \ \ 

3211: x \in \uu_2 \ \ \  \ {\rm then} \ \ \ 

3212: \uu_1 \alpha \ = \ \uu_2.$$

3213: 

3214: \noindent

3215: If $r < y$ in $\ak$ then also we have $\uu_2 \alpha = \uu_1$.

3216: Otherwise $\uu_2 \alpha = \uu_3$ which is

3217: another component of $T - \{ r \}$ which is

3218: not $\uu_1, \uu_2$.

3219: We will rule out this case, but the result will be

3220: used later on as well, so we state it in more generality:

3221: 

3222: \begin{lemma}{}{}

3223: Let $\lat$ be a local axis for $\tau$. Let $r$ in $\lat$ which

3224: is fixed by $\alpha$. Let $\uu_1$ (respectively $\uu_2$

3225: be the component of $T - \{ r \}$ containing

3226: $r \tau$ (respectively $r \tau^{-1}$). 

3227: Then $\uu_1 \alpha$ is not $\uu_2$ and

3228: $\uu_2 \alpha$ is not $\uu_1$.

3229: \label{turn}

3230: \end{lemma}

3231: 

3232: \begin{proof}{}

3233: The proof is as follows:

3234: suppose that either

3235: $\uu_1 \alpha = \uu_2$ or $\uu_2 \alpha = \uu_1$ and

3236: arrive at a contradiction.

3237: 

3238: 

3239: First assume that $\uu_1 \alpha = \uu_2$.

3240: Either $\uu_2 \alpha = \uu_1$ or $\uu_2 \alpha$ is another

3241: component $\uu_3$ of $T - \{ u \}$.

3242: 

3243: 

3244: Let $\vv_i = \uu_i \tau^{-1}$.

3245: Since 

3246: $\vv_1 \beta \ = \ \vv_1 \tau \alpha^{-1} \tau^{-1} \ = \

3247: \uu_1 \alpha^{-1} \tau^{-1} \ \not = \ \vv_1$,

3248:  we have

3249: that $\vv_1 \beta$ is contained in $\uu_2$.

3250: Therefore $r \beta $ is in $\uu_2$ and

3251: $r \beta \alpha^{m-1}$ is in 

3252: $\uu_2 \alpha^{m-1}$.

3253: Also

3254: 

3255: $$r \tau^{-1} \alpha \tau \ = \ 

3256: r \alpha \beta \alpha^{m-1} \ = \ 

3257: r \beta \alpha^{m-1}$$

3258: 

3259: \noindent

3260: As $r \tau^{-1} \in \uu_2$ then $r \tau^{-1} \alpha$ is

3261: in $\uu_2 \alpha$, which is either $\uu_1$ or $\uu_3$.

3262: Therefore $r \tau^{-1} \alpha \tau$ is either

3263: in $\uu_1 \tau \subset \uu_1$ or in $\uu_3 \tau$ again

3264: a subset of $\uu_1$. So $r \tau^{-1} \alpha \tau \in \uu_1$.

3265: Therefore $\uu_2 \alpha^{m-1} \cap \uu_1 

3266: \not = \emptyset$. But both are components of 

3267: $T - \{ r \}$, because $r \alpha = r$,

3268:  so it follows that they are equal. 

3269: As $\uu_2 = \uu_1 \alpha$ then

3270: 

3271: $$\uu_1 \alpha \alpha^{m-1} \ = \ \uu_1,

3272: \ \ \ {\rm or} \ \ \ \uu_1 \alpha^{m} = \uu_1,

3273: \ \ \uu_2 \alpha^m = \uu_2, 

3274: \ \ \uu_3 \alpha^m = \uu_3 \ \ \ {\rm if \ needed}.$$

3275: 

3276: \noindent

3277: In case $r \not = y$ this immediately implies

3278: $m$ even.

3279: 

3280: Now use $r \tau \gamma \beta \alpha^m = r \alpha \tau = r \tau 

3281: \in \uu_1$.

3282: Therefore $r \tau \gamma \beta \ \in \ \uu_1 \alpha^m = \uu_1$.

3283: It follows that

3284: 

3285: $$r \tau^{-1} \prec r \prec r \tau \gamma \beta$$

3286: 

3287: \noindent

3288: $-$ recall this means $r$ separates $r \tau^{-1}$

3289: from $r \tau \gamma \beta$.

3290: Applying $\beta^{-1}$ one gets

3291: 

3292: $$r \tau^{-1} \prec r \beta^{-1} \prec r \tau \gamma 

3293: \ \ \ \ (*)$$

3294: 

3295: \noindent

3296: Use $r \beta^{-1} = r \tau \alpha \tau^{-1}$:

3297: 

3298: $$r \tau \in \uu_1 \ \Rightarrow \ r \tau \alpha \in \uu_2, 

3299: \ \ \ r \beta^{-1} = r \tau \alpha^{-1} \tau^{-1} \in \vv_2.$$

3300: 

3301: \noindent

3302: As $r \tau^{-1}$ is an accumulation point of $\vv_2$,

3303: equation (*) above implies that

3304: $r \tau \gamma$ is in $\vv_2$ or

3305: $r \tau \gamma < r \tau^{-1}$ in $\ak$, which immediately

3306: implies $p > 2q$.

3307: % < r \tau^{-1}$

3308: %then $r \tau \gamma < u \tau^{-1}$ which implies

3309: %$p > 2q$.

3310: %Here use $r \tau \alpha \in \uu_2$.

3311: 

3312: As in the $\rrrr$-covered

3313: case, look at $r \tau \alpha$. If $r \tau \alpha$ is

3314: not in $\vv_2$ then

3315: $r \tau \alpha \tau \not \in \uu_2$ so

3316: 

3317: $$r \tau \alpha \tau \ = \ 

3318: (r \tau^2) \tau^{-1} \alpha \tau \ = \ 

3319: (r \tau^2) \gamma \beta \alpha^m \ \ \not \in \uu_2

3320: \ \ \ {\rm and} \ \ \ r \tau \gamma \beta \ \not \in \ \uu_2.$$

3321: 

3322: \noindent

3323: So 

3324: $r \tau^{-1} \prec r \preceq r \tau^2 \gamma \beta$ \ and \

3325: $r \tau^2 \gamma \preceq r \beta^{-1} \prec r \tau^{-1}$. \ 

3326: As $r \beta^{-1}  = r \tau \alpha \tau^{-1} \in \vv_2$,

3327: then 

3328: 

3329: $$r \tau^2 \gamma \ \in \ \vv_2, \ \ {\rm so} \ \ 

3330: r \tau^2 \gamma < r \tau^{-1} \ \ \ {\rm in } \ \ak.$$

3331: 

3332: \noindent

3333: As seen before this implies $p > 3q$, which is disallowed

3334: and finishes this case.

3335: 

3336: If $r \tau \alpha \in \vv_2$ then

3337: $r \beta^{-1} \in \vv_2 \tau^{-1}$.

3338: By $(*)$ 

3339: $r \tau^{-1} \prec r \beta^{-1} \prec r \tau \gamma$, so

3340: %$r \tau \gamma \prec r \beta^{-1} \prec r \tau^{-1}$

3341: %then

3342: 

3343: $$r \tau \gamma \in \vv_2 \tau^{-1} \ \ 

3344: \Rightarrow \ \ r \tau \gamma < r \tau^{-2} 

3345:  \ \ {\rm in } \ \ \ak.$$

3346: 

3347: \noindent

3348: %But $r \beta^{-1} \in \vv_2 \tau^{-1}$,

3349: %so $r \tau^{-1} \prec r \beta^{-1} \prec z \tau \gamma$.

3350: As seen before this also implies $p > 3q$ contradiction.

3351: 

3352: This finishes the analysis of the case $\uu_1 \alpha = \uu_2$.

3353: 

3354: \vskip .1in

3355: Now suppose that $\uu_2 \alpha = \uu_1$.

3356: If $\uu_1 \alpha = \uu_2$, then this is taken care by the

3357: previous situation. So now assume $\uu_2 \alpha^{-1} = 

3358: \uu_3$ which is not $\uu_1$ or $\uu_2$.

3359: 

3360: 

3361: Here use $r \tau^{-1} \alpha \tau = r \alpha \beta \alpha^{m-1}

3362: = r \tau \alpha^{-1} \tau^{-1} \alpha^{m-1}$.

3363: First 

3364: 

3365: $$r \tau^{-1} \ \in \ \uu_2 \ \ \ \Rightarrow \ \ \ 

3366: r \tau^{-1} \alpha \ \in \ \uu_2 \alpha \ = \ \uu_1 

3367: \ \ \ \Rightarrow \ \ \ r \tau^{-1} \alpha \tau \ \in \ \uu_1.$$

3368: 

3369: \noindent

3370: On the other hand

3371: 

3372: $$r \tau \ \in \ \uu_1 \ \ \Rightarrow \ \ 

3373: r \tau \alpha^{-1} \ \in \ \uu_1 \alpha^{-1}  = \uu_2

3374: \ \ \Rightarrow \ \ 

3375: r \tau \alpha^{-1} \tau^{-1} \ \in \ \uu_2 \tau^{-1} \subset \uu_2

3376: \ \ \Rightarrow \ \ 

3377: r \tau \alpha^{-1} \tau^{-1} \alpha^{m-1} \ \in \ 

3378: \uu_2 \alpha^{m-1}.$$

3379: 

3380: \noindent

3381: From which we conclude that \ $\uu_2 \alpha^{m-1} = \uu_1

3382: = \uu_2 \alpha$.

3383: % or $\uu_2 \alpha^m = \uu_2 \alpha^2$.

3384: 

3385: Now use $r \tau^{-1} \alpha \tau = r \gamma \beta \alpha^m$.

3386: The right side is in $\uu_1 = \uu_2 \alpha$.

3387: The fact that $\uu_2 \alpha^{-1}$ is not $\uu_1$ implies

3388: that $\vv_2 \beta$ is not $\vv_1$, hence 

3389: $\vv_2 \beta$ is contained in $\uu_2$.

3390: %Check the right side of the equation.

3391: We know that $r \gamma$ is $\leq r \tau^{-1}$ in $\lat$ so

3392: it is either in $\vv_2$ or is equal to $r \tau^{-1}$.

3393: Hence $r \gamma \beta$ is either $r \tau^{-1}$ or is

3394: in $\vv_2 \beta$ $-$ in either case it is in $\uu_2$.

3395: Finally $r \gamma \beta \alpha^m$ is in $\uu_2 \alpha^m$

3396: which must be $\uu_1$. But then $\uu_2 \alpha^m = \uu_2

3397: \alpha^{m-1}$ contradiction.

3398: 

3399: This finishes the analysis of the case $\uu_2 \alpha = \uu_1$

3400: and so 

3401: finishes the

3402: proof of lemma \ref{turn}.

3403: \end{proof}

3404: 

3405: This finishes the analysis of situation II.

3406: 

3407: \vskip .2in

3408: \noindent

3409: {\bf {Situation I}} $-$

3410: 

3411: 

3412: 

3413: In this case 

3414: $x \alpha^{-1} \tau^{-1} < y \alpha^{-1}$ in 

3415: $\ak$ and $y < x \tau$ in $\ak$. In addition

3416: 

3417: $$y \tau \ = \ y \alpha^{-1}, \ \ \ \ \ \ \ \

3418: x \alpha^{-1} \tau^{-1} \alpha \ = \ 

3419: x \tau \alpha^{-1} \tau^{-1} \ \ \ \ \ (*) $$

3420: %

3421: %

3422: %\noindent

3423: %Also

3424: %$$s \ \prec_{\alpha} \ r \alpha^{-1} \tau^{-1} \alpha,

3425: %\ \ \ \ s \ \prec \ r \tau \ \ \ \ {\rm and  \ so } \ \ \

3426: %r  \alpha^{-1} \tau^{-1} \ \prec \ s \alpha^{-1} = 

3427: %s \tau \ \ \ \ (1.4)$$

3428: 

3429: \noindent

3430: Here

3431: $x \alpha^{-1} \ > \ y \alpha^{-1} = y \tau$ in $\ak$

3432: (orientation

3433: reversing case) so $x \alpha^{-1} \tau^{-1}  \ > \ y$

3434: in $\ak$.

3435: Therefore $x \alpha^{-1} \tau^{-1} \ \in \ 

3436: (y, y \alpha^{-1})$.

3437: Also $x \tau \ < \ y \tau = y \alpha^{-1}$ in $\ak$,

3438: so one concludes

3439: 

3440: $$x \alpha^{-1} \tau^{-1}, \ \ \ 

3441: x \tau \ \in \ (y, y \alpha^{-1}) $$

3442: 

3443: 

3444: 

3445: 

3446: On the other hand \ $y \prec y \alpha^{-1} \prec x \alpha^{-1}$,

3447: so $y \alpha \prec y \prec x$ and $y \alpha$ is

3448: in $\ak \alpha - \ak$. It follows that

3449: $y \alpha^{-1} \prec y \prec y \alpha$ and

3450: $y$ is in a local axis $\lal$ for $\alpha$.

3451: This implies that the translates $[y \alpha^i, y \alpha^{i+1})$

3452: are all disjoint (as $i$ varies in ${\bf Z}$).

3453: %$[s, s \alpha^{-1}]  \cap  [s \alpha, s] \ =

3454: %\ \{ s \}$, with the usual conclusions of lemma

3455: %4.0.3 that

3456: %

3457: %$$[s, s \alpha^{-1}] \ \cap \ [s, s \alpha^{-1}] \alpha^n

3458: %\ = \  \emptyset 

3459: %\ \ \ {\rm if} \ \ \  |n| \geq 2 \ \ \ \ (1.6)$$ 

3460: Use the relation

3461: $\tau^{-1} \alpha \tau \ = \ 

3462: \alpha \tau \alpha^{-1} \tau^{-1} \alpha^{m-1}$

3463: in the form

3464: 

3465: $$ \alpha^{-1} \tau^{-1} \alpha \tau \alpha^{1-m} 

3466: \ = \ \tau \alpha^{-1} \tau^{-1} $$

3467: 

3468: \noindent 

3469: applied to $x$ to get

3470: 

3471: $$ (x \alpha^{-1} \tau^{-1} \alpha) \tau \alpha^{1-m} 

3472: \ = \  x \tau \alpha^{-1} \tau^{-1} \ \ \ \ \ \ (**)$$

3473: 

3474:  

3475: \noindent

3476: Now apply the second equality of $(*) $

3477: {\underline {both}} sides of $(**)$ to get

3478: 

3479: $$ (x \tau \alpha^{-1} \tau^{-1}) \tau \alpha^{1-m} \ = \ 

3480: x \alpha^{-1} \tau^{-1} \alpha \ \ \ \ \ \ \ 

3481: {\rm or }  \ \ \ \ \ \ \ 

3482: (x \tau) \alpha^{-m} \ = \ (x \alpha^{-1} \tau^{-1}) \alpha.$$

3483: 

3484: 

3485: \noindent

3486: But \ $x \tau \in (y, y \alpha^{-1})$,  \ so

3487: $x \tau \alpha^{-m} \in

3488: (y, y \alpha^{-1}) \alpha^{-m}$. 

3489: Similarly 

3490: $x \alpha^{-1} \tau^{-1} \alpha$ is in $(y, y \alpha^{-1}) \alpha$.

3491: Since they are equal then $-m = 1$ or

3492: $m = -1$, impossible.

3493: 

3494: 

3495: 

3496: 

3497: \vskip .2in

3498: \noindent

3499: {\bf {Situation III}} $-$ 

3500: 

3501: Here $x \tau = y$, $x \alpha^{-1} \tau^{-1} = y \alpha^{-1}$

3502: and 

3503: 

3504: $$\ak u \cap \ak \ = \ [y,z] \ = \ [t, y \alpha^{-1} \tau^{-1}]$$

3505: 

3506: \noindent

3507: and if $t \not = y$, then

3508: $<$, $\pu$ disagree on $\ak u \cap \ak$.

3509: 

3510: Notice that $y \leq z = y \alpha^{-1} \tau^{-1}$

3511: so $y < y \alpha^{-1}$ in $\ak$,

3512: and $y \alpha^{-1}$ is in $\ak - \ak \alpha$.

3513: Also $y \tau \leq y \alpha^{-1}$ in $\ak$.

3514: Now 

3515: 

3516: $$y \ \prec \ y \alpha^{-1} \ \prec \ x \alpha^{-1}

3517: \ \ \ \Rightarrow \ \ \ 

3518: x \ \prec \ y \ \prec \ y \alpha \ \ \ {\rm all \ in }

3519: \ \ \ak \alpha.$$

3520: %Apply $\alpha$ to obtain \ 

3521: 

3522: \noindent

3523: Hence 

3524: $y \alpha \ \pu \ y$ in $\ak$ and

3525: $y \alpha$ is in $\ak \alpha - \ak$.

3526: %, see fig.

3527: %\ref{09}, c.

3528: Hence $y$ is in $(y \alpha^{-1}, y \alpha)$ and

3529: there is a local axis $\lal$ of $\alpha$ with

3530: $y$ in $\lal$.

3531: Consider the relation $\tau^{-1} \alpha \tau = \alpha \beta

3532: \alpha^{m-1}$. Substitute $\beta = \tau \alpha^{-1} \tau^{-1}$

3533: and rearrange the terms to get

3534: $\alpha^{-1} \tau^{-1} \alpha =

3535: \tau \alpha^{-1} \tau^{-1} \alpha^{m-1} \tau^{-1}$.

3536: Now apply it to

3537: $x$:

3538: 

3539: $$y = x \alpha^{-1} \tau^{-1} \alpha \ = \ 

3540: x \tau \alpha^{-1} \tau^{-1} \alpha^{m-1} \tau^{-1},$$

3541: 

3542: \noindent

3543: or $y \tau \alpha^{1-m} = y \alpha^{-1} \tau^{-1}$.

3544: Now $y \tau \in [y, \ y \alpha^{-1}]$, so

3545: $y \tau$ is in $\lal$ and

3546: 

3547: $$y \tau \alpha^{1-m} \ \in \ [y \alpha^{1-m}, \ y \alpha^{-m}],$$

3548: 

3549: \noindent

3550: so $y \tau \alpha^{1-m}$ is not in $\ak$.

3551: But $y \alpha^{-1} \tau^{-1}$ is in $\ak$, contradiction.

3552: 

3553: This finishes the analysis of $\ak u \cap \ak = [x,y]$

3554: with $x$ not equal $y$. Consequently this finishes

3555: the analysis of Case A, $\tau$ acts freely,

3556: which we now proved cannot happen.

3557: 

3558: 

3559: 

3560: \vskip .2in

3561: 

3562: \section{Case B $-$ $\tau$ has a fixed point, 

3563: $\alpha$ acts freely}

3564: \label{caseb}

3565: 

3566: Here $\alpha$ has an (actual) axis $\ap$

3567: and so does $\beta$ with axis $\ab = \ap \tau^{-1}$.

3568: Let $Fix(\tau)$ be the set of fixed points

3569: of $\tau$.

3570: As usual there are various possibilities.

3571: This case is very interesting because the topology

3572: of the manifold $M_{p/q}$ will play a key role.

3573: 

3574: Recall that if $t$ is a point not in a connected set

3575: $B$ of the tree $T$, then the segment $[t,u]$ is

3576: the bridge from $t$ to $B$ if the subsegment

3577: $[x,u)$ does not intersect $B$ and

3578: if $u$ is either in $B$ or is an accumulation point

3579: of $B$. Again the important fact is that

3580: the bridge from $x$ to $B$ is unique:

3581: it is the only embedded path from $x$ to $B$ because

3582: $T$ is a tree. As in case $A$ this will be

3583: explored here. If $u$ is in $B$ we say that

3584: $t$ bridges to $u$ in $B$.

3585: 

3586: We say that a point $a$ is an ideal point of a local

3587: axis $l$ if $a$ is not in $l$ but is an accumulation

3588: point of $l$. Obviously this implies that $l$ is

3589: not properly embedded in $T$ in the side accumulating

3590: to $a$.

3591: 

3592: There are two main cases depending on whether $Fix(\tau)$

3593: intersects $\ap$ or not.

3594: 

3595: 

3596: \vskip .2in

3597: \noindent

3598: {\bf {Case B.1}} $-$ $Fix(\tau) \cap \ap = \emptyset$.

3599: 

3600: Then $\kappa$ also has a fixed point $s$. 

3601: Choose $s$ closest to $\ap$, that is,

3602: the bridge $[s,c]$ from

3603: $s$ to $\ap$ has no other fixed point of $\kappa$.

3604: Let $z$ in $[s,c]$ fixed by $\tau$

3605: and closest to $\ap$, that is,

3606: the bridge $[z,c]$ from $z$ to $\ap$ has no other

3607: fixed point of $\tau$ besides $z$.

3608: A priori we do not know whether $z$ is equal to $s$ or

3609: not.

3610: Let $\uu$ be the component of $T - \{ z \}$ containing

3611: $\ap$.

3612: 

3613: 

3614: Then $\ab$ is a subset of $\uu \tau \not = \uu$ and

3615: $z$ bridges to $c \tau^{-1}$ in $\ab$.

3616: 

3617: 

3618: %\blankfig{09}{1.}{

3619: %a. The case $\uu \tau \not = \uu$, \

3620: %b. The case $\lat \cap \ap = \emptyset$.}

3621: 

3622: 

3623: 

3624: \begin{figure}

3625: \centeredepsfbox{lam9.eps}

3626: \caption{

3627: a. The case $\uu \tau \not = \uu$, \

3628: b. The case $\lat \cap \ap = \emptyset$.}

3629: \label{09}

3630: \end{figure}

3631: 

3632: 

3633: 

3634: 

3635: 

3636: %, \ c.

3637: %When $\uu \gamma \not = \uu$ then $\uu \gamma \cap \uu

3638: %= \emptyset$.}

3639: 

3640: 

3641: \vskip .2in

3642: \noindent

3643: {\bf {Case B.1.1}} $-$ Suppose $\uu \tau \not = \uu$.

3644: 

3645: Then $\uu \tau^{-1} \not = \uu$ as well.

3646: Apply \ $\alpha \tau = \tau \alpha \beta \alpha^{m-1}$ \ to

3647: $z$: the point 

3648: $z$ bridges to $c$ in $\ap$, so $z \alpha$ bridges

3649: to $c \alpha$ in $\ap$. As $c \alpha$ is not $c$ then

3650: $z \alpha$ is in $\uu$, so $z \alpha \tau$ is in

3651: $\uu \tau$, see fig. \ref{09}, a.

3652: On the other hand $z \tau \alpha = z \alpha$ is in

3653: $\uu$ and hence $z$ separates

3654: it from $\ab$. It follows that $z \alpha$ also

3655: bridges to $c \tau^{-1}$ in $\ab$.

3656: Then 

3657: 

3658: $$z \alpha \tau \beta \ \ \ {\rm bridges \ to} \ \ \ 

3659: c \tau^{-1} \beta \ \ {\rm in} \ \ 

3660: \ab \ \ \ {\rm and} \ \ \  c \tau^{-1} \beta \not = 

3661: c \tau^{-1}, \ \ \ {\rm so} \ \ \  z \tau \alpha \beta \ \in \

3662: \uu \tau^{-1}.$$

3663: 

3664: \noindent

3665: Therefore $z \tau \alpha \beta$ bridges

3666: to $c$ in $\ap$, so  $z \tau \alpha \beta \alpha^{m-1}$

3667: bridges to $c \alpha^{m-1}$ in $\ap$. This implies 

3668: $z \tau \alpha \beta \alpha^{m-1}$ is in $\uu$,

3669: impossible since it is equal to $z \alpha \tau \in \uu \tau$.

3670: 

3671: \vskip .1in

3672: We conclude that $\uu \tau = \uu$, which will be assumed

3673: from now on in this proof.

3674: 

3675: Choose a prong $\eta$ at $z$ which is a subset

3676: of $[z, c]$.

3677: This prong is associated to the component $\uu$ of

3678: $T - \{ z \}$, hence the prong $\eta \tau$ also is 

3679: associated to the component $\uu = \uu \tau$ and

3680: $\eta \cap \eta \tau$ is not just $z$. Let $e$ be

3681: another point in the intersection. Then $e \tau^{-1}, e$

3682: are both in $\eta$ and $e \tau^{-1}$ is not equal $e$ $-$ 

3683: by choice of $z$ as the fixed point of $\tau$ 

3684: in $[z,c]$ closest

3685: to $\ap$.

3686: So either $e$ is in $[z, \ e \tau)$ or

3687: $e \tau$ is in $[z, e)$. In the first case (say) apply

3688: $\tau$ to get $e \tau$ is in $[z, \ e \tau^2)$ and

3689: it now follows that $e \prec e \tau \prec e \tau^2$.

3690: The same alignment of points

3691:  happens in the second case. We conclude that

3692: there is a local axis $\lat$ for $\tau$, with $e$ in 

3693: the local axis.

3694: 

3695: This construction of a local axis is crucial in case 

3696: B and also in case C of the proof.

3697: 

3698: \vskip .1in

3699: \noindent

3700: {\bf {Conclusion}} $-$ If $\uu \tau = \uu$ and there is

3701: no fixed point of $\tau$ in $(z,w]$,

3702: then there is a

3703:  local axis of $\tau$ contained in $\uu$ with one

3704: ideal point $z$.

3705: 

3706: 

3707: \vskip .2in

3708: \noindent

3709: {\bf {Case B.1.2}} $-$ Suppose that $\lat \cap \ap$

3710: is at most one point.

3711: 

3712: Let $[d,c]$ be the bridge from 

3713: $\lat$ to $\ap$ $-$ here $d = c$ if

3714: $\lat \cap \ap$ is a single point.

3715: We do the proof for $\lat \cap \ap = \emptyset$, the

3716: case of single point intersection being entirely

3717: similar.

3718: The bridge from $c \alpha \tau$ to $\lat$ is

3719: $[c \alpha \tau, d \tau]$, see fig. \ref{09}, b.

3720: Now the bridge from $c \tau \gamma$ to $\lat$ is

3721: $[c \tau \gamma, d \tau \gamma]$. Here use 

3722: $p$ odd to get

3723: $d \tau \gamma \not = d \tau^{-1}$, so the

3724: bridge from $c \tau \gamma$ to $\ab$ is

3725: $[c \tau \gamma, \ c \tau^{-1}]$.

3726: Therefore 

3727: 

3728: $$c \tau \gamma \beta \ \  \ {\rm bridges \ to} \ \

3729: c \tau^{-1} \beta \ \ {\rm in } \ \ \ab,

3730: \ \ {\rm hence \ bridges \ to} \ \ 

3731:  d \tau^{-1} \ \ \ {\rm in} \ \  \lat

3732: \ \ \ {\rm and \ to}  \ \ c \ \ \ {\rm in} \ \

3733: \ap.$$

3734: 

3735: 

3736: \noindent

3737: Finally 

3738: $c \tau \gamma \beta \alpha^m$ bridges to 

3739: $\ap$ in $c \alpha^{m}

3740: \not = c$ and so bridges to $\lat$ in $c$.

3741: 

3742: As $c \alpha \tau = c \tau \gamma \beta \alpha^m$,

3743: this implies $c = c \tau$, impossible.

3744: This rules out this case.

3745: 

3746: \vskip .1in

3747: We conclude that $\lat \cap \ap$ is more than one point.

3748: If $\lat \cap \ap$ is $(z,d]$, then either

3749: $z \alpha = z$ or $\alpha$ has a fixed point

3750: in $\lat$, both impossible.

3751: Therefore from now on in 

3752: case B.1 let 

3753: $\lat \cap \ap = [a, b]$, with $a \not = z$ and

3754: $a$ closest to $z$.

3755: By an abuse of notation $b$ can be $+\infty$, meaning

3756: the intersection is a ray in $\lat$.

3757: Put an order $<$ in $\lat$ so that $a < b$ in $\lat$.

3758: Also let $\pa$ be the order in $\ap$ with 

3759: \ $a  \ \pa  \ b$.

3760: 

3761: 

3762: 

3763: 

3764: 

3765: 

3766: 

3767: From now on in case B.1 the proof will depend on

3768: whether $\uu \gamma$ is equal to $\uu$ or not.

3769: The arguments here are also very similar to

3770: what will be needed for case C, therefore we will make

3771: the arguments in more generality so that they can

3772: be used in case C,

3773: namely when $\alpha$ has

3774: a fixed point but has a local axis with

3775: certain properties. 

3776: We first specify the conditions under which 

3777: the analysis works.

3778: 

3779: 

3780: \vskip .1in

3781: \noindent

3782: {\bf {Conditions}} $-$ Consider two conditions:

3783: 

3784: 

3785: \vskip .05in

3786: \noindent

3787: {\bf {Condition F}} $-$ $\tau$ has a fixed point $z$,

3788: $\alpha$

3789: acts freely and $z$ is not in the axis $\ap$.

3790: Let $\ap$ be in the component $\uu$ of $T - \{ z \}$.

3791: There is a fixed point $s$ of $\kappa$ so that

3792: $s$ is either $z$ or $z$ separates $s$ from $\ap$.

3793: Let $(s,c]$ be the bridge from $s$ to $\ap$.

3794: Then $(s,c]$ has no fixed point of $\kappa$

3795: and $(z,c]$ has no fixed point of $\tau$.

3796: Also $\uu \tau = \uu$ and there is a local axis

3797: $\lat$ of $\tau$ in $\uu$ with ideal point $z$.

3798: Finally $\lat \cap \ap = [a,b]$ where $a \not = z$ 

3799: and $a$ is in $(z,b)$.

3800: 

3801: 

3802: \vskip .05in

3803: \noindent

3804: {\bf {Condition N}} $-$ $\tau$ has a fixed point $z$; \

3805: $\kappa$ has a fixed point $s$ and $\alpha$ has

3806:  a fixed point $w$ so that $(s,w)$ has no fixed

3807: point of either $\kappa$ or $\alpha$. In addition

3808: either $z = s$ or $z \in (s,w)$ and

3809:  $(z,w)$

3810: has no fixed point of $\tau$.

3811: In addition let $\uu$ be $T_z(w)$ and $\vv$ be

3812: $T_w(z)$. Then $\uu \tau = \uu$ and $\vv \alpha = \vv$.

3813: There is a local axis $\lat$ of $\tau$ in

3814: $\uu$ with one ideal point $z$ and a local

3815: axis $\lal$ of $\alpha$ in $\vv$ with ideal

3816: point $w$. The intersection of $\lal$ and $\lat$ is

3817: $[a,b]$ where $a$ is the closest point

3818: to $z$ and $b$ can be $+\infty$ in $\lat$.

3819: 

3820: \vskip .05in

3821: Here condition F is for free action of $\alpha$ 

3822: (which is used here) and

3823: condition N is for non free action of $\alpha$

3824: (which is used in Case C).

3825: In either case the order $\pa$ in $\lal$ corresponds

3826: to $a \ \pa \ b$.

3827: This implies the orders $<, \pa$ coincide in the

3828: intersection.

3829: Beware that the order $\pa$ here is in $\lal$ and

3830: not in $({\cal A}_{\tau}) \alpha$ as in case A.

3831: 

3832: \vskip .05in

3833: \noindent

3834: {\bf {Caution}} $-$ 

3835: An axis is also a local axis.

3836: For the sake of simplicity and to use it

3837: for case C,

3838: we will use the notation $\lal$ even in the

3839: case of $\alpha$ acting freely for the rest of the

3840: proof of case B.1.

3841: In case B.2,

3842: we will return to use the notation $\ap$ for the

3843: axis of $\alpha$.

3844: 

3845: 

3846: 

3847: 

3848: 

3849: \vskip .2in

3850: \noindent

3851: {\bf {Case B.1.3}} $-$ $\uu \gamma \not = \uu$.

3852: 

3853: We first claim that this implies that $\uu \gamma \cap \uu$

3854: is empty. 

3855: Recall that $\partial \uu = z$ and $z \tau = z$.

3856: Notice we do not know a priori that

3857: $z \gamma = z$. If $z \gamma = z$ then $\gamma$ permutes

3858: the components of $T - \{ z \}$ so one has $\uu \gamma

3859: \cap \uu = \emptyset$.

3860: Suppose then that $z \gamma$ is not $z$.

3861: Recall that there is a fixed point $s$ of $\kappa$ with

3862: $z \in [s,w]$ $-$ maybe $s = z$.

3863: If $z \gamma \not = z$, then

3864: 

3865: $$[s, z] \ \cap \ [s, z \gamma] \ = \ 

3866: [s, t] \ \ \ {\rm with} \ \ t \in [s,z),

3867: \ \ \ {\rm hence} \ \ t \in (z, z \gamma).$$

3868: 

3869: \noindent

3870: %see fig. \ref{10}, a.

3871: In particular $z$ is not equal to $s$.

3872: Notice $t$ may be equal to $s$.

3873: Here $z$ separates $\uu$ from $s$, hence

3874: $z$ separates $\uu$ from $t$.

3875: Also $z \gamma$ separates $\uu \gamma$ from

3876: $s$, hence $z \gamma$ separates $\uu \gamma$ from

3877: $t$. It follows that $t$ separates 

3878: $\uu$ from $\uu \gamma$ and $\uu \cap \uu \gamma = \emptyset$.

3879: This proves the claim.

3880: 

3881: 

3882: %Put an order $<$ in $\lat$ so that points decrease

3883: %in order when they approach $z$.

3884: 

3885: %In case B.1.3 we know that $\lal \cap \lat = [x,y]$

3886: %is more than one point, where $x$ is the closest

3887: %to $z$ $-$ hence $x < y$ in $\lat$.

3888: %Notice $x$ is not $z$ but

3889: %a priori $y$ could be $+\infty$ in $\lat$.

3890: %Put an order $\pa$ in $\lal$ so that \ $x \ \pa  \ y$ \

3891: %in $\lal$.

3892: 

3893: 

3894: \vskip .2in

3895: \noindent

3896: {\bf {Situation I}} $-$ Suppose \ $a   \alpha  \ \pa  \ a$ \ in

3897: $\lal$.

3898: 

3899: \vskip .2in

3900: \noindent

3901: {\bf {Situation I.1}} $-$ Suppose \  $a \alpha^{-1} 

3902: >_{\alpha} \  b$ \ in $\lal$, see fig. \ref{10}, a.

3903: 

3904: This implies that $a \alpha$ is not in

3905: $\lat$, see fig. \ref{10}, a.

3906: Also this implies $b$ is finite. Notice that

3907: 

3908: $$z \tau^{-1} \alpha^{-1} \tau \ = \

3909: z \alpha^{-m} \beta^{-1} \gamma^{-1} \ = \

3910: z \alpha^{-m} \tau \alpha \tau^{-1} \gamma^{-1}$$

3911: 

3912: \noindent

3913: The point $z$ bridges to $\lal$ in $a$.

3914: Hence $z \tau^{-1} \alpha^{-1} = z \alpha^{-1}$

3915: bridges to $\lal$ in $a \alpha^{-1}$, 

3916: so $z \alpha^{-1}$ is in $\uu$ and $z \alpha^{-1} \tau$

3917: is also in $\uu$, which is invariant under $\tau$.

3918: Since $\uu \gamma \cap \uu = \emptyset$, then

3919: 

3920: $$z \alpha^{-m} \tau \alpha \ \not \in \uu

3921: \ \ \ {\rm and \ it \ bridges \ to} \ \ 

3922: \lal \ \ {\rm in} \ a \ \ \Rightarrow \ \

3923: z \alpha^{-m} \tau \ \ \ {\rm bridges \ to} \ \

3924: \lal \ \ {\rm in} \ 

3925: a \alpha^{-1}$$

3926: 

3927: \noindent

3928:  and hence bridges to $\lat$ in

3929: $b$.

3930: But $z \alpha^{-m}$ bridges to $\lal$ in $a \alpha^{-m}$

3931: so bridges to $\lat$ in $a$. So $z \alpha^{m} \tau$ 

3932: bridges to $\lat$ in $a \tau$. This implies

3933: $a \tau = b$ and also 

3934: that $\tau$ is increasing in

3935: $(\lat, <)$.

3936: 

3937: %\blankfig{10}{1.}{

3938: %The case 

3939: %$\lal \cap \lat = [a,b]$: \  a. Case \ $a \alpha  \ \pa \ a$,

3940: %\ $b \ \pa  \ a \alpha^{-1}$, \

3941: %b. Case $b = a \tau = a \alpha^{-1}$,

3942: %\ c. Case \ $a \tau > b$.}

3943: 

3944: 

3945: 

3946: \begin{figure}

3947: \centeredepsfbox{lam10.eps}

3948: \caption{

3949: The case 

3950: $\lal \cap \lat = [a,b]$: \  a. Case \ $a \alpha  \ \pa \ a$,

3951: \ $b \ \pa  \ a \alpha^{-1}$, \

3952: b. Case $b = a \tau = a \alpha^{-1}$,

3953: \ c. Case \ $a \tau > b$.}

3954: \label{10}

3955: \end{figure}

3956: 

3957: 

3958: 

3959: 

3960: In addition 

3961: 

3962: $$\lab \ = \ (\lal) \tau^{-1} \ \ \ {\rm so} \ \ \

3963: \lab \cap \lat \ = \ [a \tau^{-1}, a] \ = \ 

3964: [a \tau^{-1}, b \tau^{-1}]$$

3965: 

3966: \noindent

3967: and $a \beta^{-1}$ is not in $\lat$ and

3968: bridges to $\lat$ in $a \tau^{-1}$. So this point

3969: bridges to $\lal$ in $a$ and $a \beta^{-1} \alpha^{-1}$

3970: bridges to $\lal$ in $a \alpha^{-1}$. As

3971: a result $a \beta^{-1} \alpha^{-1}$ is in $\uu$.

3972: 

3973: Also $a \alpha^{-1}$ bridges to $\lat$ in $b = a \tau$.

3974: Hence it bridges to $\lab$ in $a$. This implies

3975: that $a \alpha^{-1} \beta^{-1}$ bridges

3976: to $\lab$ in $a \beta^{-1}$ so again

3977: $a \alpha^{-1} \beta^{-1}$ is in $\uu$.

3978: Now  \ 

3979: $(a \beta^{-1} \alpha^{-1}) \gamma \ = \ 

3980: a \alpha^{-1} \beta^{-1}$.

3981: \ Which implies $\uu \gamma \cap \uu$ is not empty.

3982: This contradicts the first claim in Case B.1.3.

3983: 

3984: Situation I.1 cannot happen.

3985: 

3986: \vskip .2in

3987: \noindent

3988: {\bf {Situation I.2}} $-$ Suppose \ $a \alpha^{-1} \leq_{\alpha}

3989: b$ \ in $\lal$.

3990: 

3991: Similarly to the arguments  in situation I.1,

3992: $z \alpha^{-1} \tau$ is in

3993: $\uu$,

3994: so $z \alpha^{-m} \tau \alpha$ is not

3995: in $\uu$ so 

3996: 

3997: $$z \alpha^{-m} \tau \alpha \ \ \ {\rm bridges \ to} \ \ \ 

3998: \lal \ \ {\rm in} \  a,  \ \ z \alpha^{-m} \tau

3999: \ \ \ {\rm  bridges \   to} \ \ 

4000: \lal \ \ {\rm in} \  \ a \alpha^{-1}.$$

4001: 

4002: 

4003: \noindent

4004: Also

4005: $a \alpha^{-1} \leq_{\alpha}  b$ 

4006: in $\lal$, hence $a \alpha^{-1}$ is

4007: in $\lat$ and $a \alpha^{-1} \leq b$ in $\lat$ as well.

4008: On the other hand $z \alpha^{-m}$ bridges to 

4009: $\lat$ in $a$ so $z \alpha^{-m} \tau$ bridges

4010: to $\lat$ in $a \tau$.

4011: From this it follows that $a \tau \geq a \alpha^{-1}$

4012: in $\lat$.

4013: There are two possibilities:

4014: 

4015: \vskip .05in

4016: The first possibility is that 

4017: $a \alpha^{-1} \not = b$.

4018: In this case $z \alpha^{-m} \tau$ 

4019: bridges to $\lal$ in $a \alpha^{-1}$

4020: which is in the interior of $[a,b]$, hence this point

4021: also bridges to

4022: $\lat$ in $a \alpha^{-1}$.

4023: It follows that

4024: 

4025: $$a \tau \ = \ a \alpha^{-1} \ \ \ \Rightarrow \ \ \ 

4026: a \beta^{-1} \ = \  a \tau^{-1} \ \ \ {\rm bridges \ to} \ \

4027: \lal \ \ {\rm in} \ a.$$

4028: 

4029: 

4030: \noindent

4031: Then $a \beta^{-1} \alpha^{-1}$ bridges to

4032: $\lal$ in $a \alpha^{-1}$ so 

4033: is in $\uu$.

4034: As before consider $a \alpha^{-1} \beta^{-1}$.

4035: Here $a \alpha^{-1}$ is either in $\lab$ or

4036: bridges to $\lab$ in $b \tau^{-1}$ (the top

4037: intersection of $\lab$ with $\lat$).

4038: If $a \alpha^{-1}$ in $\lab$ then 

4039: $a \alpha^{-1} \beta^{-1}$ is in $\lab$ so in $\uu$,

4040: as above contradiction.

4041: If it bridges to 

4042: $\lab$ in $b \tau^{-1}$ then

4043: $a \alpha^{-1} \beta^{-1}$ bridges to $\lab$ in

4044: $b \tau^{-1} \beta^{-1} = b \alpha \tau^{-1}$.

4045: Since in this case 

4046: 

4047: 

4048: $$b \alpha \ > \ a \ \ {\rm in} \ \ \lat,

4049: \ \ \ {\rm then} \ \ b \alpha \tau^{-1}  \ > \  

4050: a \tau^{-1} \ \ {\rm in} \ \lat \ \ \ 

4051: \Rightarrow \ \ \ 

4052: a \alpha^{-1} \beta^{-1} \ \in \ \uu,$$

4053: 

4054: 

4055: \noindent

4056: again

4057: a contradiction.

4058: 

4059: \vskip .05in

4060: The second possibility is that $a \alpha^{-1} = b$.

4061: Here we have to split further into two options:

4062: 

4063: Recall that $a \tau \geq a \alpha^{-1}$ in $\lat$.

4064: First consider the case that $a \tau = a \alpha^{-1}$,

4065: see fig. \ref{10}, b.

4066: %The remaining situation is $a \alpha^{-1} = a \tau = b$.

4067: We have the equalities

4068: $a \beta^{-1} = a \tau \alpha \tau^{-1} = a \tau^{-1}$.

4069: Use

4070: 

4071: $$(a \alpha^m ) \tau^{-1} \alpha^{-1} \tau

4072: \ = \ a \alpha^m \alpha^{-m} \beta^{-1} \gamma^{-1}

4073: \ = a \beta^{-1} \gamma^{-1}

4074: \ = \ a \tau^{-1} \gamma^{-1} \not \in \uu$$

4075: 

4076: \noindent

4077: Hence $a \alpha^m \tau^{-1} \alpha^{-1}$ is not in $\uu$

4078: and bridges to $\lal$ in $a$,

4079: $a \alpha^m \tau^{-1}$ bridges to $\lal$ in $a \alpha$.

4080: But 

4081: 

4082: $$a \alpha^m \ \in \ \lal \ \ \Rightarrow \ \ 

4083: a \alpha^m \tau^{-1} \ \in \ \lab \ \ \ 

4084: \Rightarrow \ \ \ 

4085: \lal \cap \lab \ = \ [a , a \alpha],$$

4086: 

4087: \noindent

4088: see fig. \ref{10}, b.

4089: Now evaluate $\gamma^{-1} = \beta \alpha \beta^{-1} \alpha^{-1}$

4090: on $a \tau^{-1}$:

4091: 

4092: $$(a \tau^{-1}) \gamma^{-1} \ = \

4093: (a \beta^{-1}) \beta \alpha \beta^{-1} \alpha^{-1}

4094: \ = \ a \alpha \beta^{-1}  \alpha^{-1}.$$

4095: 

4096: \noindent

4097: Notice that $a \alpha$ is in $\lab$ so $a \alpha \beta^{-1}$

4098: is in $\lab$. Either $a \alpha \beta^{-1}$ is in

4099: $\lal$ and then $a \alpha \beta \alpha^{-1}$  is in

4100: $\lal \subset \uu$ (contradiction) $-$

4101: or 

4102: 

4103: $$a \alpha \beta ^{-1} \ \not \in \ \lal

4104: \ \ \ {\rm so \

4105: bridges \ to} \ \  \lal \ \ \ {\rm in} \ \  a \ \ \

4106: {\rm and} \ \

4107: a \alpha \beta \alpha^{-1} \ \ \ {\rm bridges \ to} \ \ 

4108: \lal \ \ {\rm in} \ \ a \alpha^{-1}$$

4109: 

4110: 

4111: \noindent

4112:  and again

4113: this point is in $\uu$.

4114: In either case $\uu \gamma \cap \uu \not = \emptyset$,

4115: contradiction.

4116: 

4117: The last option of the second possibility

4118: $a \alpha^{-1} = b$ 

4119: is that $a \tau > b = a \alpha^{-1}$ in

4120: $\lat$.

4121: Then 

4122: 

4123: 

4124: $$b \tau^{-1} \ = \ a \tau^{-1} \beta \ < \ a

4125: \ \ {\rm in} \ \lat 

4126: \ \ \ \Rightarrow \ \ \ 

4127: \lal \ \cap \ \lab \ = \ \emptyset,$$

4128: 

4129: 

4130: \noindent

4131: see fig. \ref{10}, c.

4132: Here use $\alpha \tau = \tau \alpha \beta \alpha^{m-1}$

4133: applied to $z$: The point $z \alpha$ bridges to

4134: $a$ in $\lat$ and $z \alpha \tau$ bridges to $a \tau$

4135: in $\lat$. Since $a \tau > b$, then $z \alpha \tau$

4136: bridges to $b = a \alpha^{-1}$ in $\lal$.

4137: 

4138: On the other hand $z \alpha$ bridges to $b \tau^{-1}$ in

4139: $\lab$ hence $z \alpha \beta$ bridges to 

4140: $b \tau^{-1} \beta$ in $\lab$, hence to $a$ in

4141: $\lal$. Finally

4142: $z \alpha \beta \alpha^{m-1}$ bridges  to

4143: $a \alpha^{m-1}$ in $\lal$.

4144: Since $m$ is not $0$ this is a contradiction.

4145: 

4146: We conclude that situation I cannot happen.

4147: 

4148: 

4149: 

4150: 

4151: \vskip .2in

4152: \noindent

4153: {\bf {Situation II}} $-$ \ $a \alpha^{-1} \ \pa \ a$ \ in $\lal$.

4154: 

4155: \vskip .2in

4156: \noindent

4157: {\bf {Situation II.1}} $-$ $a \alpha^{-m}$ is not in $\lat$.

4158: Here use

4159: 

4160: $$z \alpha^{-1} \tau \ = \

4161: z \tau^{-1} \alpha^{-1} \tau \ = \

4162: z \alpha^{-m} \tau \alpha \tau^{-1} \gamma^{-1}$$

4163: 

4164: \noindent

4165: is in $\uu$,

4166: so $z \alpha^{-m} \tau \alpha$ is not in $\uu$.

4167: It bridges to $\lal$ in $a$, hence $z \alpha^{-m} \tau$ bridges

4168: to $\lal$ in $a \alpha^{-1}$ and hence bridges

4169: to $\lat$ in $a$. On the other hand $z \alpha^{-m}$

4170: bridges to $\lal$ in $a \alpha^{-m}$, so bridges

4171: to $\lat$ in $b$.

4172: It follows that $z \alpha^{-m} \tau$ bridges to

4173: $\lat$ in $b \tau$ which then must be $a$.

4174: So $a < a \tau^{-1}$ in $\lat$.

4175: 

4176: %\blankfig{11}{1.}{

4177: %Case $a \alpha^{-1} \ \pa \ a$ in

4178: %$\lal$:

4179: %a. Picture when

4180: %$a \alpha^{-m} \not \in \lat$,

4181: %$a \alpha = a \tau^{-1}$,

4182: %b. Picture when $a \alpha^{-m} \in \lat, \ 

4183: %a \tau^{-1} \beta \not \in \lal$.}

4184: 

4185: 

4186: 

4187: 

4188: 

4189: \begin{figure}

4190: \centeredepsfbox{lam11.eps}

4191: \caption{

4192: Case $a \alpha^{-1} \ \pa \ a$ in

4193: $\lal$:

4194: a. Picture when

4195: $a \alpha^{-m} \not \in \lat$,

4196: $a \alpha = a \tau^{-1}$,

4197: b. Picture when $a \alpha^{-m} \in \lat, \ 

4198: a \tau^{-1} \beta \not \in \lal$.}

4199: \label{11}

4200: \end{figure}

4201: 

4202: 

4203: 

4204: 

4205: Notice $\lab \cap \lat$ is equal to $[a \tau^{-1}, b \tau^{-1}]$

4206: and this intersects $\lal$ in $a \tau^{-1} = b$.

4207: 

4208: Suppose first that $a \alpha$ is not $a \tau^{-1} = b$.

4209: Here 

4210: 

4211: 

4212: $$a \beta^{-1} \ \ \ {\rm bridges \ to}

4213: \ \ \ \lab \ \ {\rm  in} \  a \tau^{-1} \beta^{-1},

4214: \ \ \ {\rm so \ bridges \ to} \ \ \lal \ \ {\rm in} \ a \tau^{-1}.$$

4215: 

4216: \noindent

4217: Then $a \beta^{-1} \alpha^{-1}$ bridges to 

4218: $\lal$ in $a \tau^{-1} \alpha^{-1} \not = a$.

4219: It follows that $a \beta^{-1} \alpha^{-1}$ is in $\uu$.

4220: 

4221: On the other hand $a \alpha^{-1}$ bridges to $\lab$ in $a \tau^{-1} = b$,

4222: so $a \alpha^{-1} \beta^{-1}$ bridges to $\lab$ in $b \beta^{-1}$ which

4223: is not $b$ and it follows that $a \alpha^{-1} \beta^{-1}$ is also

4224: in $\uu$. As seen before  this implies $\uu \gamma \cap \uu$ is

4225: not emptyset, contradiction.

4226: 

4227: \vskip .05in

4228: The second option in situation II.1 is that 

4229: $a \alpha = a \tau^{-1}$, see fig. \ref{11}, a.

4230: 

4231: 

4232: Apply \ $\alpha^{-m}  \beta^{-1} \gamma^{-1}

4233: \ = \ \tau^{-1} \alpha^{-1} \tau$ \  to $a \alpha^m$.

4234: The right side becomes

4235: %$ a \alpha^m \alpha^{-m} \beta^{-1} \gamma^{-1} \ = \

4236: $a \beta^{-1} \gamma^{-1}$. Here

4237: 

4238: 

4239: $$a \beta^{-1} \ \in \ \uu \ \ \ \ \Rightarrow \ \ \

4240: a \beta^{-1} \gamma^{-1} \ \not \in \ \uu \ \ \ \Rightarrow \ \ \ 

4241: a \alpha^m \tau^{-1} \alpha^{-1} \ \not \in \ \uu$$

4242: 

4243: 

4244: \noindent

4245: and bridges to $\lal$ in $a$. \

4246: It follows that $a \alpha^m \tau^{-1}$ bridges to 

4247: $\lal$ in $a \alpha = a \tau^{-1} = b$.

4248: But $a \alpha^m$ is in $\lal$,

4249: so  $a \alpha^m \tau^{-1}$ is in $\lab$.

4250: Consequently 

4251: $\lal \cap \lab = a \tau^{-1} = b$, \  see fig. \ref{11}, a.

4252: 

4253: %Now consider $a \beta^{-1} \gamma^{-1}$ which is not in $\uu$.

4254: %So $a \alpha \beta^{-1} \alpha^{-1} = a \tau^{-1} \beta^{-1} \alpha^{-1}$

4255: %is not in $\uu$.

4256: %Here $x \tau^{-1} \beta$ is in $\lab$ 

4257: %and not in $\lal$.

4258: %Under 

4259: %$\alpha^{-1}$ the component of 

4260: %$T - \{ y \}$ containing $x \tau^{-1} \beta$

4261: %(that is a ray of the axis $\lab$) intersects $T - \uu$.

4262: %In particular it contains a point  sent to $z$ under $\alpha^{-1}$.

4263: %It follows that all other components of

4264: %$T - \{ y \}$ are mapped by $\alpha^{-1}$ inside $\uu$.

4265: %

4266: %Hence all other branches go into $\uu$.

4267: %In particular $x \tau^{-2} \gamma^{-1}$ is in $\uu$.

4268: %But

4269: %

4270: %$$x \tau^{-1} \alpha^{-1} \beta^{-1} \ = \ x \beta^{-1} 

4271: %\ \ \  \ {\rm is \ in} \ \ \uu.$$

4272: %

4273: %\noindent

4274: %But 

4275: %%

4276: %$$x \tau^{-1} \alpha^{-1} \beta^{-1} \gamma^{-1}

4277: %\ = \ x \tau^{-1} \alpha^{-1} \beta^{-1} \beta \alpha \beta^{-1} \alpha^{-1}

4278: %\ = \ x \tau^{-1} \beta^{-1} \alpha^{-1} \ = \ x \alpha \tau^{-1}

4279: %\alpha^{-1} \ = \ x \tau^{-2} \alpha^{-1} \ \ \ \ \  {\rm in } \ \ \ \uu.$$

4280: %

4281: %\noindent

4282: %Again we obtain $\uu \gamma \cap \uu \not = \emptyset$, contradiction.

4283: 

4284: The point

4285: $a \beta^{-1}$ is in $\uu$, hence 

4286: 

4287: $$a \beta^{-1} \gamma^{-1} \ = \ a \alpha \beta^{-1} \alpha^{-1}

4288: \ = \ a \tau^{-1} \beta^{-1} \alpha^{-1} \ = \ 

4289: a \tau^{-2} \alpha^{-1}$$

4290: 

4291: \noindent

4292: is not in $\uu$.

4293: Not only that, but also $a \beta^{-1} \gamma^{-1}$ is

4294: not equal to $z$ $-$ else some point near $a \beta^{-1}$

4295: in $\uu$ will have image under $\gamma$ in

4296: $\uu$, which is disallowed.

4297: Then 

4298: 

4299: $$z \ \in \ (a, a \tau^{-2} \alpha^{-1})

4300: \ \ \Rightarrow \ \ 

4301: z \alpha \ \in \ (a \alpha, a \tau^{-2}) \ = \ 

4302: (a \tau^{-1}, a \tau^{-2}) \ \ \Rightarrow \ \ 

4303: z \alpha \tau \ \in \ (a, a \tau^{-1}).$$

4304: 

4305: 

4306: \noindent

4307: In particular $z \alpha \tau$ is in $\lal$ and

4308: $z \alpha \tau \alpha^{1-m}$ is in $\lal$ as well.

4309: This point is equal to $z \alpha \beta$.

4310: 

4311: On the other hand 

4312: 

4313: $$z \alpha \ \in \ (a \tau^{-1}, a 

4314: \tau^{-2}) \ = \ (a \tau^{-1}, a \tau^{-1} \beta^{-1})

4315: \ \ \ \Rightarrow \ \ \ 

4316: z \alpha \beta\ \in \ 

4317: (a \tau^{-1}, a \tau^{-1} \beta).$$ 

4318: 

4319: 

4320: \noindent

4321: But then

4322: $z \alpha \beta$ is not in $\lal$, contradiction.

4323: 

4324: This finishes the analysis of situation II.1,

4325: $a \alpha^{-m}$ is not in $\lat$.

4326: 

4327: \vskip .2in

4328: \noindent

4329: {\bf {Situation II.2}} $-$ $a \alpha^{-m}$ is in $\lat$.

4330: 

4331: In particular $a \alpha$ is in $(a,b]$.

4332: Here $z \alpha^{-m} \beta^{-1} \gamma^{-1} =

4333: z \tau^{-1} \alpha^{-1} \tau$ is in $\uu$.

4334: As usual this implies 

4335: $z \alpha^{-m} \tau \alpha$ is not in $\uu$

4336: and bridges to $\lal$  

4337: in $a$ and $z \alpha^{-m} \tau$ bridges to $\lal$

4338: in $a \alpha^{-1}$, see fig. \ref{11}, b; so

4339: $z \alpha^{-m} \tau$

4340: bridges to $\lat$ in $a$.

4341: So 

4342: 

4343: 

4344: $$z \alpha^{-m} \ \ \ {\rm bridges \ to} \ \lat \ \ {\rm in}

4345: \ a \tau^{-1} \ \ \ \Rightarrow \ \ \ 

4346: a \tau^{-1} \ > \ a \ \ {\rm in} \ \ \lat.$$

4347: 

4348: 

4349: Notice $z \alpha^{-m}$ bridges to $\lal$ in $a \alpha^{-m}$.

4350: If \ $a \alpha^{-m} \pa \ b$ \ in $\lal$, then

4351: $z \alpha^{-m}$ also bridges to $\lat$ in $a \alpha^{-m}$

4352: and $a \alpha^{-m} = a \tau^{-1}$.

4353: If 

4354: 

4355: $$a \alpha^{-m} \ = \ b \ \ \ {\rm then} \ \ 

4356: z \alpha^{-m} \ \ \ {\rm bridges \ to} \ \ 

4357: \lat \ \ \ {\rm in \ a  \ point } \ 

4358: \geq a \alpha^{-m},$$

4359: 

4360: \noindent

4361: that is, $a \tau^{-1} \geq a \alpha^{-m}$ in

4362: $\lat$.

4363: In any case 

4364: $a \alpha^{-m} \leq a \tau^{-1}$ in $\lat$ and 

4365: $a \alpha < a \tau^{-1}$ in $\lat$.

4366: 

4367: Now compute $a \gamma = a \alpha \beta \alpha^{-1} \beta^{-1}$.

4368: Here $a \alpha$ is in $[a, a \tau^{-1}]$

4369: and bridges to $\lab$ in $a \tau^{-1}$.

4370: Hence $a \alpha \beta$ bridges to $\lab$

4371: in $a \tau^{-1} \beta$.

4372: There are two options:

4373: First if $a \tau^{-1} \beta$ is not in $\lal$, then

4374: $a \alpha \beta$ bridges to a point

4375: $v$ in $\lal$ and  $v \in (a, a \tau^{-1} \beta)$ $-$

4376: see fig. \ref{12}, b.

4377: Here 

4378: $v$ could be in $\lat$. 

4379: Then 

4380: 

4381: 

4382: $$a \alpha \beta \alpha^{-1} \ \ \ {\rm bridges \ to \ a \ point} \ 

4383: \ v \alpha^{-1} \ \ {\rm in} \ \ \lal

4384: \ \ \Rightarrow \ \ {\rm it \ bridges \ a \ point} \ \ 

4385: c \ \ {\rm in}  \ \lab, \ \ c \ \in \

4386: (b \tau^{-1}, a \tau^{-1} \beta).$$

4387: 

4388: 

4389: \noindent

4390: It follows that 

4391: $a  \gamma = a \alpha \beta \alpha^{-1} \beta^{-1}$ bridges

4392: to a point in $\lab$ which is not $a \tau^{-1}$, hence

4393: $a \gamma$ is in $\uu$, contradiction.

4394: 

4395: The second option here is

4396: that $a \tau^{-1} \beta$ is in $\lal$.

4397: Here $a \tau^{-1}$ is in $\lal$.

4398: Then consider $a \tau^{-1} \alpha^{-1}$ which is in $\lal$ and hence

4399: in $\uu$. Then

4400: 

4401: $$(a \tau^{-1} \alpha^{-1}) \alpha \beta \alpha^{-1}

4402: \ = \ a \tau^{-1} \beta \alpha^{-1}$$

4403: 

4404: \noindent

4405: is in $\lal$ and 

4406: \ $a \tau^{-1} \beta \alpha^{-1}  \ \pa \ 

4407: a \tau^{-1} \beta$  \ in $\lal$.

4408: Therefore 

4409: 

4410: 

4411: $$a \tau^{-1} \beta \alpha^{-1} \ \ \ 

4412: {\rm bridges \ to \ a \ point \ in} \ 

4413: \lab \ \ \ {\rm contained \ in} \ \ 

4414: (b \tau^{-1}, a \tau^{-1} \beta).$$ 

4415: 

4416: 

4417: \noindent

4418: Apply $\beta^{-1}$ $-$ the resulting point

4419: bridges to a point in

4420: $\lab$ which is not $a \tau^{-1}$, hence

4421: $(a \tau^{-1} \alpha^{-1}) \gamma$ is in $\uu$, again a contradiction.

4422: 

4423: This finishes the analysis of situation $II$.

4424: Hence this finishes the analysis of case B.1.3,

4425: $\uu \gamma$ is not equal to $\uu$.

4426: %finishes the analysis of case B.1:

4427: %$Fix(\tau) \cap \lal = \emptyset$.

4428: 

4429: %\blankfig{13}{1.}{The case of intersection being a single point.}

4430: %

4431: %\vskip  .2in

4432: %\noindent

4433: %{\bf {Situation II}} $-$ $\lal \cap \lat = \{ x \}$.

4434: %

4435: %Here $x \tau^{-1}$ is in $\lat$ and $x  \tau^{-1} \alpha^{-1}$ 

4436: %%bridges to $\ap$  in $x \alpha$ so it is in $\uu$.

4437: %It follows that $x \tau^{-1} \alpha^{-1} \tau$ is in $\uu$.

4438: %Hence

4439: %

4440: %$$x \tau^{-1} \alpha^{-1} \tau \ = \

4441: %x \alpha^{-m} \tau \alpha^{-1} \tau^{-1} \gamma^{-1}

4442: %\ \ \ \ \ \ \ \in \uu$$

4443: %

4444: %\noindent

4445: %so $x \alpha^{-m} \tau \alpha$ is not in $\uu$.

4446: %But $x \alpha^{-m} \in \ap$ bridges to $\lat$ in $x$,

4447: %$x \alpha^{-m}  \tau$ bridges to $\lat$ in $x \tau$,

4448: %bridgesto $\ap$ in $x$, so 

4449: %$x \alpha^{-m} \tau \alpha$ bridges to $\ap$ in $x \alpha$

4450: %and $x \alpha^{-m} \tau \alpha$ is in $\uu$.

4451: %This  is a  contradiction.

4452: %

4453: %This finishes t he analysis of the case

4454: %$Fix(\tau) \cap \ap = \emptyset$.

4455: 

4456: 

4457: \vskip .2in

4458: \noindent

4459: {\bf {Case B.1.4}} $-$ Suppose $\uu \gamma = \uu$.

4460: 

4461: %This is a more complicated situation which will

4462: %have a longer proof.

4463: Since the boundary $\partial \uu$ in $T$ is the point $z$ this

4464: implies that $z \gamma = z$.

4465:  Here $(\lat) \gamma \cap \lat \not = \emptyset$,

4466: choose

4467: $c \gamma$ in this intersection.

4468: So $c, c \gamma$ are disjoint and in $\lat$. If follows

4469: that $z, c, c \gamma$ are aligned (the particular

4470: order is not important) and $c$ is in a local

4471: axis of $\gamma$. But $c \gamma^{-q} = c \tau^p $ is

4472: also in $\lat$ and it follows easily that the local

4473: axis is contained in and therefore equal to

4474: the local axis $\lat$ of $\tau$

4475: so $\gamma, \tau$ and hence $\kappa$ leaves

4476: $\lat$ invariant. This sort of argument

4477: will be used from time to time from now on.

4478: 

4479: Here the ideal would be to apply the proof of

4480: case A, where $\tau$ acted freely and $\at$

4481: was invariant by $\gamma$ and $\tau$. We already

4482: have $\lat$ invariant under $\gamma$ and $\tau$,

4483: however $\lat$ is not properly embedded in

4484: $T$ - at least in the $z$ direction.

4485: In order to apply the proof of case A, we analyse

4486: the relative positions of $(\lat) \alpha, (\lat) \alpha \tau$

4487: and so on. In particular for that analysis to work

4488: we must have $(\lat) \alpha$ contained in $\uu$ and so

4489: on. So first we 

4490: do preparation work, showing all images

4491: of the local axis are in $\uu$ and then we can

4492: apply the proof of case A.

4493: 

4494: %But this will also recur in Case C, when both

4495: %$\tau$ and $\alpha$ have local axis. Therefore

4496: %we will do the arguments here in more generality.

4497: For simplicity of notation in case B.1.4 we do the

4498: following: 

4499: $\kk$ will denote the local axis

4500: $\lat$ which is contained in $\uu$ and has an

4501: ideal point $z$.

4502: Again as we want to use this in section C as well,

4503: we will consider a local axis $\lal$ for $\alpha$.

4504: The key result is the following:

4505: %First we detailed the conditions to which this applies.

4506: 

4507: %\vskip .1in

4508: %\noindent

4509: %{\bf {Condition}} $-$

4510: %Suppose that $\tau$ has a fixed point $z$ and 

4511: %$-$ $\alpha$ acts freely with axis $\ap$ not intersecting

4512: %$z$ and if $\uu$ is the component of $T - \{ z \}$ 

4513: %containing $\ap$ then $\uu \kappa = \uu$ (hence

4514: %$\uu \tau = \uu$, $\uu \gamma = \uu$).

4515: %Here $\kk$ will take the role of the axis $\at$ in case A.

4516: %

4517: %\vskip .1in

4518: %\noindent

4519: %{\bf {Situation 0}} $-$ First suppose that $\kk \cap \ap$ has

4520: %%at most one point.

4521: %

4522: %Let $[x,y]$ be the bridge from $\kk$ to $\ap$.

4523: %Notice that $x$ is not an ideal point of $\kk$ or else

4524: %that would be fixed by $\tau$ and closer to $\ap$.

4525: %Hence $x$ is in $\kk$. 

4526: %Then $[x, x \alpha]$ is the bridge from $\kk$ to $\kk \alpha$.

4527: %In particular $\kk \cap \kk \alpha$ is empty.

4528: %In this situation one can apply the proof of 

4529: %case A.1 directly.

4530: %

4531: %\vskip .1in

4532: %\noindent

4533: %{\bf {Situation 1}} $-$ From now on in Case B.1.3 assume that

4534: %$\kk \cap \kk \alpha = [x,y]$ with

4535: %$x$ closest to $z$.

4536: %Here $y$ may be infinity.

4537: 

4538: %We collect the results we need in a lemma:

4539: 

4540: \begin{lemma}{}{}

4541: We have $\kk \alpha \subset \uu$,

4542: $\kk \alpha^{-1} \subset \uu$ and $\kk \alpha \tau \alpha^{-1}

4543: \subset \uu$.

4544: \label{probl}

4545: \end{lemma}

4546: 

4547: \begin{proof}{}

4548: We treat each case in turn:

4549: 

4550: \vskip .1in

4551: \noindent

4552: {\bf {Problem 1}} $-$ Is $\kk \alpha \subset \uu$?

4553: 

4554: Suppose not. Then as $a \alpha$ is in $\lal$ contained

4555: in $\uu$ there is $t$ in $\kk$ with $t \alpha = z$ or

4556: $z \alpha^{-1}$ is in $\kk$, see fig. \ref{13}, a.

4557: Here $z$ bridges to $a$ in $\lal$ so $z \alpha^{-1}$

4558: bridges to $a \alpha^{-1}$ in $\lal$. So $z \alpha^{-1}$

4559: can only be in $\kk$ if

4560: $b$ is in $(z, z \alpha^{-1})$

4561: and $a \alpha^{-1} = b$. In particular 

4562: \ $a \alpha \ \pa \ a$ \ in $\lal$.

4563: 

4564: %\blankfig{12}{1.}{

4565: %Analysing $z \alpha^{-1} \in \kk$:

4566: %a. Picture when $a \tau \in [z,a)$,

4567: %\ b. Picture when $a \tau^{-1} \in [z,a)$.}

4568: 

4569: 

4570: \begin{figure}

4571: \centeredepsfbox{lam12.eps}

4572: \caption{

4573: Analysing $z \alpha^{-1} \in \kk$:

4574: a. Picture when $a \tau \in [z,a)$,

4575: \ b. Picture when $a \tau^{-1} \in [z,a)$.}

4576: \label{12}

4577: \end{figure}

4578: 

4579: 

4580: 

4581: 

4582: 

4583: There are two possibilities depending on whether

4584: $\tau$ is expanding away from $z$ or not:

4585: 

4586: First suppose $a \tau$ is in $[z, a)$, see fig.

4587: \ref{12}, a.

4588: As $z \alpha$ bridges to $a$ in $\kk$ then $z \alpha \tau$

4589: bridges to $a \tau$ in $\kk$ so bridges to

4590: $a$ in $\lal$. Then $z \alpha \tau \alpha^{-m}$  bridges

4591: to $a \alpha^{-m}$ in $\lal$.

4592: The point $z \alpha \tau \alpha^{-m}$

4593: is equal to $z \beta$ and

4594: bridges to $a$ in $\kk$ so bridges to $a \tau^{-1}$ in

4595: $\lab$. But $z$ also bridges 

4596: to $a \tau^{-1}$ in $\lab$, contradiction.

4597: 

4598: The second option is $a \tau > a$ in $\kk$, see fig. \ref{12}, b.

4599: Here $z \beta^{-1} $ bridges to $a \tau^{-1} \beta^{-1}$ 

4600: in $\lab$ and

4601: so to $a$ in $\lal$.

4602: Hence 

4603: 

4604: $$z \beta^{-1} \alpha^{-1} \ \ \ {\rm bridges \ to} \ 

4605: a  \alpha^{-1} \ \ {\rm in} \  \lal

4606: \ \ \ \Rightarrow \ \ \ 

4607: z \beta^{-1} \alpha^{-1} \ \in \ \uu.$$

4608: 

4609: \noindent

4610: On the other hand $z \alpha^{-1} \beta^{-1} =

4611: z \alpha^{-1} \tau \alpha \tau^{-1}$.

4612: Here 

4613: 

4614: $$z \alpha^{-1} \tau \ \in \ \kk \ \ \Rightarrow \ \ 

4615: z \alpha^{-1} \ \in \ (z, z \alpha^{-1} \tau) \ \ \Rightarrow \ \ 

4616: z \alpha^{-1} \tau \alpha \ \not \in \ \uu \ \ \Rightarrow \ \ 

4617: z \alpha^{-1} \beta^{-1} \ \not \in \ \uu.$$

4618: 

4619: \noindent

4620: But $z \beta^{-1} \alpha^{-1} \gamma

4621: = z \alpha^{-1} \beta^{-1}$, leading

4622: to $\uu \gamma \not = \uu$,

4623: contradiction

4624: to case B.1.3.

4625: 

4626: \vskip .05in

4627: So we obtain $z \alpha^{-1} \in \uu$ is impossible.

4628: Hence $\kk \alpha \subset \uu$.

4629: If 

4630: $\kk \alpha$ intersects $\kk$ in at most one point we can

4631: use the analysis of Case B.1.2

4632: (or of case A) and disallow it.

4633: If 

4634: 

4635: 

4636: $$\kk \ \cap \ \kk \alpha \ = \ (z, t),

4637: \ \ \ {\rm then} \ \ 

4638: \kk, \ \kk \alpha \ \ \ {\rm share \ a \ ray}.$$

4639: 

4640: 

4641: 

4642: \noindent

4643: The orientations in

4644: $\kk$ and $\kk \alpha$ may agree or not. In the first

4645: case $z \alpha = z$ and in the second case there is

4646: a fixed point $r$ of $\alpha$ in $\lat = \kk$.

4647: If $z \alpha = z$, then $z$ is a global fixed point,

4648: impossible by non trivial action. In the second option 

4649: let $\uu_1$ (respectively $\uu_2$) be the component

4650: of $T - \{ r \}$ containing $r \tau$ (respectively

4651: $r \tau^{-1}$). The condition

4652: $\kk \alpha \cap \kk = (z,t)$ implies that $\uu_1 \alpha 

4653: = \uu_2$. This is now disallowed by lemma \ref{turn}.

4654: 

4655: 

4656: Now consider the situation that $\kk$ has another ideal

4657: point $v$.

4658: Then $v \kappa = v$.

4659: Suppose first that $v$ is in $\lal$. Here we split

4660: into cases: if $\alpha$ acts freely then $v$ is

4661: a fixed point of $\tau$ in the axis of $\alpha$ and

4662: this falls under case B.2. 

4663: Consider then the case that

4664: $\alpha$ does not act freely. Then $(w,v)$ has no

4665: fixed point of  $\alpha$ (as $v$ is in $\lal$) and

4666: also no fixed point of $\tau$ or $\gamma$.

4667: Also $T_w(v)$ is invariant under $\alpha$ and

4668: $T_v(w)$ is invariant under $\tau$. 

4669: Then $v$ in $\lal$ is 

4670: disallowed by lemma \ref{noal}.

4671: 

4672: \vskip .05in

4673: It follows that 

4674: $v$ has the same properties as $z$. 

4675: In any case one obtains that

4676: 

4677: 

4678: $$\kk \alpha \ \cap \ \kk \  

4679:  = \ [t, r], \ \ t  \ \not = \ r,

4680: \ \ t \ \ \ {\rm closest \ to} \ \ z$$

4681: 

4682: \noindent

4683: and if $\kk$ is not properly embedded in the

4684: other direction then $r$ is an actual point in $\kk$.

4685: Then $\kk \alpha \tau \cap \kk = [t \tau, r \tau]$.

4686: So the intersections are the same as occurred

4687: in Case A so far.

4688: 

4689: %\blankfig{12}{1.}{Situation $z \alpha \in \kk$:

4690: %a. Picture when $x \tau^{-1} < x$ in $\kk$,

4691: %b. Picture when $x \tau < x$ in $\kk$.}

4692: 

4693: 

4694: \vskip .2in

4695: \noindent

4696: {\bf {Problem 2}} $-$ Is $\kk \alpha^{-1} \subset \uu$?

4697: 

4698: 

4699: This is similar to problem 1.

4700: As before if $\kk \alpha^{-1}$ not contained in $\uu$, then

4701: $z \in \kk \alpha^{-1}$ and $z \alpha \in \kk$.

4702: This can only happen if $b \in (z, z \alpha)$,

4703: $a \alpha = b$ and $a \alpha^{-1} \ \pa \ a$ in $\lal$.

4704: 

4705: First suppose that $a \tau^{-1} \in [z,a]$.

4706: %see fig. \ref{12}, a.

4707: Then 

4708: 

4709: $$a \tau^{-1} \alpha \ \in \ [z \alpha, a \alpha] \ = \  [b, 

4710: z \alpha] \ \ \Rightarrow \ \ a \tau^{-1} \alpha \ \in \ \kk

4711: \ \ \Rightarrow \ \ a \tau^{-1} \alpha \tau \ \in \ \kk$$

4712: 

4713: 

4714: \noindent

4715: and this last point 

4716: bridges

4717: to $b$ in $\lal$. 

4718: Then $a \tau^{-1} \alpha \tau \alpha^{-m} = a \gamma \beta$

4719: bridges to $b \alpha^{-m}$ in $\lal$.

4720: But 

4721: 

4722: 

4723: $$b \alpha^{-m} \ \pa \ b \ \ {\rm in} \ \lal \ \ \ \Rightarrow \ \ \ 

4724: a \gamma \beta \ \ {\rm bridges \ to} \ \   

4725: b \tau^{-1} \ \ {\rm in} \ \lab.$$

4726: 

4727: 

4728: \noindent

4729: On the other hand $a \gamma \in [z, a \tau^{-1}]$ and

4730: bridges to $a$ in $\lab$, so $a \gamma \beta$ bridges to 

4731: $a \tau^{-1} \beta$. 

4732: Since 

4733: $a \tau^{-1} \beta$ is a point

4734: in $\lab - \kk$ it is not equal to $b \tau^{-1}$, leading

4735: to a contradiction.

4736: 

4737: 

4738: \vskip .05in

4739: The second option is $a \tau^{-1} > a$ in $\kk$.

4740: %see fig. \ref{12}, b.

4741: Here use 

4742: 

4743: $$z \beta^{-1} \ = \ z \alpha \tau^{-1} \ \in \ \kk, \ \ \ \

4744: z \alpha \ \in \ [z, z \beta^{-1}) \ \ \ \Rightarrow \ \ \ 

4745: z \beta^{-1} \alpha^{-1} \ \not \in \ \uu.$$

4746: 

4747: 

4748: \noindent

4749: On the other hand $z \alpha^{-1}$ bridges to

4750: $a \alpha^{-1}$ in $\lal$ so bridges to $a \tau^{-1}$

4751: in $\lab$. So $z \alpha^{-1} \beta^{-1}$ bridges

4752: to $a \tau^{-1} \beta^{-1}$ in $\lab$ and is in

4753: $\uu$. As above this is a contradiction.

4754: 

4755: We conclude that problem 2 does not occur.

4756: 

4757: As in problem 1, this implies that 

4758: 

4759: 

4760: $$\kk \alpha^{-1} \ \cap \ \kk \ = \ 

4761: [t', r'], \ \ \ {\rm with} \ \ t' \ \not = \ r',

4762: t' \ \not = \ z$$

4763: 

4764: \noindent

4765: and if

4766: $\kk$ not properly embedded on the other side then

4767: $r'$ has to be finite in $\kk$.

4768: 

4769: Then clearly $\kk \alpha^{-1} \tau^{-1} \subset \uu$ and

4770: intersects $\kk$ in a segment. 

4771: 

4772: The last problem is the following:

4773: 

4774: 

4775: 

4776: \vskip .15in

4777: \noindent

4778: {\bf {Problem 3}} $-$ Does $\kk \alpha \tau \alpha^{-1}

4779: \subset \uu$?

4780: 

4781: Suppose not, that is, $\kk \alpha \tau \alpha^{-1}

4782: \not \subset \uu$.

4783: We have to be careful here.

4784: First a preliminary claim:

4785: 

4786: 

4787: \vskip .08in

4788: \noindent

4789: {\underline {Claim}} $-$ $z \in \kk \alpha \tau \alpha^{-1}$.

4790: 

4791: If this is not true then

4792: $\kk \alpha \tau \alpha^{-1} \cap \uu = 

4793: \emptyset$.

4794: Notice that 

4795: 

4796: $$\kk \alpha \tau \ \cap \ \lal  \ \not = \ \emptyset

4797: \ \ \ \Rightarrow \ \ \ 

4798: \kk  \alpha \tau \alpha^{-1} \ \cap \ \lal \

4799: \not = \ \emptyset \ \ \ {\rm and} \ \ \ 

4800: \kk \alpha \tau \alpha^{-1} \

4801: \cap \ \uu \ \not = \ \emptyset,$$

4802: 

4803: \noindent 

4804: contrary to assumption here.

4805: 

4806: So consider $\kk \alpha \tau \cap \lal = \emptyset$.

4807: Also here 

4808: $\kk \alpha \tau \cap \kk$ is a non trivial

4809: segment.

4810: If $\kk \alpha \tau$ bridges to $a$ in $\lal$ then

4811: $\kk \alpha \tau \alpha^{-1}$ is contained in

4812: $\uu$ and we are done.

4813: If follows that $\kk \alpha \tau$ has to bridge

4814: to $b$ in $\lal$ and hence $z \alpha$ has to be in

4815: the this bridge. But then $z \alpha$ is in $\kk$,

4816: which was disallowed in problem 2.

4817: This proves the claim.

4818: 

4819: \vskip .05in

4820: We now analyse what happens when 

4821: 

4822: $$z \ \in \ \kk \alpha \tau \alpha^{-1} \ \ \ {\rm so} \ \ 

4823: z \tau^{-1} \ = \ z \in \kk \alpha \beta

4824: \ \ \ {\rm and} \ \ \ z \beta^{-1} \alpha^{-1} \ \in \ \kk.$$

4825: 

4826: 

4827: \noindent

4828: Also $z \beta^{-1} \alpha^{-1} \gamma = z \alpha^{-1} \beta^{-1}$

4829: is in $\kk$ as well.

4830: %So $z$ is in $\kk \tau \alpha \tau^{-1} \alpha^{-1}

4831: %= \kk \beta \alpha$ so $z \alpha^{-1} \beta^{-1}$ is in $\kk$ as 

4832: %well.

4833: 

4834: 

4835: %\blankfig{13}{1.}{

4836: %Situation \ $a \alpha^{-1} \ \pa \ a$ \ in $\lal$:

4837: %a. Picture when $a \tau < a$ in $\kk$,

4838: %b. Picture when $a \tau^{-1} < a$ in $\kk$.}

4839: 

4840: 

4841: 

4842: 

4843: \begin{figure}

4844: \centeredepsfbox{lam13.eps}

4845: \caption{

4846: Situation \ $a \alpha^{-1} \ \pa \ a$ \ in $\lal$:

4847: a. Picture when $a \tau < a$ in $\kk$,

4848: b. Picture when $a \tau^{-1} < a$ in $\kk$.}

4849: \label{13}

4850: \end{figure}

4851: 

4852: 

4853: 

4854: \vskip .2in

4855: \noindent

4856: {\bf {Situation I}} $-$ \ $a \alpha^{-1} \ \pa \ a$ in $\lal$.

4857: 

4858: \vskip .2in

4859: \noindent

4860: {\bf {Situation I.1}} $-$ $a \tau  < a$ in $\kk$.

4861: 

4862: Here $z \alpha^{-1}$ bridges to $a \alpha^{-1}$ in

4863: $\lal$, so it bridges to $a \tau^{-1}$ in $\lab$. 

4864: Also 

4865: 

4866: 

4867: $$z \alpha^{-1} \beta^{-1} \ \in \ \kk \ \ \  {\rm and} \ \ \ 

4868: a \tau^{-1} \ \prec \ a \ \prec \ 

4869: a \alpha^{-1} \ \prec \ z \alpha^{-1}.$$

4870: 

4871: 

4872: \noindent

4873: As

4874: $\beta^{-1}$ moves 

4875: points up along $\kk$, it follows that

4876: $z \alpha^{-1} \beta^{-1} > b$ in $\kk$ and

4877: $a \tau^{-1} \beta^{-1} = b \tau^{-1}$.

4878: Here $a \alpha^{-1} \in

4879: [a \tau^{-1}, z \alpha^{-1}]$, see fig. \ref{13}, a.

4880: %$x \alpha^{-1} \beta^{-1}$ is in

4881: %$[y \beta^{-1}, z \alpha^{-1} \beta^{-1})$,

4882: %see fig. \ref{12}, a.

4883: Then 

4884: 

4885: $$a \tau^{-1} \beta^{-1} \ = \ b \tau^{-1} \ \prec \

4886: a \beta^{-1} \ \prec \ a \alpha^{-1} \beta^{-1} \ = \ 

4887: v_1 \ \prec \ z \alpha^{-1} \beta^{-1} \ = \ v_2$$

4888: 

4889: \noindent

4890: and all are in $\kk$.

4891: %Hence $a \alpha^{-1} \beta^{-1}$ is in $\kk$ also

4892: %and $v_1 =

4893: %x \alpha^{-1} \beta^{-1} < z \alpha^{-1} \beta^{-1} = _2$.

4894: Also $a \beta^{-1} \in (b, \ a \alpha^{-1} \beta^{-1})

4895: \subset \kk$ and $z \beta^{-1}$ bridges to $\kk$ in

4896: $a \beta^{-1}$ so bridges to $\lal$ in $b$.

4897: Then $z \beta^{-1} \alpha^{-1} 

4898: = v_2 \gamma^{-1} \in \kk$ bridges

4899: to $a$ in $\lal$ and $a \beta^{-1} \alpha^{-1}

4900: = v_1 \gamma^{-1}$ is in $(z \beta^{-1} \alpha^{-1}, \ a)$,  \

4901: see fig. \ref{13}, a.

4902: Then

4903: 

4904: $$z \beta^{-1} \alpha^{-1} \ \prec \ a \beta^{-1} \alpha^{-1}

4905: \ \prec \ a \alpha^{-1} \beta^{-1} \ \prec \ z \alpha^{-1} \beta^{-1}$$

4906: 

4907: 

4908: 

4909: \noindent

4910: all points in $\kk$.

4911: This contradicts the fact that $\gamma$ acts as

4912: a translation in $\kk$.

4913: 

4914: \vskip .2in

4915: \noindent

4916: {\bf {Situation I.2}} $-$

4917: $a \tau > a$ in $\kk$.

4918: 

4919: Here $z \alpha^{-1}$ bridges to $a$ in

4920: $\kk$, see fig. \ref{13}, b.

4921: If $a \geq b \tau^{-1}$ in $\kk$ then

4922: $z \alpha^{-1}$ bridges to 

4923: a point $t \geq_{\beta} b \tau^{-1}$ in $\lab$, so

4924: 

4925: 

4926: 

4927: $$z \alpha^{-1} \beta^{-1} \ \ \ {\rm bridges \ to} \ \ 

4928: \lab \ \ {\rm in \ a \ point} \ 

4929: \geq_{\beta} \ b \tau^{-1} \beta \ \ \ {\rm and} \ \ \

4930: z \alpha^{-1} \beta^{-1} \ \not \in \ \kk,$$

4931: 

4932: 

4933: \noindent

4934: contradiction.

4935: Hence $a < b \tau^{-1}$ in $\kk$ and 

4936: $z \alpha^{-1}$ bridges to $a$ in $\lab$ so

4937: $z \alpha^{-1} \beta^{-1}$ bridges to

4938: $a \beta^{-1}$ in $\lab$ and as $z \alpha^{-1} \beta^{-1}$

4939: is in $\kk$ then 

4940: 

4941: $$z \alpha^{-1} \beta^{-1} \ > \ b \tau^{-1} \ \ {\rm in} \ \

4942: \kk  \ \ \ {\rm and} \ \ \  a \beta^{-1} \ = \ b \tau^{-1}

4943: \ \ \ {\rm or} \ \ 

4944: a \tau \alpha \ = \  b.$$

4945: 

4946: \noindent

4947: Now

4948: 

4949: $$a \beta^{-1} \ = \ b \tau^{-1} \ \ \ {\rm so} \ \ \ 

4950: a \alpha \ = \ a \tau^{-1} \beta^{-1} \tau \ < \ a \beta^{-1} \tau

4951: \ = \ b$$

4952: 

4953: \noindent

4954: so in particular 

4955: $a \alpha$ is in $\kk$.

4956: Also $z \beta$ bridges to $a$ in $\lal$ and so does $z$.

4957: Hence $z \beta \alpha = z \alpha \beta$ and $z \alpha$

4958: bridge to $a \alpha$ in $\lal$. 

4959: Since $a \alpha < b$ then $z \alpha, z \alpha \beta$ bridge

4960: to $a \alpha$ in $\lat$ as well.

4961: 

4962: If $a \alpha < b \tau^{-1}$ in $\kk$ then

4963: $z \alpha, z \alpha \beta$ bridge 

4964: to $a \alpha$ in $\lab$, impossible $-$ they have

4965: to bridge to distinct points in $\lab$.

4966: If 

4967: 

4968: $$b \tau^{-1} \ \in \ (a, a \alpha) \ \ \ \Rightarrow \ \ 

4969: z \alpha, \ z \alpha \beta \ \ \ {\rm bridge \ to}  \ \ 

4970: b \tau^{-1} \ \ {\rm in}  \ \lab,$$

4971: 

4972: 

4973: \noindent

4974: also contradiction. Therefore $a \alpha = b \tau^{-1}$

4975: or $a \alpha \tau = b$. 

4976: Now 

4977: 

4978: $$a \alpha \tau \alpha^{-1} \tau^{-1}

4979: \ = \ b \alpha^{-1} \tau^{-1} \ = \ a

4980: \ \ \ \ {\rm so} \ \ \ a \gamma = a \alpha^{-1} \beta^{-1}.$$

4981: 

4982: \noindent

4983: Notice $a \gamma \in [z, a \tau^{-1}]$.

4984: But $a \alpha^{-1}$ bridges to $a$ in $\lab$ so

4985: $a \alpha^{-1} \beta^{-1}$ bridges to $a \beta^{-1}

4986: = b \tau^{-1}$ in $\lab$ and

4987: $a \alpha^{-1} \beta^{-1}$ cannot be

4988: $a \gamma$, contradiction.

4989: 

4990: This finishes the analysis of situation I.

4991: 

4992: The remaining options are extremely similar and

4993: have shortened proofs.

4994: 

4995: %\blankfig{14}{1.}{The situation $x \alpha \pa x$ in 

4996: %$\ap$: a. Picture when $x \tau^{-1} < x$ in $\kk$,

4997: %b. Picture when $x \tau < x$ in $\kk$.}

4998: 

4999: 

5000: \vskip .2in

5001: \noindent

5002: {\bf {Situation II}} $-$  $a \alpha \ \pa \ a$ \ in $\lal$.

5003: 

5004: \vskip .2in

5005: \noindent

5006: {\bf {Situation II.1}} $-$ $a \tau^{-1} < a$ in $\kk$.

5007: 

5008: This is as situation I.1 above.

5009: Here $z \beta^{-1}$ bridges to $a$ in $\lal$,

5010: so $z \beta^{-1} \alpha^{-1}$ bridges to $a \alpha^{-1}$

5011: in $\lal$ and $a \alpha^{-1} = b$.

5012: It follows that

5013: 

5014: $$b \ \prec \ a \tau^{-1} \alpha^{-1} \ \prec \ 

5015: a \tau^{-1} \beta^{-1} \alpha^{-1} \ \prec \

5016: z \beta^{-1} \alpha^{-1},$$

5017: 

5018: \noindent

5019: all points in $\kk$.

5020: %Also $(a \tau^{-1}) \beta^{-1} \alpha^{-1} \in 

5021: %(b, z \beta^{-1} \alpha^{-1})$.

5022: 

5023: On the other hand $a \tau^{-1} \alpha^{-1} \in 

5024: (b, \ (a \tau^{-1}) \beta^{-1} \alpha^{-1}) \subset \kk$.

5025: The point $z \alpha^{-1}$ bridges to $(a \tau^{-1}) \alpha^{-1}$

5026: in $\kk$. 

5027: It follows that

5028: 

5029: $$z \alpha^{-1} \beta^{-1} \ \prec \ 

5030: (a \tau^{-1}) \alpha^{-1} \beta^{-1} \ \prec \ 

5031: (a \tau^{-1}) \beta^{-1} \alpha^{-1} \ \prec \ z \beta^{-1} \alpha^{-1},$$

5032: 

5033: 

5034: \noindent 

5035: all points in $\kk$.

5036: %the positioning of $z \alpha^{-1} \beta^{-1},

5037: %z \beta^{-1} \alpha^{-1}$,

5038: %$(x \tau^{-1}) \alpha^{-1} \beta^{-1}$  and

5039: %$(x \tau^{-1}) \beta^{-1} \alpha^{-1}$ in $\kk$ 

5040: As before this contradicts the fact that $\gamma$ acts as

5041: a translation in $\kk$.

5042: 

5043: 

5044: 

5045: \vskip .2in

5046: \noindent

5047: {\bf {Situation II.2}} $-$ $a \tau < a$ in $\kk$.

5048: 

5049: This is very much like situation I.2.

5050: Here $z \beta^{-1}$ bridges to $a \tau^{-1}$ in $\kk$.

5051: If $a \tau^{-1} \geq b$ in $\kk$, 

5052: then 

5053: 

5054: $$z \beta^{-1} \alpha^{-1} \ \ \ {\rm bridges \ to \ a \ point}

5055: \ \ 

5056: >_{\alpha} \  b \ \ {\rm in} \ \ \lal \ \ \ \Rightarrow \ \ \ 

5057: z \beta^{-1} \alpha^{-1} \ \not \in \ \kk,$$

5058: 

5059: \noindent

5060:  contradiction.

5061: Hence 

5062: 

5063: $$a \tau^{-1} \ <  \ b \ \ {\rm in} \ \kk, \ \ \ 

5064: z \beta^{-1} \alpha^{-1}  \ > \ b \ \ {\rm in } \  \kk

5065: \ \ \ {\rm and} \ \ \ 

5066: a \tau^{-1} \alpha^{-1} \ = \ b \ \ {\rm or} \ \ 

5067: a \ = \ b \alpha \tau.$$

5068: 

5069: 

5070: \noindent

5071: In addition 

5072: 

5073: $$z \alpha, z \ \  {\rm bridge \ to} \ \lab \ \ {\rm in} \ 

5074: a \tau^{-1}  \  \ \ \Rightarrow \  \ \

5075: z \beta \alpha  =  

5076: z \alpha \beta,  \ z \beta \ \ {\rm bridge \ to } \ 

5077: \lab \ \ {\rm in} \ a \tau^{-1} \beta$$

5078: 

5079: 

5080: \noindent

5081: and similarly to situation I.2, this implies

5082: $a \tau^{-1} \beta = b$ or

5083: $ a  = b \tau \alpha$. Then $b \alpha \beta = b$

5084: and $b \gamma =  b \alpha^{-1} \beta^{-1}$.

5085: But $b \gamma \geq b \tau^{-1}$ in $\kk$ and

5086: $b \alpha^{-1}$ bridges to $b$ in $\lab$, so

5087: $b \alpha^{-1} \beta^{-1}$ bridges to $b \beta^{-1}

5088: = a \tau^{-1}$ and cannot be equal to $b \tau^{-1}$.

5089: 

5090: This contradiction shows that problem 3 cannot occur.

5091: This finishes the proof of lemma \ref{probl}.

5092: \end{proof}

5093: 

5094: It follows from lemma \ref{probl} that 

5095: $\kk \alpha \tau \alpha^{-1} \subset \uu$,

5096: so $\kk \alpha \beta \subset \uu$ as

5097: is $\kk \gamma \beta \alpha$. So all of the

5098: sets $\kk, \ \kk \alpha, \ \kk \alpha \tau$, 

5099: \ $\kk \alpha \tau \alpha^{-1}, \

5100: \kk \alpha \beta, \ \kk \alpha^{-1}, \  \kk \alpha^{-1} \tau^{-1}$

5101: \ and $\kk \alpha^{-1} \tau^{-1} \alpha$ are contained in

5102: $\uu$ and none has $z$ as an ideal point.

5103: If $\kk$ has another ideal point $v$, then

5104: $v$ has the same properties as $z$ and

5105: the same situation

5106: occurs with respect to this other ideal point.

5107: 

5108: %One then has to analyse the relative positioning of all these

5109: %axes. 

5110: Given these facts, an analysis exactly as in

5111: case A.2 can be applied here.

5112: That analysis then shows that case B.1.3 is not possible.

5113: 

5114: %$-$ $\alpha$ has a fixed point $r$ and $\kappa$ has 

5115: %a fixed point $s$ so that $s \preceq z \prec r$ and

5116: %$(s,r)$ has no fixed point of either

5117: %$\alpha$ or $\kappa$ and $(z,r)$ has no fixed

5118: %points of $\tau$. It could be that $z = s$.

5119: %Let $\uu$ be the component of $T - \{ z \}$ containing

5120: %$r$ and $\vv$ the component of $T - \{ r \}$ containing

5121: %$z$. Then we assume that $\uu \tau 

5122: 

5123: %We can now apply the proof of case A which

5124: %also works for local axes which are 

5125: %{\underline {both}} of $\tau$ and $\gamma$.

5126: %Notice there are no fixed points of

5127: %$\alpha$ so we never have to obtain

5128: %the condition $p > 3q$ or $m$ even.

5129: Hence case B.1.4 is disallowed.

5130: This also finishes the proof of case B.1.

5131: 

5132: \vskip .05in

5133: For case B.2 we return to the study of $\alpha$

5134: acting freely using the axis $\ap$.

5135: 

5136: 

5137: 

5138: 

5139: 

5140: 

5141: 

5142: \vskip .3in

5143: \noindent

5144: {\bf  {Case B.2}} $-$ $Fix(\tau) \cap \ap \not = \emptyset$.

5145: 

5146: This is the key case of the proof for essential laminations.

5147: In this case the topology will be important, in particular, the

5148: exact condition $|p - 2q| = 1$ will be used in a crucial manner.

5149: Let $z \in Fix(\tau) \cap \ap$.

5150: Let $\uu_1$ (respectively $\uu_2$) 

5151: be the component of $T - \{ z \}$ containing

5152: $z \alpha$ (respectively $z \alpha^{-1}$).

5153: A priori we do not know whether $z$ is also a fixed point

5154: of $\gamma$.

5155: In some subcases, the tricky part will be in fact to

5156: show that $z \gamma = z$.

5157: 

5158: \vskip .1in

5159: \noindent

5160: {\bf {Case B.2.1}} $-$ $\uu_1 \tau = \uu_1$.

5161: 

5162: Notice that $\uu_1 \alpha$ is  contained in $\uu_1$.

5163: Here use 

5164: $z \alpha \tau = z \tau \gamma \beta \alpha^m =

5165: z \gamma \beta \alpha^m$.

5166: 

5167: $$z \alpha \in \uu_1 \ \Rightarrow \ z \alpha \tau \in \uu_1

5168: \ \Rightarrow \ z \alpha \tau \alpha^{-m} \in \uu_1 \alpha^{-m} \subset \uu_1

5169: \ \Rightarrow \ z \gamma \beta \in \uu_1.$$

5170: 

5171: \noindent

5172: So $z \gamma \tau \alpha^{-1} \tau^{-1}$ is in $\uu_1$ and

5173: then  $z \gamma \alpha^{-1} $ is in $\uu_1$ or

5174: $z \gamma$ is in $\uu_1 \alpha$. In particular

5175: $z \prec z \alpha \prec z \gamma$, see fig. \ref{14}, a.

5176: We stress that 

5177: in this case $z \gamma$ is not equal to $z$!

5178: 

5179: Use now 

5180: $z \alpha \tau = z \alpha \beta \alpha^{m-1} =

5181: z \alpha \tau \alpha^{-1} \tau^{-1} \alpha^{m-1}$.

5182: 

5183: $$z \alpha \tau \alpha^{1-m} \in \uu_1 \ \Rightarrow \

5184: z \alpha \tau \alpha^{-1} \tau^{-1} \in \uu_1 \ \Rightarrow \

5185: z \alpha \tau \alpha^{-1} \in \uu_1

5186: \ \Rightarrow \ z \alpha \tau \in \uu_1 \alpha.$$

5187: 

5188: \noindent

5189: In particular $z \prec z \alpha  \prec z \alpha \tau$

5190: and $z \prec z \alpha \tau^{-1} \prec z \alpha$

5191: and so $z \alpha \tau^{-1} \alpha^{-1} \in (z \alpha^{-1}, z)$.

5192: In other words

5193: 

5194: $$z \alpha \tau^{-1} \alpha^{-1} \ = \ 

5195: z \tau \alpha \tau^{-1} \alpha^{-1} \ = \ 

5196: z \beta^{-1} \alpha^{-1} \ \ \ \in \ (z \alpha^{-1}, z).$$

5197: 

5198: \noindent

5199: Then $z \beta^{-1} \alpha^{-1}$ is in $\uu_2$ so

5200: $z \beta^{-1} \alpha^{-1} \gamma$ is in $\uu_2  \gamma$.

5201: Notice $z \beta^{-1} = z \tau \alpha \tau^{-1} = z \alpha \tau^{-1}$

5202: with $z \alpha \in \uu_1$, $z \alpha \tau^{-1}$ also in $\uu_1$.

5203: 

5204: %\blankfig{14}{1.}{

5205: %Case B: a. Picture when

5206: %$\uu_1 \tau = \uu_1$,

5207: %b. Picture when $\uu_1 \tau^{-1} = \uu_2$

5208: %and

5209: %$[z, z \beta] \cap [z, z \alpha] =

5210: %[z,t]$

5211: %.}

5212: 

5213: 

5214: 

5215: \begin{figure}

5216: \centeredepsfbox{lam14.eps}

5217: \caption{

5218: Case B: a. Picture when

5219: $\uu_1 \tau = \uu_1$,

5220: b. Picture when $\uu_1 \tau^{-1} = \uu_2$

5221: and

5222: $[z, z \beta] \cap [z, z \alpha] =

5223: [z,t]$

5224: .}

5225: \label{14}

5226: \end{figure}

5227: 

5228: 

5229: 

5230: 

5231: Recall that $z \gamma \not = z$.

5232: If  $\uu_1 \gamma \subset \uu_1$ this implies 

5233: that $z$ is in a local axis for $\gamma$ contradicting

5234: $z \gamma^q = z \tau^{-p} = z$. 

5235: Therefore $\uu_1 \gamma$ is not contained in $\uu_1$ and

5236: consequently $\uu_2 \gamma$ is contained in $\uu_1$

5237: and so $z  \gamma$ separates $\uu_2 \gamma$  from $z$.

5238: Hence 

5239: 

5240: 

5241: $$z  \alpha \ \ \ {\rm separates} \ \  \uu_2 \gamma

5242: \ \ {\rm from} \   z \ \ \ \ {\rm and} \ \ \ 

5243: z \beta^{-1} \alpha^{-1} \gamma \ \in \ \uu_2 \gamma.$$

5244: 

5245: \noindent

5246: But $z \beta^{-1}  \alpha^{-1} \gamma = z \alpha^{-1} \beta^{-1}

5247: = z \alpha^{-1} \tau \alpha \tau^{-1}$.

5248: Now $z \alpha$  separates $z$ from 

5249: %$z \gamma$, hence $z \alpha$

5250: %separates $z$ from 

5251: $z \alpha^{-1} \tau \alpha \tau^{-1}$

5252: which is in $\uu_2 \gamma$. Apply $\tau$:

5253: $z \alpha \tau$ separates  $z$ from $z \alpha^{-1} \tau \alpha$.

5254: Then

5255: 

5256: $$z \alpha \tau \ \in \ \uu_1 \alpha \ \ \Rightarrow \ \ 

5257: z \alpha^{-1} \tau \alpha \ \in \ \uu_1 \alpha \ \ \Rightarrow \ \ 

5258: z \alpha^{-1} \tau \ \in \ \uu_1 \ \ {\rm and} \ \ 

5259: z \alpha^{-1} \ \in \ \uu_1 \tau^{-1} = \uu_1.$$

5260: 

5261: \noindent

5262: But this contradicts $z \alpha^{-1}$ is in $\uu_2$.

5263: This is an impossible case.

5264: 

5265: We conclude that $\uu_1 \tau \not = \uu_1$.

5266: 

5267: \vskip .2in

5268: \noindent

5269: {\bf {Case B.2.2}} $-$ $\uu_1 \tau \not = \uu_2$.

5270: 

5271: Then $z \alpha \tau$ is not in $\uu_2$, which 

5272: implies $z \alpha \tau \alpha^{1-m}$ is in $\uu_1$,

5273: or $z \alpha \beta \in \uu_1$

5274: and $z \alpha \tau \alpha^{-1} \tau^{-1}$ is

5275: in $\uu_1$.

5276: By assumption $z \alpha \tau \not \in \uu_1$, hence

5277: $z \alpha \tau \alpha^{-1} \in \uu_2$ and

5278: $z \alpha \tau \alpha^{-1} \tau^{-1} \in \uu_2 \tau^{-1}$.

5279: This would imply $\uu_2 \tau^{-1} = \uu_1$

5280: or $\uu_1 \tau = \uu_2$, so the assumption is incompatible.

5281: 

5282: We conclude that $\uu_1 \tau = \uu_2$.

5283: 

5284: \vskip .2in

5285: \noindent

5286: {\bf {Case B.2.3}} $-$ $\uu_1 \tau^{-1} = \uu_2$.

5287: 

5288: This is a very interesting case. Here we only use

5289: the fact that $p$ is odd.

5290: 

5291: 

5292: First consider $z \beta = z \tau \alpha^{-1} \tau^{-1}

5293: = z \alpha^{-1} \tau^{-1}$ which is in

5294: $\uu_2 \tau^{-1} = \uu_1$.

5295: Then $z \alpha, z \beta$ are in the component $\uu_1$,

5296: hence $[z, z \alpha]$, $[z,  z \beta]$ share

5297: a subprong.

5298: Suppose first that

5299: 

5300: $$[z, z \beta] \cap [z, z \alpha] =

5301: [z,t],  \ \ t \not = z \alpha, z \beta,

5302: \ \ \ {\rm that \ is} \ \ \ z \alpha \not \in [z, z \beta],

5303:  \  \ z \beta \not \in [z, z \alpha]$$

5304: 

5305: \noindent

5306: see fig. \ref{14}, b.

5307: Then $z \alpha \beta$ bridges to $t$ in $\ap$

5308: and  $z \alpha \beta \alpha^{m-1}$  bridges

5309: to $\ap$ in $t \alpha^{m-1}$ which is

5310: a point in $(z \alpha^{m}, z \alpha^{m-1})$.

5311: But 

5312: 

5313: $$z \alpha \beta \alpha^{m-1} = z \alpha \tau \ \ \ \Rightarrow \ \ \ 

5314: z \alpha^{-1} \tau^{-1} \ \in \ 

5315: [z, z \alpha) \ \ \ \Rightarrow \ \ \ 

5316: z \beta = z \alpha^{-1} \tau^{-1} \ \in \ [z, z \alpha),$$

5317: 

5318: \noindent

5319: contradiction.

5320: %Then $z \tau^{-1} \beta = z \alpha^{-1} \tau^{-1} \in

5321: %[z, z \alpha)$

5322: %or $z \beta \in [z, z \alpha)$, contradiction.

5323: 

5324: So either $z \beta \in [z, z \alpha]$ or

5325: $z \alpha \in [z, z \beta]$.

5326: 

5327: \vskip .2in

5328: \noindent

5329: {\bf {Situation I}} $-$ $z \alpha$ is in $[z, z \beta]$.

5330: 

5331: Use $z \beta \tau = z \tau \alpha^{-1} = z \alpha^{-1}$.

5332: As $z \alpha$ is in $[z, z \beta]$, then

5333: $z \alpha \tau \in [z, z \beta \tau] = [z, z \alpha^{-1}]$

5334: and $z \alpha \tau \alpha^{1-m} \in [z \alpha^{-m}, z \alpha^{1-m}]$.

5335: But 

5336: 

5337: $$z \alpha \tau \alpha^{1-m} \ = \ 

5338: z \tau^{-1} \alpha \tau \alpha^{1-m} \ = \ z \alpha \beta,

5339: \ \ \ \ \ {\rm so} \ \ \ z \alpha \beta \in [z \alpha^{-m}, z \alpha^{1-m}]

5340: \subset \ap.$$

5341: 

5342: \noindent

5343: We stress that $z \alpha \beta \in \ap$.

5344: Here $z \beta^{-1} \prec z \prec z  \alpha$,

5345: hence $z \prec z \beta \prec z \alpha \beta$.

5346: It follows that 

5347: 

5348: $$z \beta \ \in \ \ap \ \ \ {\rm and} \ \ \

5349: z \beta \ \in \ [z, z \alpha \beta] \ \ \ \Rightarrow \ \ \ 

5350: z \alpha \beta \alpha^{-1} \ \in \

5351: [z \alpha^{m-1}, z \alpha^{-m}].$$

5352: 

5353: 

5354: 

5355: We want $z \gamma = z$ or $z \alpha \beta = z \beta \alpha$.

5356: We first analyse the other two possibilities.

5357: 

5358: \vskip .2in

5359: \noindent

5360: {\bf {Situation I.1}} $-$ $z \alpha \beta \alpha^{-1} > z \beta$ 

5361: in $\ap$.

5362: %, fig. \ref{14}, a.

5363: 

5364: %\blankfig{14}{1.}{The situation $z$ closer.}

5365: 

5366: Then $z \beta \prec z \alpha \beta \alpha^{-1} \prec z \alpha \beta$,

5367: so $z \prec z \gamma \prec z \alpha$, or $z \gamma \in (z, z \alpha)$,

5368: so $z \gamma \in \uu_1$.

5369: Clearly $z \beta \alpha \in \ap$.

5370: %Also $z \alpha \beta \alpha^{-1} \beta^{-1} = z \gamma$ is in

5371: %$[z, z \alpha)$.

5372: Here $z \alpha \beta > z \beta \alpha$ in $\ap$.

5373: %Now $z \alpha \beta$ is in $\lab$,

5374: %$z \alpha \in \lab, z \beta \in \lab$,

5375: Then

5376: 

5377: 

5378: $$z \prec z \beta \alpha \prec z \alpha \beta

5379: \ \ \ {\rm all \ in } \ \ \ap \ \ 

5380: \Rightarrow \ z \beta^{-1} \prec z \beta \alpha \beta^{-1}

5381: \prec z \alpha \ \ \ {\rm and} \ \ 

5382: z \beta^{-1} \alpha^{-1} \prec z \gamma^{-1} \prec z.$$

5383: 

5384: 

5385: \noindent

5386: But $z \beta^{-1} = z \alpha \tau^{-1} \in \uu_2$,

5387: hence $z \beta^{-1} \alpha^{-1}$ is in $\uu_2$.

5388: %$$z \beta \alpha >_{\beta} z \beta

5389: %\ = \ z \beta \alpha \beta^{-1} \pb z$$

5390: %\noindent

5391: %but $z \beta \alpha \beta^{-1} \pb z \beta \alpha$ so

5392: %

5393: %$$z \beta \alpha \beta^{-1} \in \ap

5394: %\ \ \ {\rm and} \ \ \ z \beta \alpha \beta^{-1}

5395: %\pb z \alpha \beta \beta^{-1} = z \alpha.$$

5396: %Then $z \beta \alpha \beta^{-1} \alpha^{-1}

5397: %= z \gamma^{-1} \pa z$

5398: %which implies $z \gamma^{-1}$ is in $\uu_2$.

5399: Now $z \gamma \in \uu_1, \ z \gamma^{-1} \in \uu_2$,

5400: therefore $z$ is in a local axis for $\gamma$, hence

5401: $z \gamma^q \not = z$, contradiction.

5402: 

5403: \vskip .2in

5404: \noindent

5405: {\bf {Situation I.2}} $-$ Suppose  \ $z \alpha \beta \ 

5406: \pa  \ z \beta \alpha$.

5407: 

5408: Then 

5409: 

5410: $$z \ \prec \ z \alpha \beta \alpha^{-1} \ \prec \

5411: z \beta \ \ \ \ \Rightarrow \ \ \ \ 

5412: z \beta^{-1} \ \prec \ z \gamma \ \prec \ z.$$

5413: 

5414: \noindent

5415: As $z \beta^{-1} = z \alpha \tau^{-1}$ is in $\uu_2$,

5416: then $z \gamma$ is in $\uu_2$.

5417: 

5418: %Then $z \alpha \beta \alpha^{-1} \pa z \beta$,

5419: %so $z \alpha \beta \alpha^{-1} \pb z \beta$,

5420: %implying $z \alpha \beta \alpha^{-1} \beta^{-1}

5421: %= z \gamma \pb z $.

5422: %Notice $z \beta^{-1} = z \alpha \tau^{-1}$ is

5423: %in $\uu_2$, so $\ab$ goes into $\uu_2$.

5424: %We obtain $z \gamma$ is in $\uu_2$.

5425: 

5426: Now \ $z \alpha \beta \ \pa  \ z \beta \alpha$. \

5427: If $\ab$ contains elements in $\ap$ above 

5428: $z \alpha \beta$, that is,

5429: $\ab \cap \ap \supset [z,t)$ with

5430: $t >_{\alpha}  z \alpha \beta$.

5431: %, see fig. \ref{14}, c).

5432: Then 

5433: 

5434: $$z \ \prec \ z \alpha \ \prec \ t \beta^{-1} \ \prec \ 

5435: z \beta \alpha \beta^{-1}, \ \ \ {\rm with} \ \

5436: t \beta^{-1} \ \in \ \ap \ \ \ 

5437: \Rightarrow \ \ \ 

5438: z \ \prec \ t \beta^{-1} \alpha^{-1}

5439: \ \prec \ z \gamma^{-1}$$

5440: 

5441: \noindent

5442:  with $t \beta^{-1} \alpha^{-1}$ in

5443: $\ap$ so $z \gamma^{-1}$ is in $\uu_1$ and not in $\uu_2$.

5444: 

5445: %$z \beta \alpha \beta^{-1} > z \alpha$

5446: %that is it escapes beyond $z \alpha$.

5447: %This implies 

5448: %$z \beta \alpha \beta^{-1} \alpha^{-1} = 

5449: %z \gamma^{-1}  \in \uu_1$

5450: %so $z \gamma^{-1}$ is not in $\uu_2$.

5451: 

5452: On the other hand if

5453: $\ab$ escapes $\ap$ in $z \alpha \beta$, then

5454: %(see fig. \ref{14}, c), then

5455: $z \beta \alpha \beta^{-1}$ bridges to

5456: $\ab$ in $z \alpha$,

5457: hence bridges to $\ap$ in $z \alpha$ as $z \alpha \in (z, z \beta)$. 

5458: Hence $z \beta \alpha \beta^{-1} \not \in \uu_2 \alpha$

5459: %because $z \in \ab$%.

5460: and $z \beta \alpha \beta^{-1} \alpha^{-1} = z \gamma^{-1}$

5461: bridges to $\ap$ in $z$ and $z \gamma^{-1}$ is not

5462: in $\uu_2$.

5463: In any case $z \gamma^{-1}$ is not in $\uu_2$

5464: and $z \gamma$ is in $\uu_2$ so

5465: $z$ separates $z \gamma$ from $z \gamma^{-1}$ and

5466: $z$ is in a local axis for $\gamma$, impossible.

5467: 

5468: We conclude that $z \alpha \beta = z \beta \alpha$ 

5469: or that $z \gamma = z$.

5470: 

5471: \vskip .2in

5472: \noindent

5473: {\bf {Situation I.3}} $-$ $z \gamma = z$.

5474: 

5475: Then $\gamma$ leaves invariant the set of components

5476: of $T - \{ z \}$.

5477: Recall that $\uu_1 \tau^{-1} = \uu_2$

5478: and $\uu_1 \tau = \uu_2$ in situation I.

5479: Use 

5480: $z \beta^{-1} \alpha^{-1} \gamma \ = 

5481: \ z \alpha^{-1} \beta^{-1}$.

5482: The left side is

5483: $z \tau \alpha \tau^{-1} \alpha^{-1} \gamma =

5484: z \alpha \tau^{-1} \alpha^{-1} \gamma$.

5485: 

5486: $$z \alpha \in \uu_1 \ \Rightarrow \ 

5487: z \alpha \tau^{-1} \in \uu_1 \tau^{-1} \not = \uu_1,

5488: \ \ \ {\rm so} \ \ \ z \alpha \tau^{-1} \alpha^{-1}

5489: \in \uu_2 \ \ \ {\rm and} \ \ \

5490: z \alpha \tau^{-1} \alpha^{-1} \gamma \in \uu_2 \gamma.$$

5491: 

5492: \noindent

5493: On the other hand the right side is

5494: $z \alpha^{-1} \tau \alpha \tau^{-1}$:

5495: 

5496: $$z \alpha^{-1} \in \uu_2 

5497: \ \Rightarrow \ z \alpha^{-1} \tau \in \uu_2 \tau = \uu_1,

5498: \ \ z \alpha^{-1} \tau \alpha \in \uu_1

5499: \ \ \ {\rm and} \ \ \ z \alpha^{-1} \tau \alpha \tau^{-1}

5500: \in \uu_1 \tau^{-1} = \uu_2.$$

5501: 

5502: \noindent

5503: So $\uu_2 \gamma \cap \uu_2 \not = \emptyset$.

5504: Since $\gamma$ now preserves the set of components

5505: of $T - \{ z \}$ it follows that

5506: $\uu_2 \gamma = \uu_2$

5507: and 

5508: $ \uu_1 \gamma \ = \ \uu_2 \tau \gamma \ = \

5509: \uu_2 \gamma \tau \ = \ \uu_2 \tau \ = \ \uu_1$.

5510: %\noindent

5511: Now we use $p$ odd and $\tau^p \gamma^q = id$:

5512: 

5513: $$\uu_1 \ = \ \uu_1 \gamma^q \tau^p \ = \

5514: \uu_1 \tau^p \ = \ \uu_1 \tau^{p(mod 2)}

5515: \ = \ \uu_1 \tau.$$

5516: 

5517: \noindent

5518: This contradicts $\uu_1 \tau \not = \uu_1$ and

5519: finishes the analysis of situation $I$.

5520: 

5521: \vskip .2in

5522: \noindent

5523: {\bf {Situation II}} $-$ $z \beta \in [z, z \alpha]$.

5524: 

5525: This is very similar to the previous case if we think

5526: of it in the appropriate way. The trick here is to

5527: switch the roles of $\alpha$ and $\beta$, which

5528: can be done.

5529: Notice first that $z \beta \in \uu_1$ and

5530: $z \beta^{-1} = z \tau \alpha \tau^{-1}

5531: = z \alpha \tau^{-1}$ is in $\uu_2$.

5532: So the component of $T - \{ z \}$ containing

5533: $z \beta$ (respectively $z \beta^{-1}$)

5534: is the $\uu_1$ (respectively $\uu_2$).

5535: First rewrite the relations as

5536: 

5537: $$ \tau \alpha \tau^{-1} \ = \ \beta^{-1}

5538: \ \ \ \ \ \ \ \tau \beta \tau^{-1} \ = \

5539: \gamma^{-1} \alpha \beta^{m} \ = \ 

5540: \beta \alpha \beta^{m-1}$$

5541: 

5542: \noindent

5543: As $z \beta$ is in $[z, z \alpha]$

5544: then $z \beta \tau^{-1}$ is in

5545: $[z \tau^{-1}, z \alpha \tau^{-1}] = 

5546: [z, z \beta^{-1}]$.

5547: So 

5548: 

5549: $$z \tau \beta \tau^{-1} \beta^{1-m} \ = \

5550: z \beta \tau^{-1} \beta^{1-m} \ = \ z \beta \alpha

5551: \ \ \in \ \ [z \beta^{-m}, z \beta^{1-m}] \ \subset

5552: \ \ab.$$

5553: 

5554: \noindent

5555: As $z \beta \in [z, z \alpha]$, then $z \beta \alpha$ is

5556: in $[z \alpha, z \alpha^2]$

5557: and 

5558: 

5559: $$z \alpha \in [z, z \beta \alpha] \ \subset \ 

5560: [z, z \beta^{1-m}] \ \subset \ \ab.$$

5561: 

5562: \noindent

5563: Therefore $z \alpha$ is in $\ab$ and similarly

5564: $z \alpha \beta$, $z \beta \alpha$ are in $\ab$. 

5565: 

5566: From this point on the proof is entirely similar

5567: to the analysis in situation I:

5568: consider whether $z \alpha \beta \ \pb \ z \beta \alpha$, \ \

5569: $z \alpha \beta \ >_{\beta}  \ z \beta \alpha$, \

5570: or $z \alpha \beta = z \beta \alpha$,

5571: with completely analogous proofs. 

5572: 

5573: Therefore this case is disallowed. This finishes the

5574: analysis of case B.2.3, \ $\uu_2 \tau = \uu_1$.

5575: 

5576: \vskip .2in

5577: \noindent

5578: {\bf {Case B.2.4}} $-$ $\uu_1 \tau = \uu_2$,

5579: \ $\uu_1 \tau^{-1} \not = \uu_2$.

5580: 

5581: This is the most interesting case which relates

5582: to the topology in a crucial way.

5583: 

5584: 

5585: %\blankfig{17}{1.}{The all possibilities}

5586: 

5587: Use $z \beta^{-1} \alpha^{-1} \gamma = z \alpha^{-1}

5588: \beta^{-1}$.

5589: The right side is $z \tau \alpha \tau^{-1} \alpha^{-1} \gamma

5590: = z \alpha \tau^{-1} \alpha^{-1} \gamma$.

5591: 

5592: $$ z \alpha \in \uu_1 \ \ \Rightarrow \   z \alpha \tau^{-1}

5593: \in \uu_1 \tau^{-1} \not = \uu_1 \ \Rightarrow

5594: \ z \alpha \tau^{-1} \alpha^{-1} \in \uu_2.$$

5595: 

5596: \noindent

5597: Hence $z \beta^{-1} \alpha^{-1} \gamma$ is

5598: in $\uu_2 \gamma$.

5599: On the other hand $z \alpha^{-1} \beta^{-1} =

5600: z \alpha^{-1} \tau \alpha \tau^{-1}$:

5601: 

5602: $$z \alpha^{-1} \tau \ \in \uu_2 \tau \not = \uu_2

5603: \ \Rightarrow z \alpha^{-1} \tau \alpha \in \uu_1 \ \

5604: \Rightarrow \ 

5605: z \alpha^{-1} \tau \alpha \tau^{-1} \in \uu_1 \tau^{-1}

5606: \ \not = \ \uu_2.$$

5607: 

5608: \noindent

5609: We conclude that

5610: 

5611: $$\uu_2 \gamma \ \cap \ \uu_1 \tau^{-1} \ \not = \emptyset,

5612: \ \ \ \ {\rm or} \ \ \ \ 

5613: \uu_1 \tau \gamma \cap \uu_1 \tau^{-1} \not = \emptyset

5614: \ \ \ \ \ \ \ (*).$$

5615: 

5616: \noindent

5617: What we actually want is that these two sets

5618: are equal. A priori we have to be careful because

5619: $\gamma$ may not preserve the set of components

5620: of $T - \{ z \}$, or equivalently we may have

5621: $z \gamma \not = z$. So we first deal with this case.

5622: We will need the following useful lemma:

5623: 

5624: \begin{lemma}{}{}

5625: Let $\eta$ be a homeomorphism of a tree $V$ so that

5626: $\eta^l$ has a fixed point $a$, where $l$ is not $0$.

5627: Then there is a fixed point of $\eta$ in $[a, a \eta]$.

5628: \end{lemma}

5629: 

5630: \begin{proof}{}

5631: Consider $a \eta^2$.

5632: If $a \eta^2$ is in $[a, a \eta]$ and

5633: not equal to $a \eta$, then $\eta$ sends

5634: $[a, a \eta]$ into itself and has a fixed

5635: point there, done.

5636: If $a \eta$ is in $(a, a \eta^2)$ then $a$ is in 

5637: a local axis of $\eta$ and $a \eta^l$ is not $a$,

5638: impossible.

5639: If $a$ is in $(a \eta, a \eta^2)$, then $\eta^{-1}$ sends

5640: $[a \eta, a \eta^2]$ into itself (into $[a, a \eta]$)

5641: producing a fixed point there, done.

5642: 

5643: 

5644: %\blankfig{15}{1.}{

5645: %a.  $r \eta \in [r, a \eta]$,

5646: %b. $r \eta \in (r, a \eta^2]$.}

5647: 

5648: 

5649: \begin{figure}

5650: \centeredepsfbox{lam15.eps}

5651: \caption{

5652: a.  $r \eta \in [r, a \eta]$,

5653: b. $r \eta \in (r, a \eta^2]$.}

5654: \label{15}

5655: \end{figure}

5656: 

5657: 

5658: 

5659: 

5660: 

5661: 

5662: We can now assume $a \eta^2$ bridges to $[a, a \eta]$

5663: in a point

5664: $r$ which is in $(a, a \eta)$, see fig. \ref{15}, a.

5665: If $r \eta = r$ we are done.

5666: Assume $r \eta \not = r$.

5667: Then $r \eta$ is in $[a \eta, a \eta^2]$.

5668: 

5669: Suppose first that 

5670: $r \eta$ is in $[r, a \eta]$, see fig. \ref{15}, a.

5671: Then $r \eta^2$ is in $[r \eta, a \eta^{2}]$ so either

5672: $[r \eta, r]$ is contained in its image 

5673: under $\eta$ or vice versa.

5674: In any case there is a fixed point of 

5675: $\eta$ in $[r, r \eta]$.

5676: %Apply $\eta$ to the tripod with endpoints

5677: %$a \eta, a  \eta^2, a \eta^3$. The tripod $r \eta^2$

5678: %will be in the leg from $a \eta^2$.

5679: %It now follows $[r, r \eta]$ is contained in

5680: %$[r \eta, r \eta^2]$ and there is a fixed point

5681: %of $\eta$ in $[r, r \eta] \subset [a, a \eta]$,

5682: %see fig. \ref{15}, a.

5683: 

5684: Suppose now that $r \eta$ is in $(r, a \eta^2]$ see

5685: fig. \ref{15}, b.

5686: Hence $a \prec r \prec r \eta$ and $a \eta \prec r \eta \prec

5687: r \eta^2$.

5688: Then $r \in (a \eta, r \eta)$ and 

5689: $r \eta \in (r, r \eta^2)$, so $r$ is in a local axis for

5690: $\eta$. 

5691: %Apply $\eta$ to the tripod with endpoints

5692: %$a \eta, a \eta^2, a \eta^3$. Then $r \eta^2$ is in

5693: %the $a \eta^3$ leg. It now follows that

5694: %$r$ is in a local axis of $\eta$ an%I%d 

5695: This implies

5696: that $a \eta^t \not = a$ for any nonzero $t$ in ${\bf Z}$,

5697: contradiction.

5698: This finishes the proof.

5699: \end{proof}

5700: 

5701: We are back to case B.2.4.

5702: 

5703: 

5704: 

5705: \vskip .1in

5706: \noindent

5707: {\bf {Situation I}} $-$

5708: $z \gamma \not = z$.

5709: 

5710: %\vskip .1in

5711: %\noindent

5712: %{\bf {Situation I.1}} $-$ 

5713: Suppose first that $z \gamma \in \uu_2$.

5714: Notice $\uu_2 \tau \not = \uu_1$ and also $\not = \uu_2$.

5715: Then there is $c$ in $[z, z \gamma]$ fixed

5716: by $\gamma$ so $c$ is in $\uu_2$. 

5717: This implies 

5718: 

5719: $$\uu_2 \tau \gamma \ \subset \ 

5720: \uu_2 \ \ \ \Rightarrow \ \ \ 

5721: \uu_1 \tau^2 \gamma \ \subset \ \uu_2, \ \ {\rm or} \ \ 

5722: \uu_1 \tau \gamma \ \subset \ \uu_1.$$

5723: 

5724: \noindent

5725: But by $(*)$

5726: $\uu_1 \tau \gamma \cap \uu_1 \tau^{-1} \not = \emptyset$,

5727: which now implies 

5728: $\uu_1 \tau^{-1} \cap \uu_1 \not = \emptyset$.

5729: This is impossible and rules out this case.

5730: 

5731: %\vskip .1in

5732: %\noindent

5733: %{\bf {Situation I.2}} $-$ 

5734: The second possibility is that $z \gamma \in \uu_1$.

5735: Here $\uu_2 \gamma \subset \uu_1$ so

5736: $\uu_1 \tau \gamma \subset \uu_1$.

5737: As $\uu_1 \tau \gamma \cap \uu_1 \tau^{-1} \not

5738: = \emptyset$ then $\uu_1 \tau^{-1} \cap \uu_1 \not

5739: = \emptyset$, also impossible.

5740: 

5741: The final option is

5742: %\vskip .1in

5743: %\noindent

5744: %{\bf {Situation I.3}} $-$

5745: $z \gamma \not \in \uu_1 \cap \uu_2$, 

5746: $z \gamma \in \uu_3$ (which may be $\uu_2 \tau$ or not).

5747: Here there is $y$ fixed by $\gamma$ with

5748: $y \in \uu_3$. Here first use

5749: 

5750: \noindent

5751: $$\uu_2 \gamma \ \subset \ \uu_3, \ \ {\rm or} \ \ 

5752: \uu_1 \tau \gamma \ \subset \ \uu_3 \ \ \ \Rightarrow \ \ \ 

5753: \uu_1 \tau^{-1}  \ \cap \ \uu_3 \ \not = \ 

5754: \emptyset \ \ {\rm and} \ \ 

5755: \uu_1 \tau^{-1} \ = \ \uu_3.$$

5756: 

5757: \noindent

5758: Use $\uu_1 \gamma \subset \uu_3$, so

5759: 

5760: $$\uu_1 \tau \gamma \subset \uu_3 \tau

5761: \ \ \ {\rm and} \ \ \ 

5762: \uu_1 \tau^{-1} \cap \uu_3 \tau \not = \emptyset

5763: \ \ \ {\rm or} \ \ \ \uu_1 \tau^{-1} = \uu_3 \tau.$$

5764: 

5765: \noindent

5766: Then $\uu_3 = \uu_3 \tau$ or $\uu_1 \tau^{-1} = \uu_1 \tau^{-2}$,

5767: so $\uu_1 \tau = \uu_1$, impossible.

5768: This rules out this final option.

5769: 

5770: We conclude that:

5771: 

5772: \vskip .1in

5773: \noindent

5774: {\bf {Situation II}} $-$ $z \gamma = z$.

5775: 

5776: This is a crucial case. In fact there is an essential

5777: lamination in $M_{p/q}$ whenever $|p - 2q| \geq 2$ and this

5778: essential lamination may satisfy these properties:

5779: $\tau$ has a fixed point, $\alpha$ has an axis (or at

5780: least a local axis) which contains the fixed point of

5781: $\tau$. See more below.

5782: So here is a part of the proof where the

5783: specific condition $|p - 2q| = 1$ needs to be used. See remark

5784: below on the topological significance of this condition.

5785: 

5786: Here is the proof.

5787: Since $z \gamma = z$ , then $\gamma$ permutes components

5788: of $T - \{ z \}$. Since $\uu_1 \tau \gamma \cap

5789: \uu_1 \tau^{-1} \not = \emptyset$, it now follows

5790: that 

5791: 

5792: $$\uu_1 \tau \gamma  \ = \ \uu_1 \tau^{-1}

5793: \ \ \ \ \ {\rm or} \ \ \ \ 

5794: \uu_1 \gamma \tau^2 \ = \ \uu_1.$$

5795: 

5796: \noindent

5797: We now compute

5798: 

5799: $$\uu_1 \ = \ \uu_1 \tau^p \gamma^q \ = \ 

5800: \uu_1 \tau^{p-2q} \tau^{2q} \gamma^q \ = \

5801: \uu_1 (\gamma \tau^2)^q \tau^{p-2q} \ = \

5802: \uu_1 \tau^{p-2q}.$$

5803: 

5804: \noindent

5805: When $|p - 2q| = 1$ then either

5806: $\uu_1 = \uu_1 \tau$ or $\uu_1 = \uu_1 \tau^{-1}$ $-$

5807: so in either case $\uu_1 = \uu_1 \tau$!

5808: But this contradicts that we proved before

5809: that in case B, $\uu_1 \tau$ is not equal to $\uu_1$.

5810: This is a contradiction showing that case B.2.4 cannot happen.

5811: This is quite straightforward, but it needed all

5812: the previous steps.

5813: 

5814: This finishes the proof of case B: $Fix(\tau) \not = \emptyset$,

5815: $Fix(\alpha) = \emptyset$.

5816: 

5817: 

5818: 

5819: \vskip .2in

5820: \noindent

5821: {\bf {Remark}} $-$

5822: We now analyse the topology of this situation.

5823: Consider the original stable foliation in the

5824: torus bundle over the circle (the manifold $M$). This

5825: produces a lamination $\lambda_1$ in $M - N(\delta)$.

5826: The solid torus complementary component of $\lambda_1$

5827: have degeneracy locus $(1,2)$ that is $\gamma \tau^2$.

5828: This means the $\gamma \tau^2$ is a curve in the boundary

5829: leaf of the complementary component and it also

5830: preserves the ``outer" side of this complementary

5831: component. Now do $p/q$ Dehn filling on $M - N(\delta)$

5832: and look at the tree $T$ produced. The leaf

5833: through $\delta$ collapses to a fixed point $z$ of $\tau$

5834: (and $\gamma$ too). Usually neither $\tau$ nor $\gamma$

5835: preserves the complementary components of $z$, but

5836: the above fact about the degeneracy locus means

5837: that $\gamma \tau^2$ does preserve these components $-$

5838: if $\uu_1$ is one such component of $T - \{ z \}$

5839: then $\uu_1 \gamma \tau^2 = \uu_1$

5840: %, that is

5841: %$\gamma$ goes once around and $\tau$ 

5842: %goes only half around in the opposite direction.

5843: After $(q,p)$ Dehn surgery,

5844: the leaf space $T$ of the lamination has a singularity

5845: at $z$ with exactly

5846: $|p - 2q|$ prongs. 

5847: The transformation $\tau$ rotates by one in the set of

5848: prongs, hence

5849: $\tau^{p - 2q}$ preserves each of

5850: the prongs.

5851: This is also detected by $\gamma \tau^2$ preserving the

5852: set of prongs and $\tau^p \gamma^q$ being null homotopic.

5853: All is well when $|p - 2q| \geq 2$, because we have 

5854: 2 or more prongs and the lamination is essential and

5855: the action is very nice.

5856: However when $|p - 2q| = 1$ there is only

5857: one prong and the lamination is not essential.

5858: It is amazing that this sort of difficulty 

5859: can still be detected on

5860: the level of group action on trees.

5861: Notice that this is exactly what the proof shows that

5862: $\uu_1 \tau = \uu_1$, which must happen if there is

5863: only one prong.

5864: 

5865: \vskip .2in

5866: 

5867: \section{Case C $-$ $\alpha$ has a fixed point

5868: and $\tau$ has a fixed point}

5869: 

5870: Let $s$ in $Fix(\kappa)$, $w$ in $Fix(\alpha)$ with

5871: $(s,w] \cap Fix(\kappa) = \emptyset$ and

5872: $[s,w) \cap Fix(\alpha) = \emptyset$.

5873: The following notation will be very useful in

5874: this section. Given $u \not = v$ in $T$ let

5875: 

5876: $$T_u(v) \ = \ \{ {\rm component \ of} \ \  T - \{ u \}

5877: \ \ \ {\rm containing} \ \ v \ \}.$$

5878: 

5879: \noindent

5880: Let $\we = T_s(w)$, $\vv = T_w(s)$.

5881: First in this section we will try to prove that

5882: $\we$ is invariant under $\tau$ and $\vv$ is

5883: invariant under $\alpha$. This will produce local

5884: axes for $\alpha$ and (eventually) for $\tau$ and we will

5885: see how the 2 axes interact.

5886: 

5887: \vskip .2in

5888: \noindent

5889: {\bf {Case C.1}} $-$ Suppose $\we \tau \not = \we$.

5890: 

5891: 

5892: \vskip .2in

5893: \noindent

5894: {\bf {Case C.1.1}} $-$ Suppose $w \in [s, s \alpha]$.

5895: 

5896: 

5897: 

5898: This is equivalent to $\vv \alpha \not = \vv$.

5899: Notice $s \alpha \not = w$. We know $s \alpha \beta = s \beta \alpha$,

5900: and $s \beta \alpha = s \alpha^{-1} \tau^{-1} \alpha$.

5901: Then 

5902: 

5903: $$s \alpha^{-1} \not \in \vv \ \ \Rightarrow \ \ 

5904: s \alpha^{-1} \in \we \ \ \Rightarrow \ \ 

5905: s \alpha^{-1} \tau^{-1}  \in  

5906: \we \tau^{-1} \subset \vv \ \ \Rightarrow \ \ 

5907: s \alpha^{-1} \tau^{-1} \alpha \in \vv \alpha \subset \we.$$

5908: 

5909: 

5910: 

5911: On the other hand

5912: $s \alpha \beta = s \alpha \tau \alpha^{-1} \tau^{-1}$.

5913: Here 

5914: 

5915: $$s \alpha\ \in \ \vv \alpha \ \subset \ \we \ \ \ \Rightarrow \ \ \

5916: s \alpha \tau \ \in \ \we \tau \ \subset \ \vv \ \ \ \Rightarrow \ \ \ 

5917: s \alpha \tau^{-1} \alpha^{-1} \ \in \ 

5918: \vv \alpha^{-1} \ \subset \ \we$$ 

5919: 

5920: 

5921: \noindent

5922: and $s \alpha \beta$ is

5923: in $\we \tau$.

5924: These two facts together imply $\we = \we \tau$,

5925: contrary to assumption.

5926: 

5927: Conclusion: if $\we \tau \not = \we$, then $\vv \alpha = \vv$.

5928: 

5929: \vskip .2in

5930: \noindent

5931: {\bf {Case C.1.2}} $-$ $s \alpha^{-1} \not \in [s,w]$,

5932: \ $s \alpha \not \in [s, w]$.

5933: 

5934: This implies 

5935: $s \alpha, s \alpha^{-1}$ are in $\we$.

5936: For otherwise if $s \alpha$ is not in $\we$, then

5937: $s$ is in $(w,s \alpha]$ and

5938: so $s \alpha^{-1}$ is in $[w,s]$.

5939: 

5940: In this case $s \alpha^{-1}$ bridges to $[s,w]$ in

5941: a point $r$ with $r \in (s,w)$ $-$ the important

5942: fact

5943: is that

5944: $r$ is not one of the endpoints which would occur

5945: if $s \alpha^{-1}$ is not in $\we$ or $\vv$.

5946: Then 

5947: 

5948: $$r \ \in \ [w, s] \ \cap \ [w, s \alpha^{-1}] \ \ \ \Rightarrow \ \ \ 

5949: r \alpha^{-1} \ \in \

5950: [s, w \alpha^{-1}].$$

5951: 

5952: \noindent

5953: Notice $r \alpha^{-1}$ is not equal to $r$.

5954: If $r \alpha^{-1}$ is in $(r, s \alpha^{-1})$,

5955: then $s \alpha^{-2}$ bridges to $[r, s \alpha^{-1}]$

5956: in $r \alpha^{-1}$, hence $s \alpha^{-2}$ bridges

5957: to $[s,w]$ in $r$. The same happens for all

5958: $s \alpha^n$ with $n$ negative.

5959: If on the other hand $r \alpha^{-1}$ is in $(w,r)$

5960: then $s \alpha^{-2}$ bridges to $r \alpha^{-1}$

5961: in $[s,w]$ and $s \alpha^n$ bridges to $[s,w]$

5962: in $r \alpha^{n+1}$ for all $n$ negative. Notice

5963: then $r \alpha^n$ are all in $(w,r) \subset (w,s)$.

5964: The important conclusion is that under the hypothesis

5965: $s \alpha, s \alpha^{-1}$ both not in $[s,w]$ then

5966: any $s \alpha^n$ bridges to $[s,w]$ in a point

5967: in the interior of $[s,w]$, Hence all $s \alpha^n$

5968: are in $\we$ and $\vv$.

5969: 

5970: 

5971: Use $s \tau^{-1} \alpha \tau  = s \gamma \beta \alpha^m$.

5972: Here $s \alpha$ is in $\we$, so $s \alpha \tau$ is in

5973: $\we \tau$.

5974: Also $s \beta = s \alpha^{-1} \tau^{-1}$ is in $\we \tau^{-1}$

5975: and bridges to $s$ in $[s,w]$. Hence

5976: $s \beta \alpha^m$ bridges to $s \alpha^m$ in

5977: $[s \alpha^m, w]$. 

5978: But $s \alpha^m$ is in

5979: $\we$ and bridges to $[s,w]$ in

5980: a point in the interior of $(s,w)$.

5981: This implies $s \beta \alpha^m$ is in $\we$,

5982: contradiction.

5983: 

5984: This case is impossible.

5985: 

5986: \vskip .2in

5987: \noindent

5988: {\bf {Case C.1.3}} $-$ Suppose $s \alpha \in [s,w]$.

5989: 

5990: This implies for instance that $\we \alpha \subset \we$ and \

5991: $T_s(w \tau^{-1}) \beta^{-1} \ \subset \ T_s(w \tau^{-1})$.

5992: 

5993: \vskip .2in

5994: \noindent

5995: {\bf {Case C.1.3.1}} $-$ Suppose $s \alpha^{-1} \in \we \tau$.

5996: 

5997: Then $s \beta^{-1} = s \alpha \tau^{-1}$

5998: is in $(s, w \tau^{-1}) \subset \we \tau^{-1}$.

5999: % \cup \{ s \}$.

6000: Also $s \alpha^{-1} = s \beta \alpha^{-1} \beta^{-1}$.

6001: Here $s \beta = s \alpha^{-1} \tau^{-1}$ is in $\we$.

6002: 

6003: In this case suppose first that 

6004: $s \beta$ is not in $\vv$. 

6005: Then 

6006: 

6007: 

6008: $$w \ \in \

6009: [w \tau^{-1}, s \beta] \ \ \ {\rm and} \ \ \

6010: w \beta^{-1} \ \in \ [w \tau^{-1}, s] \ \ \ \Rightarrow \ \ \

6011: w \beta^{-1} \alpha^{-1} \ \in \ \we \tau,$$

6012: 

6013: \noindent

6014: as $s \alpha^{-1}$ is in $\we \tau$.

6015: Notice $w \beta^{-1} \alpha^{-1}$ is not $s$.

6016: Then

6017: 

6018: $$w \beta^{-1} \alpha^{-1} \gamma \ = \

6019: w \alpha^{-1} \beta^{-1} \ = \ 

6020: w \beta^{-1} \ \ \ {\rm is \ in} \ \  \we \tau^{-1}.$$

6021: 

6022: \noindent

6023: Notice if $w \beta^{-1} = s$, then

6024: 

6025: 

6026: $$ w \beta^{-1} \alpha^{-1} \  = \ 

6027: w \beta^{-1} \gamma^{-1} \ = \ s \gamma^{-1}  \ =  \ s

6028: \ = \ w \beta^{-1},$$

6029: 

6030: 

6031: \noindent

6032: contradiction

6033: because $s$ is not fixed by $\alpha$.

6034: 

6035: Collecting all of this together: $w \beta^{-1} \alpha^{-1}

6036: \gamma$ is in $\we \tau \gamma$. This point is equal

6037: to $w \beta^{-1}$ which is in $\we \tau^{-1}$.

6038: Therefore 

6039: 

6040: $$\we \tau \gamma  \ = \ \we \tau^{-1} \ \ \ {\rm or} \ \ \ 

6041: \we \tau^2 \gamma = \we, \ \ \ \ \ {\rm impossible \ when} \

6042: |p - 2q| = 1,$$

6043: 

6044: \noindent

6045: as in case B.2.4.

6046: 

6047: The second option in case C.1.3.1 is that

6048: $s \beta \in \vv$.

6049: Recall that $s \alpha^{-1} \tau^{-1} = s \beta$ is in

6050: $\we$.

6051: Notice that 

6052: 

6053: $$\lab \ = \ (\lal) \tau^{-1} \ \ \ {\rm has \ a \ segment} \ \ 

6054: [w \tau^{-1}, s] \ \subset \ 

6055: \we \tau^{-1} \cup \{ s \}$$

6056: 

6057: 

6058: \noindent

6059:  and

6060: then it goes into $\we$, as $s \beta$ is in $\we$.

6061: Then either $s \beta = t \in (w,s)$ or

6062: $s \beta$ bridges to $[w,s]$ in $t \in (w,s)$,

6063: so bridges to $t$ in $\lal$.

6064: %Here $s \alpha^{-1}$ is in $\we \tau$,

6065: %$s \alpha^{-1} \gamma = s \beta \alpha^{-1} \beta^{-1}$,

6066: %see fig. \ref{16}, a.

6067: In either case $s \beta \alpha^{-1}$ bridges 

6068: to $t \alpha^{-1}$ in $\lal$ or is $t \alpha^{-1}$.

6069: If $t \alpha^{-1}$ is in $[w,s)$, then

6070: $s \beta \alpha^{-1}$ bridges to $t \alpha^{-1}$

6071: in $\lab$, see fig. \ref{16}, a. Here $t \alpha^{-1}$ is

6072: in $[w \tau^{-1}, s \beta)$.

6073: If

6074: 

6075: $$s \ \in \ [t \alpha^{-1}, w] \ \ \ {\rm then} \ \ \

6076: s \beta \alpha^{-1} \ \ \ {\rm bridges \ to} \ \ 

6077: \lab \ \ {\rm in} \ \ r, \ \ {\rm with} \ \ 

6078: r \ \in \ [s,w \tau^{-1}].$$

6079: 

6080: 

6081: \noindent

6082: This depends for instance on

6083: whether $\we\tau = \we \tau^{-1}$ or not.

6084: In any case 

6085: $s \beta \alpha^{-1}$ bridges to $\lab$ in a point

6086: in $[w \tau^{-1}, s \beta)$.

6087: It follows that 

6088: $s \beta \alpha^{-1} \beta^{-1}$ bridges

6089: to a point $t$ in $\lab$ with $t$ in $[w \tau^{-1}, s)$,

6090: that is, $s \beta \alpha^{-1} \beta^{-1}$ is in

6091: $\we \tau^{-1}$.

6092: Then

6093: 

6094: $$ s \alpha^{-1} \ \in \ \we \tau, \ \

6095: s \beta \alpha^{-1} \beta^{-1} \ = \ 

6096: s \alpha^{-1} \gamma \ \in \ \we \tau^{-1}

6097: \ \ \  \Rightarrow \ \ 

6098: \we \tau \gamma  \ = \  \we \tau^{-1},$$

6099: 

6100: \noindent

6101: contradiction when $|p - 2q| = 1$.

6102: 

6103: This shows that case C.1.3.1 cannot occur.

6104: 

6105: %\blankfig{16}{1.}{

6106: %a. Case C.1.3.1, b. Case C.1.3.2.}

6107: 

6108: 

6109: 

6110: 

6111: \begin{figure}

6112: \centeredepsfbox{lam16.eps}

6113: \caption{

6114: a. Case C.1.3.1, b. Case C.1.3.2.}

6115: \label{16}

6116: \end{figure}

6117: 

6118: 

6119: \vskip .2in

6120: \noindent

6121: {\bf {Case C.1.3.2}} $-$ $s \alpha^{-1}$ is not in $\we \tau$.

6122: 

6123: Here $s \beta = s \alpha^{-1} \tau^{-1}$ is not in $\we$.

6124: Also $s \beta^{-1} = s \alpha \tau^{-1}$ is not in $\we$.

6125: It follows that 

6126: 

6127: $$\lab  \ \cap \ [w,s]  \ = \ \{ s \},$$

6128: 

6129: \noindent

6130: so $s \alpha$ bridges to

6131: $\lab$ in $s$ and 

6132: $s \alpha \beta = s \beta \alpha$ bridges to $\lab$ in $s \beta$.

6133: Hence $[s, s \beta] \subset (s \alpha, s \beta \alpha)$

6134: and there is a fixed point $r$ of $\alpha$ 

6135: in $(s, s \beta)$, see fig. \ref{16}, b.

6136: It also implies that 

6137: 

6138: $$s \alpha^{-1} \in 

6139: [s, r] \ \ \ {\rm and} \ \ \ 

6140: s \alpha^{-1} \ \in  \ T_s (s \beta) \ = \ 

6141: T_s(s \beta)   \tau,$$

6142: 

6143: \noindent

6144: because $s \beta^{-1} = s \alpha \tau^{-1}$.

6145: Now  apply

6146: $\tau \alpha \beta = \alpha \tau \alpha^{1-m}$ to

6147: $r$: \

6148: $r \tau \alpha \beta = r \tau \alpha^{1-m}$.

6149: 

6150: As $s \beta \tau = s \alpha^{-1}$ and as

6151: $r \in (s, s \beta)$, then

6152: 

6153: $$r \tau \ \in \ (s, s \alpha^{-1})

6154: \ \  \Rightarrow \ \ 

6155: r \tau \alpha \ \in \ (s, s \alpha) \ \ \Rightarrow \ \

6156: r \tau \alpha \beta  \ \in  \ (s \beta, s \alpha \beta)

6157: \ \subset \ T_r(s \beta).$$

6158: 

6159: \noindent

6160: As $r \tau \alpha$ is in $(s, s \alpha^{-1}) \subset

6161: T_r(s)$ this implies $r \tau \alpha^{1-m}$ is also

6162: in $T_r(s)$.

6163: %This also implies that $u \tau \alpha^{1-m} \in 

6164: %(w, u) \subset T_u(s)$.

6165: %But $u \tau \alpha^{1-m} \in T_u(s)$. 

6166: Therefore 

6167: $r$ separates $r \tau \alpha^{1-m}$ from

6168: $r \tau \alpha \beta$,

6169: contradiction.

6170: 

6171: This shows that case C.1.3, $s \alpha \in [s,w]$ cannot

6172: occur.

6173: Finally consider:

6174: 

6175: 

6176: 

6177: \vskip .2in

6178: \noindent

6179: {\bf {Case C.1.4}} $-$ Suppose $s \alpha^{-1} \in [s,w]$.

6180: 

6181: This implies that $\we \alpha^{-1} \subset \we$ and

6182: \ $(\we \tau^{-1}) \beta \ \subset \ (\we \tau^{-1})$.

6183: 

6184: 

6185: \vskip .2in

6186: \noindent

6187: {\bf {Case C.1.4.1}} $-$ Suppose $s \alpha \not \in \we \tau^{-1}$.

6188: 

6189: 

6190: This case is very similar to case C.1.3.2.

6191: Here $s \beta \in T_s(w \tau^{-1})$ which is not equal

6192: to either $T_s(s \alpha)$ or $T_s(s \alpha^{-1})$.

6193: Hence $s \beta$ bridges to $\lal$ in $s$ and $s \beta \alpha

6194: = s \alpha \beta$ bridges to $\lal$ in $s \alpha$.

6195: Hence 

6196: 

6197: $$s \beta \ \prec \ s \ \prec \ s \alpha \ \prec \ s \alpha \beta$$

6198: 

6199: \noindent

6200: and there is a fixed point $r$ of $\beta$ in

6201: $(s, s \alpha)$.

6202: Then $s \beta^{-1} \in (s, r) \subset (s, s \alpha)$.

6203: Now use $\beta \tau^{-1} \beta^{1-m} = \tau^{-1} \beta \alpha$

6204: applied to $r$: \ $r \tau^{-1} \beta^{1-m} = r \tau^{-1} 

6205: \beta \alpha$. As $s \alpha \tau^{-1} = s \beta^{-1}$ then

6206: 

6207: $$r \tau^{-1} \ \in  \ (s, s \beta^{-1})

6208: \ \ \ \ \

6209: {\rm so} \ \ \ \ \ r \tau^{-1} \beta^{1-m} \ \in \ 

6210: (r, s \beta^{1-m}) \ \subset \ T_r(s).$$

6211: 

6212: \noindent

6213: On the other hand $r \tau^{-1} \beta \alpha$ is in

6214: $(s \alpha, s \beta \alpha) \subset T_r(s \alpha)$.

6215: As $T_r(s \alpha) \not = T_r(s)$, this is a contradiction,

6216: ruling out this case.

6217: 

6218: 

6219: 

6220: %The point 

6221: %$s \beta^{-1} = s \alpha \tau^{-1}$ is not in $T_s(s \alpha)$,

6222: %so $s \alpha$ bridges to $s$ in $\lab$.

6223: %Then $s \alpha \beta$ bridges to $s \beta$ in $\lab$ so

6224: %to $s$ in $\lal$, see fig. \ref{18}, a.

6225: %Then $s \alpha \beta \alpha^{m-1}$ bridges to

6226: %$s \alpha^{m-1}$ in $\lal$ hence 

6227: %$s \alpha \beta \alpha^{m-1}$ is in $\we$. But then

6228: %$s \tau^{-1} \alpha \tau = s \alpha \tau$ is in $\we$,

6229: %so $s \alpha \in \we \tau^{-1}$, contrary to hypothesis.

6230: 

6231: 

6232: \vskip .2in

6233: \noindent

6234: {\bf {Case C.1.4.2}} $-$

6235: $s \alpha$ is in $\we \tau^{-1}$.

6236: 

6237: This is similar to case C.1.3.1.

6238: Suppose first that  $\we \tau^{-1} = \we \tau$.

6239: Then $s \alpha \tau^{-1} = s \beta^{-1}$ is

6240: in $\we$.

6241: Also $\we \beta^{-1}$ is contained in $\we$.

6242: It follows that

6243: 

6244: $$s \alpha^{-1} \beta^{-1} \ \in \ \we

6245: \ \ \ {\rm and} \ \ \ 

6246: s \alpha^{-1} \beta^{-1} \gamma^{-1} \ = \ s \beta^{-1} 

6247: \alpha^{-1} \ \in \ \we.$$

6248: 

6249: \noindent

6250: Hence $\we \gamma = \we$, $\we \tau^2 = \we$, leading

6251: to contradiction when $p$ is odd.

6252: 

6253: Suppose now that 

6254: $\we \tau^{-1} \not = \we \tau$.

6255: Then $s \alpha \in \we \tau^{-1}$ and $s \alpha \tau^{-1}

6256: = s \beta^{-1}$ is not in $\we$.

6257: Also $s \beta^{-1}$ is in $\we \tau^{-2}$.

6258: So $s \beta^{-1}$ bridges to $s$ in $\lal$ and

6259: $s \beta^{-1} \alpha^{-1}$ bridges to

6260: $s \alpha^{-1}$ in $\lal$

6261: implying $s \beta^{-1} \alpha^{-1}$ is in $\we$.

6262: 

6263: Also $s \beta^{-1} \alpha^{-1} \gamma = s \alpha^{-1} \beta^{-1}$.

6264: Here $s \alpha^{-1}$ bridges to $s$ in $\lab$,

6265: $s \alpha^{-1} \beta^{-1}$ bridges to $s \beta^{-1}$ in 

6266: $\lab$.

6267: But 

6268: 

6269: $$s \beta^{-1} \ \in \ \we \tau^{-2} \ \ \ \Rightarrow \ \ \ 

6270: s \alpha^{-1} \beta^{-1} \ \in \ \we \tau^{-2} \ \ \ \Rightarrow \ \ \ 

6271: \we \gamma \ = \ \we \tau^{-2},$$

6272: 

6273: 

6274: \noindent

6275: again impossible

6276: when $|p-2q| = 1$.

6277: 

6278: \vskip .1in

6279: This finishes the analysis of case C.1.4,  \

6280: $s \alpha^{-1} \in [s,w]$.

6281: 

6282: We conclude that case C.1, $\we \tau \not = \we$ is

6283: impossible. This implies $\we \tau = \we$.

6284: We stress that this does not yet produce

6285: a local axis of $\tau$ in $\we$, because

6286: we may have other fixed points of $\tau$ in

6287: $(s,w)$. 

6288: 

6289: \vskip .2in

6290: \noindent

6291: {\bf {Case C.2}} $-$ Suppose that $\vv \alpha \not = \vv$.

6292: 

6293: Here we will use $s \alpha \tau = s \beta \alpha^m 

6294: = s \alpha \beta \alpha^{m-1}$ many times.

6295: 

6296: \vskip .2in

6297: \noindent

6298: {\bf {Case C.2.1}} $-$ Suppose $w \tau, w \tau^{-1}$ are

6299: not in $[s,w]$.

6300: 

6301: The bridge from $w \tau$ to $[s,w]$ is

6302: $[w \tau, t]$, where $t$ is in $(s,w)$.

6303: Since $s \alpha \not \in \vv$, then

6304: $s \alpha \tau$ bridges to $t$ in $[s,w]$,

6305: so $s \alpha \tau$ is in $\vv$.

6306: Hence $s \alpha \tau \alpha^{-m}$ is 

6307: in $\vv \alpha^{-m}$.

6308: This point is equal to $s \beta = s \alpha^{-1} \tau^{-1}$.

6309: In the same way $s \alpha^{-1}$ is not in $\vv$ and

6310: bridges to $[s,w]$ in $w$. It follows that 

6311: $s \alpha^{-1} \tau^{-1}$ bridges to a point $r$

6312: in $[s,w]$, where $r$ is in

6313: in $(s,w)$, hence $s \beta \in \vv$.

6314: Therefore $\vv \alpha^m = \vv$.

6315: 

6316: On the other hand

6317: 

6318: 

6319: $$s \alpha \tau \ = \ s \alpha \beta \alpha^{m-1} \ = \ 

6320: s \alpha \tau \alpha^{-1} \tau^{-1} \alpha^{m-1}.$$

6321: 

6322: \noindent

6323: The point $s \alpha \tau$ is in $\vv$ and

6324: bridges to $t$ in $[s,w]$. So $s \alpha \tau \alpha^{-1}$

6325: is in $\vv \alpha^{-1}$ and bridges to $w$ in 

6326: $[s,w]$ so $s \alpha \tau \alpha^{-1} \tau^{-1}$ bridges

6327: to $r$ in $[s,w]$ ($r$ as above) and as a result

6328: this point is in $\vv$. Hence $s \alpha \beta \alpha^{m-1}$

6329: is in $\vv \alpha^{m-1}$ and

6330: $\vv \alpha^m = \vv \alpha^{m-1}$, contradicting

6331: $\vv \alpha \not = \vv$.

6332: 

6333: %\blankfig{17}{1.}{

6334: %a. Case C.2.2.1, b. Case C.2.2.3.}

6335: 

6336: 

6337: 

6338: \begin{figure}

6339: \centeredepsfbox{lam17.eps}

6340: \caption{

6341: a. Case C.2.2.1, b. Case C.2.2.3.}

6342: \label{17}

6343: \end{figure}

6344: 

6345: 

6346: 

6347: 

6348: \vskip .2in

6349: \noindent

6350: {\bf {Case C.2.2}} $-$ $w \tau^{-1} \in [s,w]$.

6351: 

6352: 

6353: Here $\vv \tau^{-1}$ is contained in $\vv$.

6354: 

6355: The condition implies that 

6356: $w$ is in a local axis $\lat$ of $\tau$

6357: (this case will be ruled out, we only establish

6358: the existence of a local axis of $\tau$ in $\we$ later).

6359: Put an order $<$ in $\lat$ so 

6360: $c < d$ in $\lat$ in $\lat$ if $s \prec c \prec d$ -

6361: the order decreases

6362: as points get closer to $s$.

6363: 

6364: \vskip .2in

6365: \noindent

6366: {\bf {Case C.2.2.1}} $-$

6367: $w \tau \in \vv \alpha$, \ $w \tau \not \in \vv \alpha^{-1}$,

6368: see fig. \ref{17}, a.

6369: 

6370: Here $\vv \alpha \tau \subset \vv \alpha$.

6371: 

6372: The conditions imply in particular that

6373: $\vv \alpha \not = \vv \alpha^{-1}$.

6374: Here $s \alpha \tau \in \vv \alpha$,

6375: so $s \beta \alpha^m \in \vv \alpha$. 

6376: Also $s \alpha^{-1}$ bridges to

6377: $\lat$ in $w$

6378: so $s \beta = s \alpha^{-1} \tau^{-1}$ bridges

6379: to $\lat$ in  $w \tau^{-1}$.

6380: It follows that $s \beta$ is in $\vv$ and

6381: $s \beta \alpha^m$ is in $\vv \alpha^m$.

6382: Hence $\vv \alpha^m = \vv \alpha$.

6383: 

6384: On the other hand

6385: $s \alpha \tau = s \alpha \beta \alpha^{m-1}$.

6386: Use 

6387: $s \alpha \beta \ = \ s \alpha \tau \alpha^{-1} \tau^{-1}$.

6388: Here

6389: 

6390: 

6391: $$s \alpha \in \vv \alpha \ \ \Rightarrow \ \ 

6392: s \alpha \tau \in \vv \alpha \ \ \Rightarrow \ \ 

6393: s \alpha \tau \alpha^{-1} \in \vv \ \ \Rightarrow \ \ 

6394: s \alpha \tau \alpha^{-1} \tau^{-1} \in \vv.$$

6395: 

6396: \noindent

6397: Finally $s \alpha \beta \alpha^{m-1}$ is in

6398: $\vv \alpha^{m-1}$. So $\vv \alpha^{m-1} = \vv \alpha$

6399: and $\vv = \vv \alpha$, again contradicting the

6400: assumption

6401: in this case.

6402: 

6403: \vskip .2in

6404: \noindent

6405: {\bf {Case C.2.2.2}} $-$

6406: Suppose $w \tau$ is not in $\vv \alpha$ and

6407: $w \tau$ is not in $\vv \alpha^{-1}$.

6408: 

6409: Then $w \tau$ is in $\re$ another  component

6410: of $T - \{ w \}$.

6411: Then $s \alpha \tau$ is in $\re$.

6412: Now $s \beta \alpha^m = s \alpha^{-1} \tau^{-1} \alpha^m$.

6413: But

6414: 

6415: $$w \tau \ \not \in \ \vv \alpha^{-1} \ \ \ \Rightarrow \ \ \ 

6416: s \alpha^{-1} \ \  \ {\rm bridges \ to} \ \ 

6417: \lat \ \ {\rm in} \ w \ \ \ \Rightarrow \ \ \

6418: s \alpha^{-1} \tau^{-1} \ \ {\rm bridges \ to}  \ \ \lat 

6419: \ \ {\rm in } w \tau^{-1}$$

6420: 

6421: 

6422: \noindent

6423: and

6424: $s \beta$ is in $\vv$. Therefore

6425: $s \beta \alpha^m \in \vv \alpha^m = \re$.

6426: Notice $\re \alpha^{-1}

6427: \not = \re$ because $\re = \vv \alpha^m$ and

6428: $\vv \alpha^{-1} \not = \vv$.

6429: Use 

6430: 

6431: $$s \alpha \tau \ = \ s \alpha \beta \alpha^{m-1} \ = \

6432:  s \alpha \tau \alpha^{-1} \tau^{-1} \alpha^{m-1}

6433: \ \ \ \ {\rm and} \ \ \ \ 

6434: s \alpha \tau \alpha^{-1}  \ \in 

6435: \re \alpha^{-1}

6436: \not = \re.$$

6437: 

6438: \noindent

6439: Hence $s \alpha \tau \alpha^{-1}$ bridges

6440: to $\lat$ in a point $\leq w$ in $\lat$ 

6441: (it is in $[s,w]$) and $s \alpha \beta$

6442: bridges to $\lat$ in

6443: a point $\leq w \tau^{-1}$ in $\lat$.

6444: Hence 

6445: 

6446: $$s \alpha \beta \in

6447: \vv \ \ \ \Rightarrow \ \ \ s \alpha \beta \alpha^{m-1}

6448: \in \vv \alpha^{m-1} \ \ \  \ \Rightarrow \ \  \ \ 

6449: \vv \alpha^m = \vv \alpha^{m-1},$$ 

6450: 

6451: \noindent

6452: contradiction.

6453: Notice that here it doesn't matter whether

6454: $\vv \alpha = \vv \alpha^{-1}$ or not.

6455: 

6456: \vskip .2in

6457: \noindent

6458: {\bf {Case C.2.2.3}} $-$ $w \tau$ is in $\vv \alpha^{-1}$,

6459: see fig. \ref{17}, b.

6460: 

6461: This implies $\vv \alpha^{-1} \tau$ is a subset of $\vv \alpha^{-1}$.

6462: 

6463: Use $s \alpha \tau = s \beta \alpha^m = s \alpha^{-1}

6464: \tau^{-1} \alpha^m$. 

6465: Here

6466: 

6467: $$s \alpha \not \in \vv \ \ \ \Rightarrow \ \ \ 

6468: s \alpha \tau \in T_w(w \tau) = \vv \alpha^{-1}

6469: \ \ \ \Rightarrow \ \ \ 

6470: s \alpha \tau \alpha^{-1} \ \in \ \vv \alpha^{-2} 

6471: \not = \vv \alpha^{-1},$$

6472: 

6473: 

6474: \noindent

6475: so it bridges to a point $r$ in $\lat$ with $r \leq w$ in $\lat$.

6476: Hence $s \alpha \beta$ is in $\vv$ and

6477: $s \alpha \beta \alpha^{m-1}$ is in $\vv \alpha^{m-1}$.

6478: Hence $\vv \alpha^{m-1} = \vv \alpha^{-1}$ or

6479: $\vv \alpha^m = \vv$.

6480: 

6481: %\blankfig{20}{1.}{a. Case C.2.2.3, b. ?????}

6482: 

6483: On the other hand 

6484: $s \alpha \tau = s \beta \alpha^m$

6485: is in $\vv \alpha^{-1}$,

6486: so 

6487: 

6488: $$s \alpha^{-1} \tau^{-1} \ = \

6489:  s \beta  

6490: \ \ \ \ {\rm is \ in} \ \ \ \ 

6491: %$s \alpha^{-1} \tau^{-1}$ is 

6492: \vv \alpha^{-1-m}  \ =  \

6493: \vv \alpha^{-1}.$$

6494: 

6495: % \ \ \ {\rm and} \ \ \ 

6496: %s \alpha^{-1} \in \vv \alpha^{-1} \tau.$$

6497: 

6498: 

6499: \noindent

6500: Then $s \beta$

6501: bridges to a point 

6502: %is in $\vv \alpha^{-1}$ so it bridges to a point

6503: $> w$ in $\lat$. But $s \beta = s \alpha^{-1} \tau^{-1}$,

6504: so $s \alpha^{-1}$ bridges to a point $> w \tau$ in

6505: $\lat$, which implies $w \tau \in (w, s \alpha^{-1})$.

6506: It follows that $w \tau \alpha \in (w, s)$ 

6507: and

6508: $w \beta^{-1} = w 

6509: \tau \alpha \tau^{-1}$ is in $(w \tau^{-1}, s)$

6510: is in $\we$ and in $\vv$.

6511: 

6512: The following arguments use the strategy of case R.2:

6513: 

6514: Now $w \beta^{-1} \alpha^{-1} \gamma = w \beta^{-1}$ is in

6515: $\we$ and 

6516: $w \beta^{-1} \alpha^{-1} = w \tau^{-1} \alpha^{-1} \tau$.

6517: Use 

6518: 

6519: $$w \tau^{-1} \alpha^{-1} \ \in \ \vv \alpha^{-1} \

6520:  = \ T_w(w \tau) \ \ \ {\rm so} \ \ \ 

6521: w  \tau^{-1} \alpha^{-1} \tau \ \ \ {\rm is \ in} \ 

6522: T_{w \tau}(w \tau^2) = \vv \alpha^{-1} \tau^{-1}

6523: \subset \vv \alpha^{-1} \ \subset \ \we.$$

6524: 

6525: \noindent

6526: %bridges to

6527: %a point $> w \tau$ in $\lat$. 

6528: Hence

6529: $w \beta^{-1} \alpha^{-1} \in \we$.

6530: From this it follows that $\we \gamma = \we$.

6531: As usual this implies that $(\lat) \gamma = \lat$

6532: so $\gamma, \tau$ have the common local axis $\lat$.

6533: In addition $w \beta^{-1} \alpha^{-1} \gamma = w \beta^{-1}$

6534: and as $w \beta^{-1}$ is in $\lat$, so does

6535: %so $w \beta^{-1} \in \lat$ (through $w$) and

6536: $w \beta^{-1} \alpha^{-1}$.

6537: 

6538: %Notice that $w \tau \in [w, s \alpha^{-1}]$,

6539: %hence $w \tau \alpha \in [w,s]$.

6540: If $w \tau \alpha \leq w \tau^{-1}$ in $\lat$ then

6541: $w \beta^{-1} = w \tau \alpha \tau^{-1} \leq w \tau^{-2}$

6542: in $\lat$. 

6543: Also $w \tau, w \beta^{-1} \alpha^{-1}$ are in $\lat$

6544: and $w \tau < w \beta^{-1} \alpha^{-1}$ in $\lat$.

6545: Hence 

6546: 

6547: $$w \tau \gamma \ < \ w \beta^{-1} \alpha^{-1} \gamma \ = \ 

6548: w \beta^{-1} \leq w \tau^{-2} 

6549: \ \ \ {\rm in } \ \ \lat \ \ \ \ \Rightarrow \ \

6550: p > 3q,$$

6551: 

6552: \noindent

6553: contradiction to $|p-2q| = 1$.

6554: 

6555: If $w \tau \alpha > w \tau^{-1}$ in $\lat$ then 

6556: $w \tau \alpha \tau^{-1} = w \beta^{-1} \in (w \tau^{-2}, 

6557: w \tau^{-1})$. 

6558: Here use 

6559: 

6560: $$(w \tau^2) \gamma \beta \alpha^m \ = \  w \tau \alpha \tau

6561: \ 

6562: \in T_w(w \tau) = \vv \alpha^{-1} \ \ \ \Rightarrow \ \ 

6563: w \tau^2 \gamma \beta \ \in \vv \alpha^{-1},$$

6564: 

6565: 

6566: 

6567: %As $w \beta \in \vv$, then $w \beta \alpha^{m-1} \in \vv

6568: %\alpha^{m-1} = \vv \alpha^{-1}$, so we conclude

6569: %%$\vv \alpha^m = \vv$. It follows that

6570: %$w \tau^2 \gamma \beta \alpha^m \in \vv \alpha^{-1}$ 

6571: %and so 

6572: \noindent

6573: because 

6574: $\vv \alpha^m = \vv$.

6575: Therefore $w \tau^2 \gamma \beta$ bridges to $v$ in

6576: $\lat$ with $v > w$ in $\lat$.

6577: Hence $w \tau^2 \gamma < w \beta^{-1}$ in $\lat$ and

6578: as $w \beta^{-1} < w \tau^{-1}$ we also obtain

6579: $p > 3q$, contradiction.

6580: 

6581: This rules out the case C.2.2.3 and hence finishes

6582: the analysis of case C.2.2, \ $w \tau^{-1} \in [s,w]$.

6583: The next case is:

6584: 

6585: \vskip .2in

6586: \noindent

6587: {\bf {Case C.2.3}} $-$ $w \tau \in [s,w]$.

6588: 

6589: This implies that $\vv \tau \subset \vv$. The case is

6590: similar to case C.2.2.

6591: 

6592: 

6593: \vskip .2in

6594: \noindent

6595: {\bf {Case C.2.3.1}} $-$ 

6596: $w \tau^{-1} \in \vv \alpha^{-1}, 

6597: \ w \tau^{-1} \not \in \vv \alpha$.

6598: 

6599: This implies that 

6600: $\vv \alpha^{-1} \tau^{-1} \subset \vv \alpha^{-1}$.

6601: 

6602: Here $w \tau^{-1} \alpha$ is in $\vv$,  \

6603: $w \tau^{-1} \alpha \tau$ is in $\vv$ so

6604: $w \alpha \beta \alpha^{m-1} = w \beta \alpha^{m-1}$

6605: is in $\vv$.

6606: Also 

6607: %$w \beta = w \tau \alpha^{-1} \tau^{-1}$,

6608: 

6609: $$w \tau \alpha^{-1} \ \in  \ \vv \alpha^{-1} \ \Rightarrow \

6610: w \beta = w \tau \alpha^{-1} \tau^{-1} \in \vv \alpha^{-1}

6611: \ \Rightarrow \

6612: w \beta \alpha^{m-1} \in \vv \alpha^{m-2}$$ 

6613: 

6614: \noindent

6615: which

6616: must be equal to $\vv$.

6617: 

6618: On the other hand $s \alpha \tau = s \beta \alpha^m$.

6619: Here $s \alpha \in \vv \alpha$ and bridges to $w$ in $\lat$,

6620: so $s \alpha \tau$ bridges to $w \tau$ in $\lat$

6621: and $s \alpha \tau \in \vv$.

6622: Also 

6623: 

6624: 

6625: $$s \beta \ = \ s \alpha^{-1} \tau^{-1}  \ \in \ \vv \alpha^{-1}

6626: \ \ \ {\rm and} \ \ \ 

6627: s \beta \alpha^m \ \in \ \vv \alpha^{m-1}.$$

6628: 

6629: 

6630: \noindent

6631: It follows that $\vv \alpha^{m-1} = \vv \alpha^{m-2}$,

6632: contradiction to $ \vv \not = \vv \alpha$.

6633: 

6634: 

6635: \vskip .2in

6636: \noindent

6637: {\bf {Case C.2.3.2}} $-$

6638: $w \tau^{-1} \not \in \vv \alpha^{-1},

6639: \ w \tau^{-1} \not \in \vv \alpha$,

6640: see fig. \ref{18}, a.

6641: 

6642: Use $s \alpha \tau = s \beta \alpha^m 

6643: = s \alpha \beta \alpha^{m-1}$.

6644: In this case the point 

6645: $s \alpha$ brides to $w$ in $\lat$

6646: and $s \alpha \tau \in \vv$.

6647: Also 

6648: $s \alpha^{-1}$ bridges to $w$ in $\lat$

6649: and $s \beta = s \alpha^{-1} \tau^{-1}$ bridges 

6650: $w \tau^{-1}$ in $\lat$ so

6651: 

6652: $$s \beta \ \ \ {\rm is \ in} \ \

6653: \re = T_w(w \tau^{-1}) \ \not = \ 

6654: \vv \alpha, \vv \alpha^{-1} \ \ \Rightarrow \ \

6655: s \beta \alpha^m \in \re \alpha^m = \vv.$$

6656: 

6657: \noindent

6658: So in particular $\re \not = \re \alpha$.

6659: 

6660: On the other hand $s \alpha \tau \alpha^{-1} \in \vv \alpha^{-1}$

6661: and bridges to $w$ in $\lat$ so $s \alpha \beta = 

6662: s \alpha \tau \alpha^{-1} \tau^{-1}$ bridges

6663: to $w \tau^{-1}$ in $\lat$ and is in $\re$.

6664: Then $s \alpha \beta \alpha^{m-1} \in \re \alpha^{m-1}

6665: = \vv \alpha^{-1}$. This would imply $\vv = \vv \alpha^{-1}$,

6666: contradiction. 

6667: 

6668: The final case in C.2.3 is:

6669: 

6670: 

6671: \vskip .2in

6672: \noindent

6673: {\bf {Case C.2.3.3}} $-$ $w \tau^{-1} \in \vv \alpha$.

6674: 

6675: Let $[s \alpha, r]$

6676: be the bridge from

6677: $s \alpha$ to $\lat$

6678: with $r$ in $\lat$. Then $r > w$ in $\lat$.

6679: Here we have to subdivide.

6680: 

6681: \vskip .2in

6682: \noindent

6683: {\bf {Situation I}} $-$ $r$ is in $(w, w \tau^{-1})$.

6684: 

6685: Then $s \alpha \tau$ bridges to $\lat$ in

6686: $r \tau \in (w, w \tau)$ and $s \alpha \tau \in \vv$.

6687: Hence 

6688: 

6689: $$s \alpha \tau \alpha^{-1} \not \in \vv \ \ \Rightarrow \ \ 

6690: s \alpha \tau \alpha^{-1} \tau^{-1} \ = \ s \alpha \beta \

6691: \in 

6692: \vv \alpha \ \ \Rightarrow \ \ 

6693: s \alpha \beta \alpha^{m-1} \ \in \ \vv \alpha^m

6694: \ \ \Rightarrow \ \ 

6695: \vv = \vv \alpha^m.$$

6696: 

6697: On the other hand $s \beta \alpha^m = s \alpha^{-1} \tau^{-1} 

6698: \alpha^m$. Here $s \alpha^{-1} \tau^{-1}$ is in $\vv \alpha$

6699: so $s \beta \alpha^m$ is in $\vv \alpha^{m+1}$, 

6700: implying $\vv \alpha^m = \vv \alpha^{m+1}$ again a contradiction.

6701: 

6702: \vskip .2in

6703: \noindent

6704: {\bf {Situation II}} $-$ $r = w \tau^{-1}$.

6705: 

6706: Here $s \alpha \tau$ bridges to $\lat$ in $w$ hence

6707: $s \alpha \tau \not \in \vv \alpha$ and $s \alpha \tau

6708: \not \in \vv$.

6709: So $s \alpha \tau$ is in $\re$, another component

6710: of $T - \{ w \}$.

6711: Also 

6712: 

6713: $$s \alpha^{-1} \ \not \in \vv \ \ \Rightarrow \ \ 

6714: s \beta  \ = \ s \alpha^{-1} \tau^{-1} \ \in \  \vv \alpha

6715: \ \ \Rightarrow \ \ 

6716: s \beta \alpha^m \ \in \ \vv \alpha^{m+1} \ \ \Rightarrow \ \ 

6717: \re = \vv \alpha^{m+1}.$$

6718: 

6719: 

6720: 

6721: On the other hand $s \alpha \beta \alpha^{m-1} \in \vv \alpha^{m+1}$,

6722: so $s \alpha \beta \in \vv \alpha^2$.

6723: Now $\vv \alpha^2 \not = \vv \alpha$ so $\vv \alpha^2 \tau$

6724: is contained in $\vv$. Hence 

6725: $s \alpha \tau \alpha^{-1} = s \alpha \beta \tau$ is in

6726: $\vv$.

6727: This would imply $s \alpha \tau$ is in $\vv \alpha$,

6728: contradiction to the first conclusion in

6729: this case.

6730: 

6731: %\blankfig{18}{1.}{

6732: %a. Case C.2.3, b. Case C.2.3.3, Situation III.}

6733: 

6734: 

6735: 

6736: \begin{figure}

6737: \centeredepsfbox{lam18.eps}

6738: \caption{

6739: a. Case C.2.3, b. Case C.2.3.3, Situation III.}

6740: \label{18}

6741: \end{figure}

6742: 

6743: 

6744: 

6745: 

6746: \vskip .2in

6747: \noindent

6748: {\bf {Situation III}} $-$ $w  \tau^{-1} < r$ in $\lat$. 

6749: 

6750: This is a little more tricky.

6751: Here $s \alpha \tau \in \vv \alpha$, see fig. \ref{18}, b.

6752: Also 

6753: 

6754: $$w \beta^{-1}  \ = \  w \tau \alpha \tau^{-1} \ \in \ \vv \alpha

6755: \ \subset \we.$$

6756: 

6757: \noindent

6758: Now use 

6759: $w \beta^{-1} \alpha^{-1}

6760: = w \tau^{-1} \alpha^{-1} \tau$.

6761: Here

6762: 

6763: $$w \tau^{-1} \in [w, s \alpha] \ \ \Rightarrow \ \ 

6764: w \tau^{-1} \alpha^{-1} \in [w,s] \ \ \Rightarrow \ \ 

6765: w \tau^{-1} \alpha^{-1} \tau \in

6766: [s, w \tau] \subset \lat \subset \we.$$

6767: 

6768: \noindent

6769: So $w \beta^{-1} \alpha^{-1}, w \beta^{-1}$ are both

6770: in $\we$, with the usual implications

6771: that $\we \gamma = \we$ and $\gamma$

6772: leaves $\lat$ invariant.

6773: As $w \beta^{-1} \alpha^{-1}$ is in $\lat$ then 

6774: $w \beta^{-1}$ is in $\lat$ as well.

6775: Also

6776: 

6777: %In addition this implies that $w \tau^{-1} \alpha \tau$

6778: %is in $\lat$ and $w \beta^{-1}$ also is.

6779: 

6780: $$w \beta^{-1} \alpha^{-1}  \ = \ 

6781: w \tau^{-1} \alpha^{-1} \tau < w \tau \ \in \

6782: \lat \ \ \ \Rightarrow \ \ \ 

6783: w \tau \alpha \ \in \ \lat.$$

6784: 

6785: The proof is now analogous to previous arguments.

6786: If 

6787: 

6788: 

6789: $$w \ \prec \ w \tau^{-1} \ \prec \ w \tau \alpha

6790: \ \ \ \Rightarrow \ \ \ 

6791: w \tau^{-1} \  \prec \ w \tau^{-2} \ \prec \ w \tau \alpha \tau^{-1}

6792: \ = \ w \beta^{-1}.$$

6793: 

6794: %$w \tau \alpha \leq w \tau^{-1}$ in $\lat$, then

6795: %$w \tau \alpha \tau^{-1}  \leq w \tau^{-2}$ in $\lat$,

6796: %so $w \beta^{-1} \leq w \tau^{-2}$.

6797: \noindent

6798: But 

6799: 

6800: $$w \beta^{-1} \alpha^{-1} \gamma \ = \ w \beta^{-1}

6801: \ \ \ {\rm and} \ \ \ 

6802: w \beta^{-1} \alpha^{-1} \ \in \ (s, w \tau)$$

6803: 

6804: 

6805: \noindent 

6806: implies

6807: as before that

6808: $p > 3q$, contradiction.

6809: 

6810: On the other hand if $w \prec w \tau \alpha \prec w \tau^{-1}$,

6811: then $w \tau^{-1} \ \prec \ w \tau \alpha \tau^{-1} = w \beta^{-1}

6812: \ \prec \ w \tau^{-2}$

6813: all in $\lat$.

6814: Here $s \alpha \tau \in \vv \alpha$. Now

6815: $s \beta \alpha^m \ = \ s \alpha^{-1} \tau^{-1} \alpha^m$.

6816: Also

6817: 

6818: $$s \alpha^{-1}  \ \not \in \ \vv \ \ \Rightarrow \ \ 

6819: s \alpha^{-1} \tau^{-1} \ \in \ \vv \alpha \ \ \Rightarrow \ \ 

6820: s \beta \alpha^m \ \in \ \vv \alpha^{m+1} \ \ \ \Rightarrow \ \ \ 

6821: \vv \alpha = \vv \alpha^{m+1} \ \ {\rm or}  \

6822: \vv \ = \ \vv \alpha^m.$$

6823: 

6824: 

6825: Now use $w \tau^2 \gamma \beta \alpha^m = w \tau \alpha \tau$.

6826: Here 

6827: 

6828: $$w \tau \ \prec \ w \tau \alpha \tau 

6829: \ \prec  \ w \ \ \ {\rm in} \ \ \lat \ \ \ \Rightarrow \ \ \ 

6830: w \tau \alpha \tau \ \in \ \vv, \ \ 

6831: w \tau^2 \gamma \beta \ \in \ \vv \alpha^{-m} \ = \ \vv.$$

6832: 

6833: \noindent

6834: So $w \tau^2 \gamma \prec w \beta^{-1} \prec w \tau^{-1}

6835: \prec w$,

6836: implying again $p > 3q$, contradiction.

6837: 

6838: This finishes the analysis of case C.2.3, $w \tau \in [s,w]$

6839: and so proves that the case $\vv \alpha \not = \vv$ cannot

6840: occur.

6841: From now on in case C assume:

6842: 

6843: 

6844: \vskip .2in

6845: \noindent

6846: {\bf {Case C.3}} $-$ 

6847: $\we \tau = \we$ and $\vv \alpha = \vv$.

6848: 

6849: 

6850: 

6851: Since there is no other fixed point

6852: of $\alpha$ in $(s,w)$,

6853: this immediately implies there is a local axis $\lal$ of

6854: $\alpha$ contained in $\vv$ 

6855: with $w$ as an ideal point of $\lal$.

6856: We stress that at this point we do not yet have an

6857: axis for $\tau$, because there may be other fixed points

6858: of $\tau$ in $(s,w)$.

6859: 

6860: \begin{lemma}{}{}

6861: $s \alpha, s \alpha^{-1} \in \we$,

6862: so $s \alpha, s \alpha^{-1}$ are not in $[s,w)$.

6863: \label{all}

6864: \end{lemma}

6865: 

6866: \begin{proof}{}

6867: Suppose first that $s \alpha$ is not in $\we$.

6868: Then 

6869: 

6870: 

6871: $$s \alpha^{-1} \ \in \ (s,w) \ \subset \ \we \ \ \ \Rightarrow \ \ \

6872: s \alpha^{-1} \tau^{-1} \ \in \ \we \tau \ = \ \we.$$

6873: 

6874: \noindent

6875: So $s \beta \in \we$ and bridges to $[s,w]$ in

6876: a point $r$ which is in $(s,w]$.

6877: Then $s \beta \alpha^m$ bridges to $[s,w]$ in

6878: $r \alpha^m$ and $s \beta \alpha^m$ is in $\we$.

6879: Therefore $s \alpha \tau$ is in $\we$ and

6880: $s \alpha$ is in $\we \tau^{-1} = \we$, contradiction.

6881: 

6882: On the other hand suppose that $s \alpha^{-1} \not

6883: \in \we$.

6884: Then $s \alpha \in (s,w]$. Also $s \beta =

6885: s \alpha^{-1} \tau^{-1} \not \in \we$, so bridges to $[s,w]$ in

6886: $s$. Then $s \beta \alpha^m$ bridges to $[s \alpha^m, x]$

6887: in $s \alpha^m$. Since $s \alpha^m \not \in \we$ this

6888: implies $s \beta \alpha^m \not \in \we$, 

6889: therefore $s \alpha \tau \not \in \we$. But then

6890: $s \alpha$ is not in $\we$, contradiction.

6891: This finishes the proof.

6892: \end{proof}

6893: 

6894: 

6895: We conclude that $s \alpha, s \alpha^{-1}$ are

6896: in $\we \cap \vv$. Let $s \alpha$ bridge to $r$

6897: in $[s,w]$, hence $r \in (s,w)$ and $s \alpha^{-1}$

6898: bridges to $[s,w]$ in a point $t$ also in  $(s,w)$.

6899: 

6900: Let $z$ be the fixed point of $\tau$ in $[s,w]$ which is

6901: closest to $w$. \ Then $z$ may be equal to $s$, but is not $w$.

6902: Let $\uu = T_z(w)$. 

6903: One important goal is to prove that $\uu \tau = \uu$.

6904: 

6905: \begin{lemma}{}{}

6906: Let $\uu = T_z(w)$. Then $\uu \tau = \uu$.

6907: If $z \not = s$  then

6908: $z \gamma, w \gamma \not \in \we$, and

6909: $z \alpha, \ z \alpha^{-1} \not \in (z,w)$.

6910: \label{prel}

6911: \end{lemma}

6912: 

6913: \begin{proof}{}

6914: If $z = s$ then $\uu = \we$ and the result follows

6915: from Case C.1.

6916: For the rest of the proof of the lemma assume

6917: that $s \not = z$.

6918: 

6919: We first analyse the possibility that $z \gamma \in \we$.

6920: As $\kappa$ fixes $s$ then $z \gamma^{-1} \in \we$ also.

6921: If $z \gamma = z$, then $z \kappa = z$, contradiction.

6922: 

6923: Suppose that $z \gamma$ or $z \gamma^{-1}$ is in

6924: $[s,z)$.

6925: Then as $s \gamma = s$, it follows that $z$ is in

6926: a local axis for $\gamma$ and $z \gamma^q \not = z$,

6927: contradiction to $z$ fixed by $\tau$.

6928: Hence $z \gamma, z \gamma^{-1} \not \in [s,z]$.

6929: 

6930: Let $[z \gamma, r]$ be the bridge from

6931: $z \gamma$ to $[s,z]$. Notice that

6932: $r$ is in $(s,z)$, because $z \gamma, z \gamma^{-1}$

6933: are not in $[s,w]$.

6934: Then

6935: 

6936: $$r \ \in \ [s, z] \cap [s, z \gamma] \ \ \Rightarrow \ \ 

6937: r \gamma^{-1} \in \ [s, z].$$

6938: 

6939: \noindent

6940: If $r \gamma = r$, then $r \tau^p = r \gamma^{-q} = r$.

6941: But $([s,z]) \tau = [s,z]$,

6942: % $-$ $\tau$ acts on

6943: %an interval, 

6944: so this would imply $r \tau = r$.

6945: Together these imply $r \kappa = r$, contradiction

6946: to $s$ the fixed point of $\kappa$ in $[s,w]$

6947: which is closest to $w$.

6948: 

6949: We conclude that $r \gamma \not = r$.

6950: But as $s \gamma = s$, this implies that

6951: $r$ is in a local axis $\lag$ of $\gamma$.

6952: Compute $r^{nq}, n \in {\bf Z}$.

6953: Assume without loss of generality 

6954: that $r^{nq}$ moves away from $s$ as

6955: $n \rightarrow +\infty$.

6956: Then 

6957: 

6958: $$r \gamma^{nq} \ =  \ r \tau^{-np} \in 

6959: [s,w], \ \forall n \ \ \ {\rm and} \ \ \

6960: r \gamma^{nq} \rightarrow \ c  \in 

6961: (s,z] \ {\rm as} \ n \rightarrow +\infty.$$

6962: 

6963: 

6964: \noindent

6965: Then $c \gamma = c$ and also $c \tau = c$, contradiction.

6966: 

6967: This contradiction shows that $z \gamma \in \we$

6968: is impossible.

6969: Notice that if $z \gamma$ is not in $\we$, then

6970: $z \gamma$ separates $\we \gamma$ from $s$ and

6971: hence from $\we$. It follows that

6972: $\we \gamma \cap \we

6973: = \emptyset$, so $w \gamma \not \in \we$.

6974: This proves one assertion of lemma \ref{prel}.

6975: 

6976: We now consider where $z \alpha$ and $z \alpha^{-1}$ are.

6977: Notice they are both in $\vv$.

6978: Remember that for the rest of the proof $s \not= z$.

6979: 

6980: \vskip .1in

6981: \noindent

6982: {\bf {Situation I}} $-$

6983: Suppose first that $z \alpha \in (z,w)$.

6984: 

6985: Use $\alpha \tau = \tau \gamma \beta \alpha^m$, applied

6986: to $z$. Here $z \alpha$ is in

6987: $\uu$ so $z \alpha \tau$ is in $\uu \tau$.

6988: Suppose first that $\uu \tau \not = \uu \alpha^{-1}$.

6989: Then $z \alpha \tau$ bridges to $\lal$ in a point

6990: in $[z,w]$ and

6991: hence $a = z \alpha \tau \alpha^{-m}$ bridges to

6992: $\lal$ in a point in

6993: $[z \alpha^{-m},w]$ and $a$ is in $\uu$.

6994: Here

6995: 

6996: $$z \alpha \tau \alpha^{-m} \ = \ 

6997: z \gamma \beta \ = \ z \gamma \alpha^{-1} \tau^{-1}

6998: \ \ \ \Rightarrow \ \ \ 

6999: z \gamma \alpha^{-1} \ \in \ \uu \tau \not = \uu.$$

7000: 

7001: \noindent

7002: Again $z \gamma \alpha^{-1}$

7003: bridges to $\lal$ in a point in $[z,w]$ and it follows that

7004: $z \gamma$ is in $\uu$, hence 

7005: $z \gamma \in \we$ contradicting $\we \gamma \cap \we = \emptyset$.

7006: 

7007: The remaining possibility is $\uu \tau = T_z( z \alpha^{-1})$,

7008: so in particular $\uu \tau \not = \uu$, see fig. \ref{19}, a.

7009: Consider $w \tau^{-1} \alpha^{-1} \tau$. The point

7010: $w \tau^{-1}$ is not in $\uu$, hence it bridges

7011: to $\lal$ in a point not in $(z,w]$. Therefore

7012: $w \tau \alpha^{-1}$ bridges to $\lal$ in

7013: a point not in $(z \alpha^{-1}, w]$, so $w \tau

7014: \alpha^{-1}$ is in $T_z(z \alpha^{-1}) = \uu \tau$.

7015: Hence 

7016: 

7017: $$w \tau^{-1} \alpha^{-1} \tau  \ = \ w \beta^{-1} \alpha^{-1}

7018: \ \ \

7019: {\rm is \ in} \ \ 

7020: \uu \tau^2

7021: \not = \ \uu \tau, \ T_z(s).$$

7022: 

7023: \noindent

7024: Notice that

7025: 

7026: $$(T_z(s)) \tau \ = \ T_z(s), \ \ {\rm since } \ \ 

7027: s \tau  \ = \ s, \ \ \ {\rm so} \ \ T_z(s) \ \not = \uu \tau^2.$$

7028: 

7029: \noindent

7030:  In particular

7031: $w \beta^{-1} \alpha^{-1}$ is in $\we$ and also bridges

7032: to $\lal$ in a point which is in $[z,w]$. Then

7033: $w \beta^{-1}$ bridges to $\lal$ in a point

7034: which is in $[z \alpha, w]$ so in particular

7035: $w \beta^{-1}$ is in $\uu \subset \we$.

7036: But then $w \beta^{-1} \alpha^{-1}$ and $w \alpha^{-1}$ 

7037: are both in $\uu$, contradicting 

7038: $\we \gamma \cap \we = \emptyset$.

7039: 

7040: %to $w \alpha^{-1}$ in $\lal$ and $s \alpha^{-1}$ is

7041: %in $T_w(w \alpha^{-1} = \uu \tau$. Then

7042: %$s \beta = s \alpha^{-1} \tau^{-1}$ is in $\uu$.

7043: %Use $\alpha \tau \alpha^{-m} \tau^{-1} = \tau \gamma$.

7044: %Here $w \alpha$ is not in $T_w(s)$ invariant under $\tau$,

7045: %hence $w \alpha \tau$ is not in $T_w(s)$.

7046: %Therefore it bridges to $[s,w]$ in a point which

7047: %is not in $[s,w)$. So $w \alpha \tau \alpha^{-m}$

7048: %bridges to $ahw$

7049: %First if $w \alpha^{-1}$ is not in $\we$, then

7050: %$s \alpha \in [w,w]$ and

7051: %$s \alpha \tau \in \we$.

7052: %On the other hand $s \beta = s \alpha^{-1} \tau^{-1}$

7053: %and $s \alpha^{-1}$ is not in $\we$ $-$ the $\alpha^{-1}$ is

7054: %expanding $w$ away from 

7055: 

7056: 

7057: This finishes the analysis of possibility $z \alpha \in (z,w)$.

7058: 

7059: %\blankfig{19}{1.}{

7060: %a. Situation I, Situation III.}

7061: 

7062: 

7063: 

7064: \begin{figure}

7065: \centeredepsfbox{lam19.eps}

7066: \caption{

7067: a. Situation I, Situation III.}

7068: \label{19}

7069: \end{figure}

7070: 

7071: 

7072: % b. Situation II when

7073: %$w \alpha \in T_w(w \tau^{-1})$.}

7074: 

7075: \vskip .1in

7076: \noindent

7077: {\bf {Situation II}} $-$  Suppose $z \alpha^{-1} \in (z,w)$.

7078: 

7079: 

7080: Consider first the case when $z \alpha \in \uu \tau^{-1}$,

7081: that is $T_z( z \alpha) = T_z (w \tau^{-1})$. This is very

7082: similar to Situation I, second part.

7083: Since $z \alpha$ is not in $\uu$, this in particular

7084: implies $\uu \tau \not = \uu$.

7085: %Consider $w \beta^{-1} = w \tau \alpha \tau^{-1}$.

7086: Here $w \tau \not \in \uu$, hence it bridges to $\lal$ in

7087: a point which is not in $(z,w]$. It follows that

7088: $w \tau \alpha$ bridges to $\lal$ in a point

7089: which is not in $(z \alpha, w]$.

7090: %, see fig. \ref{21}, b.

7091: This implies  that $w \tau \alpha$ is in

7092: $T_z(z \alpha) = T_z(w \tau^{-1})$.

7093: Hence 

7094: 

7095: 

7096: $$w \beta^{-1} \ = \ w \tau \alpha \tau^{-1} 

7097: \ \ \ {\rm is \ in} \ \

7098: T_z(w \tau^{-2}) \not = \ 

7099: T_z(s), \ T_z(w \tau^{-1}).$$

7100: 

7101: 

7102: \noindent

7103: The first

7104: fact means that $w \beta^{-1}$ is in $\we$.

7105: The second fact means that $w \beta^{-1}$ is not

7106: in $T_z(z \alpha)$,  hence $w \beta^{-1}$

7107: bridges to $\lal$ in

7108: a point contained in $[z,w]$. Hence $w \beta^{-1}

7109: \alpha^{-1}$ bridges to $\lal$ in a point contained

7110: in $[z \alpha^{-1}, w]$ and is in $\we$. 

7111: As $w \beta^{-1} \alpha^{-1} \gamma = w \beta^{-1}$,

7112: this would imply $\we \gamma = \we$, again 

7113: contradiction.

7114: Hence this cannot occur.

7115: 

7116: Now we know $z \alpha$ is not in $T_z(w \tau^{-1})$.

7117: The point $z \beta =

7118: z \alpha^{-1} \tau^{-1}$ is in $T_z(w \tau^{-1})$,

7119: hence it bridges to $\lal$ in a point

7120: contained in $[z,w]$. It follows that $z \beta \alpha^m$

7121: bridges to $\lal$ in a point contained in $[z \alpha^m, w]$.

7122: But 

7123: 

7124: $$z \alpha^m  \in \ \uu \ \ \Rightarrow \ \  

7125: z \beta \alpha^m \ \in \ \uu \ \ \Rightarrow \ \ 

7126: z \gamma^{-1} \alpha \tau \ \in \ \uu \ \ {\rm or} \ \ 

7127: z \gamma^{-1} \alpha \ \in  \ T_z(w \tau^{-1})$$

7128: 

7129: \noindent

7130: and bridges to $\lal$ in a point in $[z,w]$.

7131: It follows that $z \gamma^{-1}$ bridges to $\lal$

7132: in a point in $[z \alpha^{-1}, w]$, hence

7133: $z \gamma^{-1} \in \uu \subset \we$, impossible.

7134: 

7135: We conclude that situation II cannot occur.

7136: This proves the last 2 assertions of the lemma \ref{prel}.

7137: It also implies that the following situation must occur:

7138: 

7139: %\blankfig{22}{1.}{a. Situation III, b. }

7140: 

7141: \vskip .2in

7142: \noindent

7143: {\bf {Situation III}} $-$ $z \alpha \not \in (z,w),

7144: z \alpha^{-1} \not \in (z,w)$, see fig. \ref{19}, b.

7145: 

7146: What is left to prove of lemma \ref{prel} is that

7147: $\uu \tau = \uu$. So suppose that $\uu \tau \not = \uu$.

7148: %, see

7149: %fig. \ref{22}, a.

7150: 

7151: Here $z \alpha^{-1}$ bridges to $[z,w]$ in a point

7152: $r$ which is in $(z,w)$. Also $z \alpha$ bridges to $t$

7153: in $[z,w]$ with $t$ also in $(z,w)$.

7154: 

7155: The point $w \gamma$ is not in $\we$, so it is in $T_z(s)$

7156: and bridges to $[w \tau^{-1}, z]$ in $z$.

7157: Hence $w \gamma \beta$ bridges to $[w \tau^{-1}, z \beta]$

7158: in $z \beta$. But $z \beta = z \alpha^{-1} \tau^{-1}$

7159: bridges to $[z, w \tau^{-1}]$ in $r \tau^{-1}$.

7160: Then $w \gamma \beta$ bridges to $[z,w]$ in $z$

7161: (this uses $\uu \tau \not = \uu$!).

7162: Then 

7163: 

7164: $$w \gamma \beta \alpha^m \ \ \ {\rm bridges \ to} \ \ 

7165: [z,w] \ \ {\rm in  \ a \ point \ in } \ \ 

7166: (z,w) \ \ \ {\rm so} \ \ \ 

7167: w \gamma \beta \alpha^m \ \in \uu.$$

7168: 

7169: 

7170: On the other hand $w \tau^{-1}$ bridges to $[z,w]$ 

7171: in $z$ so  $ w \tau^{-1} \alpha$ bridges to $[z,w]$

7172: in a point in $(z,w)$

7173: and $w \tau^{-1} \alpha$ is in $\uu$.

7174: Then $w \tau^{-1} \alpha \tau$ is in $\uu \tau$.

7175: Of course this implies $\uu \tau = \uu$, contrary to assumption.

7176: 

7177: So in any case we conclude that $\uu \tau = \uu$.

7178: This finishes the proof of lemma \ref{prel}.

7179: \end{proof}

7180: 

7181: 

7182: This lemma is very useful.

7183: Since there is no fixed point of $\tau$ in $(z,w)$ and

7184: $T_z(w) \tau = T_z(w)$ it follows that there is

7185: a local axis $\lat$ of $\tau$ contained in $\uu = T_z(w)$

7186: with an ideal point $z$.

7187: 

7188: 

7189: \begin{lemma}{}{}

7190: $w$ is not in $\lat$.

7191: \end{lemma}

7192: 

7193: \begin{proof}{}

7194: Suppose not, that is, $w \in \lat$.

7195: Here we will use lemma \ref{chax}:

7196: Suppose that $\lat$ is a local axis for $\tau$ and

7197: $w$ is a point in

7198: $\lat$ with $w \alpha = w$. Then

7199: at least one of the components of $T - \{ w \}$

7200: containing

7201: $w \tau, w \tau^{-1}$ is not invariant under $\alpha$.

7202: 

7203: 

7204: %\blankfig{23}{1.}{a. Situation $w \tau^{-1} \in (w,z)$,

7205: %b. Situation $w \tau \in (w, z)$.}

7206: 

7207: \vskip .2in

7208: \noindent

7209: {\bf {Situation I}} $-$

7210:  $w \tau^{-1} \in [z,w)$.

7211: 

7212: Here $\vv = T_w(z) = T_w(w \tau^{-1})$ is invariant under

7213: $\alpha$. By lemma \ref{chax}, the set

7214: ${\cal R} = T_w(w \tau)$ is not invariant

7215: under $\alpha$. Notice that $\ro \alpha$ is not equal

7216: to $\vv$ either.

7217: 

7218: Use $w \alpha \tau = w \tau = w \tau \alpha \beta \alpha^{m-1}$.

7219: Here 

7220: 

7221: $$w \tau \ \in \ \ro \ \ \Rightarrow \ \ 

7222: w \tau \alpha  \ \in \ \ro \alpha \ \not = \ \vv 

7223: \ \ \Rightarrow \ \ 

7224: w \tau \alpha \tau \ \in \ \ro \tau \ \subset \ \ro

7225: \ \ \Rightarrow \ \ 

7226: c \ = \ w \tau \alpha \tau \alpha^{-1} \ \in \ \ro \alpha^{-1}

7227: \ \not = \ \ro.$$

7228: 

7229: 

7230: \noindent

7231: So $c$ bridges to $w$ in $\lat$ and

7232: then $w \tau \alpha \tau \alpha^{-1} \tau^{-1} = w \tau \alpha

7233: \beta$ bridges to $w \tau^{-1}$ in $\lat$ and is then

7234: in $\vv$. Finally $w \tau \alpha \beta \alpha^{m-1}$

7235: is in $\vv \alpha^{m-1} = \vv$. This is not

7236: $\ro$, contradiction.

7237: 

7238: \vskip .2in

7239: \noindent

7240: {\bf {Situation II}} $-$ $w \tau \in (z,w)$.

7241: 

7242: 

7243: Here $\vv = T_w(w \tau) = T_w(z)$ is invariant under $\alpha$.

7244: Let $\ro = T_w (w \tau^{-1})$, which is not

7245: invariant under $\alpha$.

7246: Use $w \tau^{-1} \alpha \tau = w \alpha \beta \alpha^{m-1}$.

7247: Then $w \tau^{-1}$ is in $\ro$, so $w \tau^{-1} \alpha$

7248: is not in $\ro$ or $\vv$ and bridges to $w$ in $\lat$.

7249: Then $w \tau^{-1} \alpha \tau$ bridges to $w \tau$ in

7250: $\lat$ and is in $\vv$. It follows that

7251: 

7252: $$w \tau^{-1} \alpha \tau \alpha^{1-m} 

7253: \ = \  w \alpha \beta \ = \ 

7254: w \beta = w \tau \alpha^{-1} \tau^{-1} \ \ \ {\rm is \ in}

7255: \ \vv.$$

7256: 

7257: \noindent

7258: Hence $w \tau \alpha^{-1}$ is in $\vv \tau$.

7259: This implies 

7260: 

7261: 

7262: $$w \tau \alpha^{-1} \prec w \tau \prec w

7263: \  \ \ \ \Rightarrow \ \  \ \ 

7264: w \tau \prec w \tau \alpha \prec w \ \ \ \ \Rightarrow \ \  \ \ 

7265: w \prec w \tau \alpha \tau^{-1} = 

7266: w \beta^{-1} \prec w \tau^{-1}.$$

7267: 

7268: \noindent

7269: In particular $w \beta^{-1}$ is in $\ro$ and $w \beta^{-1} \alpha^{-1}$

7270: is in $\ro \alpha^{-1}$ which is not equal to $\vv$.

7271: Also $w \beta^{-1} \alpha^{-1} = w \tau^{-1} \alpha^{-1}

7272: \tau$. Here $w \tau^{-1} \alpha^{-1}$ is in $\ro \alpha^{-1}$ 

7273: and bridges to $w$  in $\lat$ and so $w \tau^{-1} \alpha^{-1} \tau$

7274: bridges to $w \tau$ in $\lat$ and so is in $\vv$.

7275: As $\vv$ is not equal to $\ro \alpha^{-1}$, this

7276: is a contradiction.

7277: 

7278: We conclude that situation II cannot happen either.

7279: This finishes the proof of the lemma.

7280: \end{proof}

7281: 

7282: Now we know that $w$ is not in $\lat$.

7283: 

7284: 

7285: \begin{lemma}{}{}

7286: $z$ is not in $\lal$.

7287: \label{noal}

7288: \end{lemma}

7289: 

7290: \begin{proof}{}

7291: Suppose not, that is, $z \in \lal$.

7292: This implies that either 

7293: $z \alpha$ or $z \alpha^{-1}$ is in $(z,w)$.

7294: Then lemma \ref{prel} implies that

7295: $s   = z$.

7296: %In this proof we consider $s$ instead of $z$.

7297: 

7298: 

7299: 

7300: Suppose first that $z \alpha \in (z,w]$.

7301: So $z \alpha^{-1} \not \in T_z(w) = \uu$.

7302: Use $z \alpha \tau = z \beta \alpha^m$

7303: As $z \alpha \in \uu$, then $z \alpha \tau$ is

7304: in $\uu$ also.

7305: Then

7306: %On the other hand $z \beta = z \alpha^{-1} \tau^{-1}$.

7307: 

7308: $$z \alpha^{-1} \ \not \in \ \uu

7309: \ \ \Rightarrow \ \  z \alpha^{-1} 

7310: \tau^{-1} \ \not \in \ \uu \ \ \Rightarrow \ \ 

7311: z \beta \ \ \ {\rm bridges \ to} \ \  [z,w] 

7312: \ \ {\rm in} \ z$$

7313: 

7314: 

7315: \noindent

7316:  and

7317: $z \beta \alpha^m$ bridges to $[z \alpha^m, w] \supset [z,w]$

7318: in $z \alpha^m$. It follows that $z \beta \alpha^m$ is

7319: not in $\uu$, contradiction.

7320: 

7321: Suppose now that $z \alpha^{-1}$ is in $[z,w]$. Then

7322: $z \alpha^{-1} \tau^{-1} = z \beta$ is in $\uu$ and

7323: bridges to $[z,w]$ in a point $t$ which is not

7324: $z$. Then $z \beta \alpha^m$ bridges to $[z,w]$ in

7325: $t \alpha^m$ and $z \beta \alpha^m$ is in $\uu$.

7326: On the other hand $z \alpha$ is not in $\uu$ and

7327: so $z \alpha \tau$ is not in $\uu$ either.

7328: This is a contradiction.

7329: 

7330: This finishes the proof of the lemma.

7331: \end{proof}

7332: 

7333: 

7334: 

7335: \vskip .1in

7336: \noindent

7337: {\bf {Summary in Case C.3}} $-$ So far we 

7338: have proved: suppose

7339:  that $w \alpha = w$, $s \kappa = s$,

7340: no fixed points of $\kappa$ or $\alpha$  in $(s,w)$.

7341: Let $z \in [s,w)$, the closest to $w$ with $z \tau = z$.

7342: Then 

7343: 

7344: $$T_z(w) \tau \ = \ T_z(w), \ \ \ \ \ T_w(z) \alpha \ = \ T_w.$$

7345: 

7346: \noindent

7347: If $\lat, \lal$ are the corresponding local axes 

7348: of $\tau$ and $\alpha$ then

7349: $z \not \in \lal$, $w \not \in \lat$.

7350: 

7351: \vskip .2in

7352: \noindent

7353: {\bf {Case C.3.0}} $-$ Suppose that $\lal \cap \lat$ has

7354: at most one point.

7355: 

7356: This is very simple.

7357: Let $[c,d]$ be the bridge from $\lat$ to $\lal$,

7358: where $c = d$ if the intersection is one point.

7359: We do the proof for $c \not = d$, the other is very 

7360: similar.

7361: Use $z \tau^{-1} \alpha \tau = z \alpha \beta \alpha^{m-1}$.

7362: The right side is $z \alpha \tau$. Here $z \alpha$ bridges

7363: to $\lal$ in $d \alpha$, hence bridges to $\lat$ in

7364: $c$. So $z \alpha \tau$ bridges to $\lat$ in $c \tau$.

7365: 

7366: So $z \alpha \tau$ bridges to $\lal$ in $d$ so

7367: $z \alpha \tau \alpha^{-1}$ bridges to $\lal$ in $d \alpha^{-1}$

7368: and to $\lat$ in $c$. So $z \alpha \tau \alpha^{-1} \tau^{-1}

7369: = z \alpha \beta$

7370: bridges to $\lat$ in $c \tau^{-1}$ hence to 

7371: $\lal$ in $c$. Finally $z \alpha \beta \alpha^{m-1}$ bridges

7372: to $\lal$ in $d \alpha^{m-1}$ hence to $\lat$ in

7373: $c$. Since $c \not = c \tau$ this is a contradiction.

7374: 

7375: \vskip .2in

7376: \noindent

7377: {\bf {Case C.3.1}} $-$

7378: Now assume $\lal \cap \lat$ has more than one

7379: point. 

7380: We will use the analysis done in case B.

7381: 

7382: If $\uu \gamma$ is not equal $\uu$ then we use the

7383: proof of case B.1.3 $-$ which was done also for

7384: local axis of $\alpha$. This disallows this case.

7385: 

7386: The remaining case is that $\uu \gamma$ is equal to

7387: $\uu$. As explained in case B.1.4 this implies

7388: $\gamma$ leaves $\lat$ invariant.

7389: Here we consider the intersection $\bb = \lal \cap \lat$.

7390: First notice that $z$ is not in $\bb$. 

7391: If $z$ were a limit point of $\bb$ then $\bb$ would

7392: be $(z,r]$ (recall that $w$ is not in $\lat$).

7393: Then as $\alpha$ leaves invariant $\lal$ we would

7394: have $z \alpha = z$ also ruled out by non trivial

7395: action of the group on $T$.

7396: If $\lat$ is not properly embedded on the other

7397: side let $v$ be the other ideal point of $\lat$.

7398: Then 

7399: 

7400: $$v \kappa \ = \ v, \ \ \ (T_w(v)) \alpha \ = \ T_w(v),

7401: \ \ \ (T_v(w)) \tau \ = \ T_v(w).$$

7402: 

7403: \noindent

7404: Also $(w,v)$ has no fixed points of $\tau$.

7405: Suppose that $v$ is in $\lal$. Then $(w,v)$ also

7406: has no fixed points of $\alpha$. But then

7407: $v$ has the same properties as $z$ and this

7408: case is ruled out by lemma \ref{noal}.

7409: It follows that $v$ is not in $\lal$. 

7410: So if $\lat$ has another ideal point $v$, then

7411: $\bb$ is $[r,t]$ with $t$ an actual point

7412: in $\lat$.

7413: 

7414: Now we can apply the analysis of case B.1.4 which

7415: was also done for $\alpha$ with a local axis.

7416: The analysis rules out this situation.

7417: 

7418: This shows that case C.3.1 cannot happen either.

7419: 

7420: This finishes the proof of the main theorem.

7421: 

7422: 

7423: \section{Remarks}

7424: 

7425: 

7426: There are a lot of interesting questions still open.

7427: First we discuss some internal questions about the proofs

7428: in this article. The proof of the $\rrrr$-covered case

7429: uses $p > 3q$ for $\alpha$ orientation reversing.

7430: It would be useful to get a more general proof $-$

7431: for instance showing that $p$ must be equal to $4$

7432: or that $p$ has to be even. We obtained some preliminary

7433: results, but not conclusive.

7434: The same argument and condition $p > 3q$ are then used

7435: in various places of the article so it would be

7436: very good to discover a more general proof.

7437: 

7438: Also the best possible result for the manifolds $M_{p/q}$ 

7439: described in this article would be the following:

7440: If $p \geq q$, $p$ odd, $m \leq -4$ then the

7441: only possible essential laminations are those 

7442: coming from either stable or unstable lamination

7443: in the original manifold $M$ $-$ these remain

7444: essential whenever $|p - 2q| \geq 2$.

7445: One way to interpret such a goal is a rigidity

7446: result $-$ all laminations in this manifold have to

7447: be of this type. Notice that Brittenham's results

7448: for Seifert fibered spaces \cite{Br1} are of

7449: this form. Also Hatcher and Thurston's results

7450: for surgery on 2-bridge links \cite{Ha-Th} 

7451: are along these lines.

7452: 

7453: Now on for more general goals: 

7454: How far can the methods of this article be generalized?

7455: Can they be used whenever $M$ is a punctured torus

7456: bundle over $S^1$ with Anosov monodromy and

7457: degeneracy locus $(1,2)$?

7458: Probably a mixture of topological methods and group

7459: action methods needs to be used.

7460: How about surface bundles, where the surface has

7461: higher genus? What about other degeneracy locus

7462: as discovered by Gabai-Kazez \cite{Ga-Ka1}?

7463: 

7464: Since essential laminations do not exist in

7465: every closed hyperbolic $3$-manifold, one

7466: looks for useful generalizations. One possible 

7467: idea was introduced by Gabai in \cite{Ga5}:

7468: a lamination $\lambda$ in $M$, compact, orientable,

7469: irreducible is {\em loosesse} if $\lambda$ satisfies:

7470: 

7471: 0) \  $\lambda$ has no sphere leaves and

7472: 

7473: 1) \ for any leaf $L$ of $\lambda$, the homomorphism

7474: $\pi_1(L) \rightarrow \pi_1(M)$ induced by inclusion

7475: is injective, and for any closed complementary region

7476: $V$, the homomorphism $\pi_1(V) \rightarrow \pi_1(M)$

7477: induced by inclusion is injective.

7478: 

7479: Gabai \cite{Ga5} conjectured that under

7480: these conditions and $M$ closed then

7481: $\widetilde \lambda$ is a product

7482: lamination and $\mi$ is homeomorphic to $\rrrr^3$.

7483: One test case is the class of manifolds $M_{p/q}$ studied

7484: in this article.  When

7485: $|p-2q| = 1$ the lamination coming from the stable

7486: lamination has monogons. The leaves are either

7487: planes or have ${\bf Z}$ fundamental group.

7488: The complementary region is a solid torus.

7489: So to check for loosesse one only needs to 

7490: understand if leaves inject in the fundamental group

7491: level.

7492: 

7493: 

7494: 

7495: 

7496: 

7497: {\footnotesize

7498: {

7499: \setlength{\baselineskip}{0.01cm}

7500: 

7501: \begin{thebibliography}{Ga-Ka1}

7502: 

7503: {

7504: 

7505: %\setlength{\baselineskip}{0pt}

7506: %\baselineskip0pt

7507: 

7508: 

7509: 

7510: \bibitem[Bk1]{Bk1} M. Baker, {\em Covers of Dehn fillings on

7511: 	once punctured bundles}, Proc. A.M.S. {\bf 105} (1989) 747-754.

7512: 

7513: \bibitem[Bk2]{Bk2} M. Baker, {\em Covers of Dehn fillings on

7514: 	once punctured bundles, II}, Proc. A.M.S. {\bf 110} (1990) 1099-1108.

7515: 

7516: \bibitem[Ba1]{Ba1} T. Barbot, {\em Flots d'Anosov sur les vari\'{e}t\'{e}s

7517: 	graph\'{e}es au sens de Waldhausen"}, Ann. Inst. Fourier

7518: 	Grenoble {\bf 46} (1996) 1451-1517.

7519: 

7520: \bibitem[Ba2]{Ba2} T. Barbot, {\em Actions de groups sur les 

7521: 	$1$-vari\'{e}t\'{e}s non s\'{e}par\'{e}es et feuilletages de

7522: 	codimension un}, Ann. Fac. Sci. Toulose Math. {\bf 7} (1998) 559-597.

7523: 

7524: \bibitem[BNS]{BNS} R. Bieri, W. Neumann and R. Strebel, {\em A geometric

7525: 	invariant of discrte groups}, Inven. Math. {\bf 90} (1987) 451-477.

7526: 

7527: \bibitem[Bl-Ca]{Bl-Ca} S. Bleiler and A. Casson, {\em Automorphims of

7528: 	surfaces after Nielsen and Thurston}, Cambridge Univ. Press, 1988.

7529: 	

7530: 

7531: 

7532: \bibitem[Br1]{Br1} M. Brittenham, {\em Essential laminations in Seifert 

7533: 	fibered spaces},  Topology {\bf 32} (1993) 61-85.

7534: 

7535: \bibitem[Br2]{Br2} M. Brittenham, {\em Essential laminations and deformations

7536: 	of homotopy equivalences: from essential pullback to

7537: 	homeomorphism}, Topol. Appl. {\bf 60} (1994) 249-265.

7538: 

7539: 

7540: \bibitem[Br3]{Br3} M. Brittenham, {\em Persistently laminar tangles},

7541: 	Jour. Knot Th. Ram. {\bf 8} (1999) 415-428.

7542: 

7543: \bibitem[Br4]{Br4} M. Brittenham, {\em Persistent laminations from

7544: 	Seifert surfaces}, Jour. Knot Th. Ram. {\bf 10} (2001) 1155-1168.

7545: 

7546: \bibitem[Bu-Zi]{Bu-Zi} G. Burde and H. Zieschang, {\em Knots},

7547: 	Studies in Mathematics {\bf 5}, de Gruyter, Berlin, 1985.

7548: 

7549: 

7550: %\bibitem[Cal1]{Cal1} D. Calegari, {\em The geometry of $\rrrr$-covered

7551: %foliations}, Geom. Topol. {\bf 4} (2000) 457-515 (eletronic).

7552: 

7553: %\bibitem[Cal2]{Cal2} D. Calegari, {\em $\rrrr$-covered foliations of

7554: %hyperbolic $3$-manifolds}, Geom. Topol. {\bf 3} (1999) 137-153 (eletronic).

7555: 

7556: \bibitem[Ca-Du]{Ca-Du} D. Calegari and N. Dunfield, {\em Laminations

7557: 	and groups of homeomorphisms of the circle}, preprint, 2002.

7558: 

7559: 

7560: \bibitem[CJR]{CJR} M. Culler, W. Jaco and H. Rubinstein,

7561: 	{\em Incompressible surfaces in once punctured torus bundles},

7562: 	Proc. L.M.S. {\bf 45} (1982) 385-419.

7563: 

7564: 

7565: \bibitem[Cu-Vo]{Cu-Vo} M. Culler and K. Vogtman, {\em A group theoretic

7566: 	criterion for property FA}, Proc. A.M.S. {\bf 124} (1996) 

7567: 	676-683.

7568: 

7569: 

7570: 

7571: \bibitem[De-Ro]{De-Ro} C. Delman and R. Roberts, {\em Alternating knots

7572: 	satisfy strong property P}, Comm. Math. Helv. {\bf 74}

7573: 	(1999) 376-397.

7574: 

7575: 

7576: 

7577: 

7578: 

7579: 

7580: 

7581: %\bibitem[Fe1]{Fe1} S. Fenley, {\em The structure of branching

7582: %in Anosov flows of $3$-manifolds}, 

7583: 	Comm. Math. Helv. {\bf 73} (1998) 259-297.

7584: 

7585: 

7586: \bibitem[Fe]{Fe} S. Fenley, {\em Pseudo-Anosov flows and incompressible

7587: 	tori}, submitted.

7588: 

7589: \bibitem[Fl-Ha]{Fl-Ha} W. Floyd and A. Hatcher, {\em Incompressible

7590: 	surfaces in punctured torus bundles}, Topol. Appl. 

7591: 	{\bf 13} (1982) 263-282.

7592: 

7593: 

7594: 

7595: 

7596: \bibitem[Ga1]{Ga1} D. Gabai, {\em Foliations and the topology of 3-manifolds},

7597: 	J. Diff. Geo. {\bf 18} (1983) 445--503.

7598: 

7599: \bibitem[Ga2]{Ga2} D. Gabai, {\em Foliations and the topology 

7600: 	of 3-manifolds II},

7601: 	J. Diff. Geo. {\bf 26} (1987) 461--478.

7602: 

7603: \bibitem[Ga3]{Ga3} D. Gabai, {\em Foliations and the topology 

7604: 	of 3-manifolds III},

7605: 	J. Diff. Geo. {\bf 26} (1987) 479--536.

7606: 

7607: \bibitem[Ga4]{Ga4} D. Gabai, {\em $8$ problems in foliations

7608: 	and laminations}, in Geometric Topology, W. Kazez, ed.,

7609: 	Amer. Math. Soc. 1987, 1-33.

7610: 

7611: \bibitem[Ga5]{Ga5} D. Gabai, {\em $3$ lectures on foliations

7612: 	and laminations on $3$-manifolds}, Laminations and foliations

7613: 	in dynamics, geometry and topology (Stony Brook, NY 1998),

7614: 	87-109, Contemp. Math., 269, Amer. Math. Soc. Providence, 2001.

7615: 

7616: \bibitem[Ga-Ka1]{Ga-Ka1} D. Gabai and W. Kazez, {\em Pseudo-Anosov 

7617: 	maps and surgery on fibred $2$-bridge knots}, Topol. Appl.

7618: 	{\bf 37} (1990) 93-100.

7619: 

7620: \bibitem[Ga-Ka2]{Ga-Ka2} D. Gabai and W. Kazez, {\em Order trees and

7621: 	laminations of the plane}, Math. Res. Lett. {\bf 4} (1997) 603-616.

7622: 

7623: \bibitem[Ga-Ka3]{Ga-Ka3} D. Gabai and W. Kazez, {\em Homotopy, isotopy

7624: 	and genuine laminations of $3$-manifolds}, in Geometric Topology

7625: 	(Athens, GA 1993) 123-138, AMS/IP Stud. Adv. Math., 2.1,

7626: 	Amer. Math. Soc., Providence, RI, 1997.

7627: 

7628: \bibitem[Ga-Ka4]{Ga-Ka4} D. Gabai and W. Kazez, {\em Group negative curvature

7629: 	for $3$-manifolds with genuine laminations}, Geom. Topol.

7630: 	{\bf 2} (1998) 65-77.

7631: 

7632: 

7633: \bibitem[Ga-Oe]{Ga-Oe} D. Gabai and U. Oertel, {\em Essential

7634: 	laminations and $3$-manifolds}, Ann. of Math.

7635: 	{\bf 130} (1989) 41-73.

7636: 

7637: \bibitem[Gh-Ha]{Gh-Ha} {\em Sur les groupes hyperboliques d'apr\`{e}s

7638: 	Mikhael Gromov}, E. Ghys and P. De la Harpe, editors,

7639: 	Birkh\"{a}user, 1990.

7640: 

7641: \bibitem[Gr]{Gr} M. Gromov, {\em Hyperbolic groups}, in Essays on

7642: 	Group theory, Springer-Verlag, 1987, pp. 75-263.

7643: 

7644: 

7645: \bibitem[Hak1]{Hak1} W. Haken, {\em Theorie der normal fl\"{a}chen},

7646: 	Acta Math. {\bf 105} (1961) 245-375.

7647: 

7648: \bibitem[Hak2]{Hak2} W. Haken, {\em Some results on surfaces

7649: 	in $3$-manifolds}, Studies in modern topology, Math. Assoc.

7650: 	Amer., distributed by Prentice Hall (1968) 39-98.

7651: 

7652: \bibitem[Hat1]{Hat1} A. Hatcher, {\em On the boundary curves

7653: 	of incompressible surfaces}, Pac. Jour. Math. {\bf 99} 

7654: 	(1982) 373-377.

7655: 

7656: \bibitem[Hat2]{Hat2} A. Hatcher, {\em Some examples of essential

7657: 	laminations in $3$-manifolds}, Ann. Inst. Fourier Grenoble

7658: 	{\bf 42} (1992) 313-325.

7659: 

7660: \bibitem[Ha-Oe]{Ha-Oe} A. Hatcher and U. Oertel, {\em Boundary slopes 

7661: 	for Montesinos knots}, Topology {\bf 28} (1989) 453-480.

7662: 

7663: \bibitem[Ha-Th]{Ha-Th} A. Hatcher and W. Thurston, {\em Incompressible

7664: 	surfaces in $2$-bridge knot complements}, Inven. Math.

7665: 	{\bf 79} (1985) 225-246.

7666: 

7667: 

7668: 

7669: \bibitem[He]{He} J. Hempel, {\em 3-manifolds}, Ann. of Math. Studies 86,

7670: 	Princeton University Press, 1976.

7671: 

7672: \bibitem[Jo]{Jo} T. Jorgenson, {\em Compact $3$-manifolds of

7673: 	constant negative curvature}, Annals of Math. 

7674: 	{\bf 106} (1977) 61-72.

7675: 

7676: 

7677: \bibitem[Mo-Sh1]{Mo-Sh1} J. Morgan and P. Shalen, {\em Valuations, trees

7678: 	and degenerations of hyperbolic structures, I}, Ann. of Math.

7679: 	{\bf 120} (1984) 401-476.

7680: 

7681: \bibitem[Mo-Sh2]{Mo-Sh2} J. Morgan and P. Shalen, {\em Degenerations of

7682: 	hyperbolic structures, II. Measured laminations in $3$-manifolds},

7683: 	Ann. of Math. {\bf 127} (1988) 403-456.

7684: 

7685: 

7686: \bibitem[Mo-Sh3]{Mo-Sh3} J. Morgan and P. Shalen, {\em Degenerations of

7687: 	hyperbolic structures, III. Actions of $3$-manifold groups

7688: 	on trees and Thurston's compactness theorem}, Ann. of Math.

7689: 	{\bf 127} (1988) 457-519.

7690: 

7691: 

7692: 

7693: 

7694: \bibitem[Mo1]{Mo1} L. Mosher, {\em Dynamical systems and the homology norm

7695: 	of a $3$-manifold I. Efficient interesection of surfaces and flows},

7696: 	Duke Math. Jour.  {\bf 65} (1992) 449--500.

7697: 

7698: %\bibitem[Mo2]{Mo2} L. Mosher, {\em Dynamical systems and the homology 

7699: %norm of a $3$-manifold II},

7700: %	Invent. Math. {\bf 107} (1992) 243--281.

7701: 

7702: %\bibitem[Mo3]{Mo3} L. Mosher, {\em Examples of quasigeodesic flows on

7703: %hyperbolic $3$-manifolds}, in Proceedings of the Ohio State

7704: %University Research Semester on Low-Dimensional topology,

7705: %W. de Gruyter, 1992.

7706: 

7707: \bibitem[Mo2]{Mo2} L. Mosher, {\em Laminations and flows transverse

7708: 	to finite depth foliations}, Part I: Branched surfaces and

7709: 	dynamics $-$ available from

7710: 	http://newark.rutgers.edu:80/~mosher/, 

7711: 	Part II in preparation.

7712: 

7713: 

7714: \bibitem[Na]{Na} R. Naimi, {\em Foliations transverse to fibers of

7715: 	Seifert manifolds}, Comm. Math. Helv., {\bf 69} (1994) 155-162.

7716: 

7717: 

7718: \bibitem[No]{No} S. P. Novikov, {\em Topology of foliations}, 

7719: 	Trans. Moscow Math. Soc. {\bf 14} (1963) 268-305.

7720: 

7721: \bibitem[Oe]{Oe} U. Oertel, {\em Closed incompressible surfaces

7722: 	in complements of star links}, Pac. Jour. Math. {\bf 111}

7723: 	(1984) 209-230.

7724: 

7725: 

7726: \bibitem[Re]{Re} G. Reeb, {\em Sur certains propri\'{e}t\'{e}s

7727: 	topologiques des vari\'{e}t\'{e}s feuillet\'{e}es},

7728: 	Actual. Sci. Ind. {\bf 1183}, Hermann, Paris, 1952.

7729: 

7730: 

7731: 

7732: \bibitem[Ro1]{Ro1} R. Roberts, {\em Constructing taut foliations},

7733: 	Comm. Math. Helv. {\bf 70} (1995) 516-545.

7734: 

7735: \bibitem[Ro2]{Ro2} R. Roberts, {\em Taut foliations in

7736: 	punctured surfaces bundles, I}, Proc. London Math.

7737: 	Soc. (3) {\bf 82} (2001) 747-768.

7738: 		

7739: \bibitem[Ro3]{Ro3} R. Roberts, {\em Taut foliations in

7740: 	punctured surface bundles, II}, Proc. London Math.

7741: 	Soc. (3) {\bf 83} (2001) 443-471.

7742: 

7743: 

7744: \bibitem[Ro-St1]{Ro-St1} R. Roberts and M. Stein, {\em Exceptional

7745: 	Seifert group actions on $\rrrr$}, Jour. Knot Th. Ram.

7746: 	{\bf 8} (1999) 241-247.

7747: 

7748: \bibitem[Ro-St2]{Ro-St2} R. Roberts and M. Stein, {\em Group actions

7749: 	on order trees}, Topol. Appl. {\bf 115} (2001) 175-201.

7750: 

7751: \bibitem[RSS]{RSS} R. Roberts, J. Shareshian and M. Stein,

7752: 	{\em Infinitely many hyperbolic $3$-manifolds which

7753: 	contains no Reebless foliation}, preprint, 2001.

7754: 

7755: 

7756: \bibitem[Rol]{Rol} D. Rolfsen, {\em Knots and links}, Publish or Perish, 1976.

7757: 	

7758: 

7759: \bibitem[Sc]{Sc} P. Scott, {\em The geometries of $3$-manifolds},

7760: 	Bull. London Math. Soc. {\bf 15} (1983) 401-487.

7761: 

7762: \bibitem[Se]{Se} J. P. Serre, {\em Trees}, Springer, Berlin 1980.

7763: 

7764: \bibitem[Th1]{Th1} W. Thurston, {\em The geometry and topology of 3-manifolds},

7765: 	Princeton University Lecture Notes, 1982.	

7766: 

7767: \bibitem[Th2]{Th2}  W. Thurston, {\em Three dimensional manifolds, Kleinian

7768: 	groups and hyperbolic geometry}, Bull. A.M.S. {\bf 6} (1982) 357-381.

7769: 

7770: \bibitem[Th3]{Th3}  W. Thurston, {\em Hyperbolic structures on $3$-manifolds II:

7771: 	Surface groups and $3$-manifolds that fiber over the circle}, preprint.

7772: 

7773: \bibitem[Th4]{Th4}  W. Thurston, {\em On the geometry and dynamics of

7774: 	diffeomorphisms of surfaces}, Bull. A.M.S. {19} (1988) 417-431.	

7775: 

7776: 

7777: \bibitem[Th5]{Th5}  W. Thurston, {\em 3-manifolds, foliations and circles II:

7778: 	Transverse asymptotic geometry of foliations}, preprint.

7779: 

7780: \bibitem[Th6]{Th6}  W. Thurston, {\em The universal circle for foliations

7781: 	and transverse laminations}, lectures in M.S.R.I., 1997.

7782: 

7783: 

7784: 

7785: 

7786: 

7787: \bibitem[Wa]{Wa} K. Waldhausen, {\em On irreducible $3$-Manifolds which are sufficiently

7788: 	large}, Ann. of Math. {\bf 87} (1968) 56-88. 

7789: 

7790: 

7791: 

7792: 

7793: }

7794: \end{thebibliography}

7795: 

7796: 

7797: \noindent

7798: Florida State University

7799: 

7800: \noindent

7801: Tallahassee, FL 32306-4510

7802: 

7803: }

7804: }

7805: 

7806: \end{document}

7807: 

7808: