One can study stability properties of a closed geodesic
$c$
with the help of the linearized Poincaré mapping
${P}_{c},$
which is a linear symplectic map of an
$(2n2)$
dimensional vector space. It can be defined using the Jacobi fields along this geodesic, cf. [
BTZ1,ch.1]
, [
Ra2]
. In the most unstable case no (complex) eigenvalue of
${P}_{c}$
lies on the unit circle, then the closed geodesic is called hyperbolic. For example all closed geodesics on a Finsler manifold with negative flag curvature are hyperbolic. We obtain a result similar to [
BTZ1,Thm.B]
:
Theorem 4
Let
$F$
be a Finsler metric on a compact manifold with reversibility
$\lambda $
and flag curvature
$0<\delta \le K\le 1.$
There exists a nonhyperbolic closed geodesic if the
$l$
–th homotopy group
${\pi}_{l}\left(M\right)$
is nontrivial for some
$l\ge 2$
and
$\sqrt{\delta}>\frac{l1}{n1}\frac{\lambda}{\lambda +1}.$
In particular on the
$n$
sphere
${S}^{n}$
with a Finsler metric satisfying
${\lambda}^{2}/(\lambda +1{)}^{2}<K\le 1$
there exists a nonhyperbolic closed geodesic. A more detailed analysis also produces existence results for closed geodesics of ellipticparabolic type. Here a closed geodesic is called of ellipticparabolic type if the linearized Poincaré map splits into twodimensional rotations and a part whose eigenvalues are
$\pm 1.$
Following the ideas of Thorbergsson [
Th]
and Ballmann, Thorbergsson and Ziller [
BTZ1]
we obtain as another consequence of the length estimate Theorem 1 :
Theorem 5
On a compact Finsler manifold
$M$
with reversibility
$\lambda $
and flag curvature
$0<\delta \le K\le 1$
there exists a closed geodesic of ellipticparabolic type if one of the following conditions is satisfied:

(a)
$M={S}^{n}$
and
$\delta >\frac{9}{4}\frac{{\lambda}^{2}}{(\lambda +1{)}^{2}}$
with
$\lambda <2.$

(b)
$M=\mathbb{R}{P}^{n}$
and
$\delta >\frac{{\lambda}^{2}}{(\lambda +1{)}^{2}}.$
It is mentioned in [
BTZ3,p.61]
that most of the results presented in the Riemannian case generalize to Finsler metrics. For example it is stated that a Finsler metric with
$9/16<K\le 1$
carries a short closed geodesic of ellipticparabolic type. But the arguments only work for reversible Finsler metrics respectively under the additional assumption that a shortest closed geodesic has length
$\ge 2\pi .$
Another setting in which one can show the existence of a closed geodesic of ellipticparabolic type is in the presence of an isometric
${S}^{1}$
action. For example the above mentioned Katok metrics
${F}_{\epsilon}$
on the
$2$
sphere carry an isometric
${S}^{1}$
action. For irrational parameter
$\epsilon $
there are exactly two geometrically distinct closed geodesiscs which both are elliptic and invariant under the
${S}^{1}$
action. As an analogous result to [
BTZ1,TheoremA(iii)]
we obtain:
Theorem 6
On a compact manifold with Finsler metric with an isometric
${S}^{1}$
action there exist at least two geometrically distinct closed geodesics. These closed geodesics are
${S}^{1}$
invariant and they are of ellipticparabolic type.
2 Critical Point Theory for Closed Geodesics
Here we list a couple of results of the critical point theory for closed geodesics, general references are the survey article [
Ba]
by Bangert, the book [
Kl1]
by Klingenberg and [
Ra2]
.
If
$c:{S}^{1}:=[0,1]/\{0,1\}\to M$
is a closed geodesic on the Finsler manifold
$(M,F)$
of length
$L\left(c\right)$
then for every positive integer
$m$
the
$m$
fold cover
${c}^{m}:{S}^{1}\to M;{c}^{m}\left(t\right)=c\left(mt\right)$
is a closed geodesic, too. If
$L\left(c\right)={\int}_{0}^{1}F\left({c}^{\prime}\right(t\left)\right)dt$
denotes the length, then we have
$L\left({c}^{m}\right)=mL\left(c\right).$
We call a closed geodesic prime if it is not the cover
${c}_{0}^{m}$
of another closed geodesic
${c}_{0}$
with
$m>1.$
Closed geodesics are the critical points of the energy functional
$$E:\Lambda M\to \mathbb{R};E\left(c\right)=\frac{1}{2}{\int}_{0}^{1}{F}^{2}\left({c}^{\prime}\left(t\right)\right)dt$$
on the Hilbert manifold
$\Lambda M$
of closed curves which is the set of all absolutely continuous closed curves with a squareintegrable derivative.
The Morse index
$ind\left(c\right)$
of a closed geodesic is the index of the hessian
${d}^{2}E\left(c\right)$
of the energy functional. On the space
$\Lambda M$
there is a
${S}^{1}$
action given by changing the initial point. The energy functional is invariant under this group action. We call two closed geodesics
${c}_{1},{c}_{2}$
of a nonreversible Finsler metric geometrically equivalent, if their traces
${c}_{1}\left({S}^{1}\right)={c}_{2}\left({S}^{1}\right)$
and their orientation coincide. Otherwise we call them geometrically distinct. In contrast to the reversible case resp. the case of a Riemannian metric for a closed geodesic
$c$
the curve
${c}^{1}$
with
${c}^{1}\left(t\right)=c(1t)$
defined by reversing the orientation in general is not a geodesic. Hence a prime closed geodesic
$c$
produces infinitely many critical orbits
${S}^{1}.{c}^{m};m\ge 1$
of the energy functional consisting of all geometrically equivalent closed geodesics. If a closed geodesic
$c$
is the
$m$
fold cover
$c={c}_{0}^{m}$
of a prime closed geodesic
${c}_{0}$
then we call
$m=mul\left(c\right)$
the multiplicity of the closed geodesic
$c.$
Therefore a prime closed geodesic
$c$
produces a tower
${S}^{1}.{c}^{m};m\ge 1$
of closed geodesics resp. critical orbits of the energy functional.
We can view the hessian of the energy functional also as a selfadjoint endomorphism.
Then the index is the sum of the dimensions of negative eigenvalues and we call the
nullity
$null\left(c\right)$
the dimension of the kernel
$ker{d}^{2}E\left(c\right)$
minus
$1.$
Note that the dimension of the kernel is always at least
$1$
provided
$L\left(c\right)>0$
since there is a
$1$
dimensional group leaving the energy functional invariant. A closed geodesic
$c$
is called nondegenerate if
$null\left(c\right)=0.$
Geometrically the nullity is the dimension of periodic Jacobi fields along
$c$
which are orthogonal to the velocity field
${c}^{\prime}.$
Therefore
$null\left(c\right)\le 2n2.$
The sequence
$ind\left({c}^{m}\right)$
grows almost linearly, we call the limit
$${\alpha}_{c}:={lim}_{m\to \infty}\frac{ind\left({c}^{m}\right)}{m}$$
introduced by Bott [
Bo,Cor.1]
the average index and
${\overline{\alpha}}_{c}={\alpha}_{c}/L\left(c\right)$
the mean average index. We have the following estimate for the sequence
$ind\left({c}^{m}\right):$
$$\begin{array}{c}\leftind\left({c}^{m}\right)m{\alpha}_{c}\right\le n1,\end{array}$$ 
(1)

cf. [
Ra1,(1.4)]
. By a Rauch comparison argument as in the Riemannian case one obtains
Lemma 1
([
Ra4,Lem.3]
) Let
$c$
be a closed geodesic on a Finsler manifold
$(M,F)$
of dimension
$n$
with positive flag curvature
$K\ge \delta $
for some
$\delta \in {\mathbb{R}}^{+}.$

(a)
The mean average index is bounded from below:
${\overline{\alpha}}_{c}\ge \sqrt{\delta}(n1)/\pi .$

(b)
If the length
$L\left(c\right)$
satisfies
$L\left(c\right)>k\pi /\sqrt{\delta}$
for some positive integer
$k$
then
$ind\left(c\right)\ge k(n1).$
Combining Lemma 1 with the length estimate Theorem 1 for a closed geodesic we obtain:
Lemma 2
Let
$c$
be a closed geodesic on a compact and simplyconnected Riemannian manifold of dimension
$n$
with a nonreversible Finsler metric with reversibility
$\lambda $
and flag curvature
$0<\delta \le K\le 1$
where
$\delta >\frac{{\lambda}^{2}}{(\lambda +1{)}^{2}}$
if
$n$
is odd.
Then
$${\alpha}_{c}\ge \sqrt{\delta}\frac{\lambda +1}{\lambda}\left(n1\right).$$
Proof. Since
$L\left(c\right)\ge \pi \left(1+1/\lambda \right)$
by Theorem 1 the claim follows from Lemma 1 .
$\square $
Now we come to the Morse Inequalities of the
${S}^{1}$
invariant functional
$E:\Lambda M\to \mathbb{R}.$
Let
$${\overline{b}}_{j}:={b}_{j}\left(\Lambda M/{S}^{1},{\Lambda}^{0}M/{S}^{1};\mathbb{Q}\right)$$
where for
$a\ge 0$
we denote
${\Lambda}^{a}M:=\left\{\sigma \in \Lambda M\leftE\right(a)\le a\right\}$
the sublevel sets and
${b}_{j}$
is the
$j$
th Betti number. In particular
${\Lambda}^{0}M$
is the set of point curves which can be identified with the manifold
$M.$
Since it is the fixed point set of the
${S}^{1}$
action one can also identify the quotient space
${\Lambda}^{0}M/{S}^{1}$
with the manifold
$M.$
Given a closed geodesic
$c$
we use the following notation
$$\Lambda \left(c\right):=\left\{\sigma \in \Lambda M\leftE\right(\sigma )<E(c)\right\}.$$
Then we call
$${\overline{C}}_{*}\left(c\right)={H}_{*}\left(\left(\Lambda \left(c\right)\cup {S}^{1}.c\right)/{S}^{1},\Lambda \left(c\right)/{S}^{1};\mathbb{Q}\right)$$
the
${S}^{1}$
critical group of the closed geodesic
$c$
and let
${\overline{b}}_{j}\left(c\right)=dim{\overline{C}}_{j}\left(c\right).$
We collect the information about the
${S}^{1}$
critical groups in the following two lemmas:
Lemma 3
([
Ra1,ProofofProp.2.2]
) Let
$c$
be a nondegenerate closed geodesic with
$i=ind\left(c\right),m=mul\left(c\right).$
Then
$c={c}_{1}^{m}$
for a prime closed geodesic
${c}_{1}$
and
$${\overline{b}}_{j}\left(c\right)=\{\begin{array}{ccc}1& ;& j=i;m\equiv 1(mod2)\\ 1& ;& j=i;m\equiv 0(mod2)\\ & & \text{and}ind\left({c}_{1}^{2}\right)\equiv ind\left({c}_{1}\right)(mod2)\\ 0& ;& \text{otherwise}\end{array}$$
In the general case we obtain the following cases:
Lemma 4
[
Ra2,Satz6.13]
Let
$c$
be a closed geodesic with
$i=ind\left(c\right);l=null\left(c\right).$
Then we have the following statements:

(a)
${\overline{b}}_{j}\left(c\right)=0$
for
$j<i$
or
$j>i+l.$

(b)
${\overline{b}}_{i}\left(c\right)+{\overline{b}}_{i+l}\le 1$
and if
${\overline{b}}_{i}\left(c\right)+{\overline{b}}_{i+l}=1$
then
${\overline{b}}_{j}\left(c\right)=0$
for all
$j$
with
$i+1\le j\le i+l1.$
As a consequence of the formula for the sequence
$ind\left({c}^{m}\right),m\ge 1$
given by Bott [Bo] we conclude:
Lemma 5
Let
$c$
be a closed geodesic on a surface (i.e.
$n=dimM=2$
) with
$ind\left(c\right)=1$
and average index
${\alpha}_{c}>1.$
Then for all
$m\ge 1$
the indices
$ind\left({c}^{m}\right)$
are odd.
Proof. Let
${P}_{c}$
be the linearized Poincaré mapping, i.e. the linearization of the return map of the closed orbit of the geodesic flow corresponding to the closed geodesic. There is a function
$${I}_{c}:{S}^{1}:=\left\{z\in \mathbb{C};\leftz\right=1\right\}\to {\mathbb{Z}}^{\ge 0}$$
with the following properties, cf. [Bo,Thm.A,C] ,[Lo,ch.9] ,[Ra2,ch.4] :

(a)
$ind\left({c}^{m}\right)={\sum}_{{z}^{m}=1}{I}_{c}\left(z\right)$

(b)
Define
${N}_{c}:{S}^{1}\to {\mathbb{Z}}^{\ge 0}:$
${N}_{c}\left(z\right)=dimker({P}_{c}zId).$
Then
$$null\left({c}^{m}\right)={\sum}_{{z}^{m}=1}{N}_{c}\left(z\right).$$

(c)
The function
${I}_{c}$
is constant in a neighborhood of points
$z$
with
${N}_{c}\left(z\right)=0.$
For the splitting numbers
$${S}_{c}^{\pm}\left(z\right);={lim}_{\phi \to \pm 0}{I}_{c}(zexp(i\phi \left)\right){I}_{c}\left(z\right)$$
of the function
${I}_{c}$
the following estimate holds:
$$0\le {S}_{c}^{\pm}\left(z\right)\le {N}_{c}\left(z\right)$$

(d)
${I}_{c}\left(z\right)={I}_{c}\left(\overline{z}\right),{N}_{c}\left(z\right)={N}_{c}\left(\overline{z}\right)$
It follows that
$ind\left({c}^{m}\right)\equiv ind\left(c\right)(mod2)$
for all odd
$m$
and
$ind\left({c}^{m}\right)\equiv {I}_{c}\left(1\right)+{I}_{c}(1)=ind\left({c}^{2}\right)(mod2)$
for all even
$m.$
It was also shown by Bott that the splitting numbers only depend on the symplectic normal form of the linearized Poincaré map
${P}_{c},$
for a detailed discussion see [BTZ1,(2.13)] , [Lo,ch.IV] , [Ra2,ch.4] .
Now we come to the case
$n=2,$
then for the eigenvalues
$z$
of the linearized Poincaré map there are the following cases:

(a)
$z\notin {S}^{1},$
i.e.
$z$
is a real number with
$\leftz\right\ne 1.$
Then also
${z}^{1}$
is an eigenvalue, in this case the closed geodesic is called hyperbolic and
$ind\left({c}^{m}\right)=mind\left(c\right).$
In particular the average index satisfies
${\alpha}_{c}=1$
in contradiction to our assumption.

(b)
If
$z=1$
then
${S}^{+}\left(1\right)={S}^{}\left(1\right)$
and
${\alpha}_{c}={I}_{c}(1)=ind\left({c}^{2}\right)ind\left(c\right)=ind\left(c\right)+{S}^{+}\left(1\right)=1+{S}^{+}\left(1\right)>1$
by assumption, hence we conclude
${S}^{+}\left(1\right)=1.$
It follows that
$ind\left({c}^{2}\right)=3$
and
${\alpha}_{c}=2.$

(c)
If
$z=1$
then
${\alpha}_{c}=ind\left(c\right)=1$
in contradiction to our assumption.

(d)
If
$z=exp\left(\sqrt{1}\pi \rho \right)$
with
$\rho \in (0,1)$
then we conclude from [BTZ1,(2.13)] or [Ra2,Thm.4.3] :
${S}^{+}\left(z\right)+{S}^{}\left(z\right)=1.$
Since
${\alpha}_{c}=ind\left(c\right)+\left({S}^{+}\right(z){S}^{}(z\left)\right)\rho >1$
we conclude
${S}^{+}\left(z\right)=1,{S}^{}\left(z\right)=0,{\alpha}_{c}\in (1,2)$
Hence in this case
$ind\left({c}^{2}\right)={I}_{c}\left(1\right)+{I}_{c}(1)=2ind\left(c\right)+{S}^{+}\left(z\right)=3.$
Therefore
$ind\left({c}^{2}\right)=3$
which implies that for all even
$m$
the indices
$ind\left({c}^{m}\right)$
are odd, too.
$\square $
Now Lemma 4 and Lemma 5 imply the following
Corollary 1
Let
$c$
be a prime closed geodesic on a surface (i.e.
$n=2$
) with Finsler metric with index
$ind\left(c\right)=1$
and average index
${\alpha}_{c}>1.$
Then for every
$m\ge 1:$
$${\sum}_{j\equiv 1(mod2)}{\overline{b}}_{j}\left({c}^{m}\right)\le 1.$$
The Morse inequalities relate the critical groups as local information about the critical points with the global topological information given by the Betti numbers of the space on which the Morse function is defined.
Lemma 6
([
Ra1,2.6]
) The rational Betti numbers
${\beta}_{i}:={b}_{i}\left(\Lambda {S}^{2}/{S}^{1},{\Lambda}^{0}{S}^{2}/{S}^{1};\mathbb{Q}\right),$
of the pair of quotient spaces
$\left(\Lambda {S}^{2}/{S}^{1},{\Lambda}^{0}{S}^{2}/{S}^{1}\right)$
are given by:
$${\beta}_{i}=\{\begin{array}{ccc}2& ;& i=2m+1,m\ge 1\\ 1& ;& i=1\\ 0& ;& i=2m,m\ge 0\end{array}$$
3 Proof of Theorem 2
Let
$N$
be the odd integer satisfying
$$\begin{array}{c}N2\le \frac{1}{\sqrt{\delta}\frac{\lambda +1}{\lambda}1}<N.\end{array}$$ 
(2)

We assume that there is only one class of geometrically equivalent closed geodesics whose indices are bounded from above by
$N.$
Hence there is a prime closed geodesic
$c$
such that every closed geodesic
$\stackrel{~}{c}$
with
$ind\left(\stackrel{~}{c}\right)\le N$
is up to the choice of the initial point of the form
${c}^{m}$
respectively
$\stackrel{~}{c}\in {S}^{1}.{c}^{m}$
for some
$m\ge 1.$
We define for all
$i$
with
$0\le i\le N:{v}_{i}:={\sum}_{m\ge 1}{\overline{b}}_{i}\left({c}^{m}\right),$
then the Morse Inequalities for the
${S}^{1}$
invariant energy functional
$E:\Lambda {S}^{2}\to \mathbb{R}$
yield (cf. [
Ra2,ch.6.1]
):
$$\begin{array}{c}{v}_{i}\ge {\beta}_{i}\end{array}$$ 
(3)

for all
$i$
with
$0\le i\le N.$
In particular we conclude from
${\beta}_{1}=1$
that for some
$m\ge 1:ind\left({c}^{m}\right)\le 1.$
Since
$L\left(c\right)\ge \pi \frac{\lambda +1}{\lambda}$
and
$K\ge \delta >{\left(\frac{\lambda}{\lambda +1}\right)}^{2}$
we obtain from Lemma 1 that
$ind\left({c}^{m}\right)\ge 1$
for all
$m\ge 1.$
Hence we have finally shown:
$ind\left(c\right)=1.$
As an estimate for the average index we obtain from Lemma 1 :
${\alpha}_{c}\ge \sqrt{\delta}(\lambda +1)/\lambda >1.$
Inequality ( 1 ) and Inequality ( 2 ) imply:
$$ind\left({c}^{N}\right)\ge N{\alpha}_{c}1\ge N\sqrt{\delta}\frac{\lambda +1}{\lambda}1=N\left(\sqrt{\delta}\frac{\lambda +1}{\lambda}1\right)+N1>N.$$
Therefore Corollary 1 implies:
$${\sum}_{\begin{array}{c}0\le i\le N\\ i\equiv 1(mod2)\end{array}}{v}_{i}={\sum}_{\begin{array}{c}0\le i\le N;m\ge 1\\ i\equiv 1(mod2)\end{array}}{\overline{b}}_{i}\left({c}^{m}\right)\le \#\left\{mind\left({c}^{m}\right)\le N\right\}<N.$$
This contradicts the Morse Inequalities ( 3 ) since by Lemma 6 for
$N$
odd:
$${\sum}_{\begin{array}{c}0\le i\le N\\ i\equiv 1(mod2)\end{array}}{v}_{i}\ge {\sum}_{\begin{array}{c}0\le i\le N\\ i\equiv 1(mod2)\end{array}}{\beta}_{i}=N.$$
Hence there are at least two geometrically distinct closed geodesics
${c}_{1},{c}_{2}$
with
$ind\left({c}_{1}\right)=1;ind\left({c}_{2}\right)\le N.$
We conclude from Lemma 1 and Inequality ( 2 ):
$$L\left({c}_{2}\right)\le \frac{\pi}{\sqrt{\delta}}(N+1)\le \frac{\pi}{\sqrt{\delta}}\left(\frac{1}{\sqrt{\delta}\frac{\lambda +1}{\lambda}1}+3\right)$$
Since
$ind\left({c}_{1}\right)=1$
Lemma 1 (b) implies that
$L\left({c}_{1}\right)\le 2\pi /\sqrt{\delta}.$
For the given curvature bounds not only the length of a closed geodesic but also the length of a geodesic loop is bounded from above by
$\pi (1+{\lambda}^{1}),$
cf. [
Ra4,Thm.1]
, hence
${c}_{1}$
is simple since
$2\pi /\sqrt{\delta}<2\pi (1+{\lambda}^{1}).$
$\square $
4 Existence results in higher dimensions
We consider a compact and simplyconnected manifold
$M$
whose rational cohomology algebra is generated by a single element
$x\in {H}^{d}(M;\mathbb{Q})$
of degree
$d,$
with the relation
${x}^{m+1}=0.$
Hence the cohomology algebra
${H}^{*}(M;\mathbb{Q})$
is isomorphic to the truncated polynomial algebra
$\sim ={T}_{d,m+1}\left(x\right)=\mathbb{Q}\left[x\right]/({x}^{m+1}=0)$
and the dimension of
$M$
equals
$dm.$
The main examples are the compact rank one symmetric spaces, i.e. spheres
${S}^{d}$
of dimension
$d$
(then
$m=1$
),
$m$
dimensional complex projective spaces
$\mathbb{C}{P}^{m}$
with
$d=2,$
$m$
dimensional quaternionic projective spaces
$\mathbb{H}{P}^{m}$
with
$d=4$
and the Cayley plane
$\mathbb{C}a{P}^{2}$
with
$d=8,m=2.$
Then we obtain from [
Ra3,Thm.5.11]
:
Proposition 1
Let
$M$
be a simplyconnected and compact manifold whose rational cohomology algebra is generated by a single element of order
$d,$
i.e.
${H}^{*}(M;\mathbb{Q})={T}_{d,m+1}\left(x\right)$
endowed with a Finsler metric. Then there is a sequence
${c}_{k};k\ge 1$
of prime closed geodesics and a sequence
${m}_{k};k\ge 1$
of positive integers such that the sequence
${S}^{1}.{c}_{k}^{{m}_{k}};k\ge 1$
is a sequence of
${S}^{1}$
orbits of closed geodesics which are pairwise distinct (although in general not geometrically distinct) and whose lengths and indices satisfy the following proerties for all
$k\ge 1:$
properties:

(a)
${m}_{k}L\left({c}_{k}\right)=L\left({c}_{k}^{{m}_{k}}\right)\le L\left({c}_{k+1}^{{m}_{k+1}}\right)={m}_{k+1}L\left({c}_{k+1}\right)$

(b)
$2k(2m1)d+1\le ind\left({c}_{k}^{{m}_{k}}\right)\le 2k+d1$
Theorem 7
Let
$M$
be a simplyconnected and compact manifold whose rational cohomology algebra is generated by a single element
$x\in {H}^{d}(M;\mathbb{Q})$
, i.e.
${H}^{*}(M;\mathbb{Q})={T}_{d,m+1}\left(x\right).$
We assume that the manifold
$M$
carries a Finsler metric with reversibility
$\lambda $
whose flag curvature
$K$
is positive and satisfies
$$0<\delta \le K\le 1,$$
where
$\sqrt{\delta}>\lambda /(\lambda +1)$
if
$n$
is odd. Then the number of geometrically distinct closed geodesics of length
$\le L$
is bounded from below by
$$A(m,d,\delta ,\lambda ,L):=\frac{1}{2}\frac{\lambda +1}{\lambda}\sqrt{\delta}\left(md1\frac{\pi}{L}d\right)$$
Remark 2

(a)
If we are not interested in the length of the closed geodesics we obtain as bound:
$$A(m,d,\delta ,\lambda ,\infty )={lim}_{L\to \infty}A(m,d,\delta ,\lambda ,L)=\frac{1}{2}(md1)\sqrt{\delta}\frac{\lambda +1}{\lambda}$$

(b)
The maximal value of the bound is attained if the flag curvature
$K$
is constant (
$\delta =1$
) and the metric is reversible (
$\lambda =1$
):
$$A(m,d,1,1,\infty )=md1=n1$$

(c)
Theorem 3 is a direct consequence for
$m=1,d=n.$
Proof of Theorem 7 : We consider the sequence
${\left({S}^{1}.{c}_{k}^{{m}_{k}}\right)}_{k\ge 1}$
of pairwise distinct critical orbits of closed geodesics satisfying the properties of Proposition 1 . Hence
${c}_{k},k\ge 1$
are prime closed geodesics and
$\left({m}_{k}\right)$
is a sequence of positive integers. Fix a number
$L>0$
and let
${a}_{L}:=\#\{k\ge 1L\left({c}_{k}\right){m}_{k}\le L\}.$
By the comparison result part (b) of Lemma 1 :
$${a}_{L}\ge {b}_{L}:=\#\left\{k\ge 1ind\left({c}_{k}^{{m}_{k}}\right)<\frac{L}{\pi}\sqrt{\delta}(n1)\right\}.$$
We conclude from Proposition 1 , (b) that for
$2k<L\sqrt{\delta}(n1){\pi}^{1}(d1):ind\left({c}_{k}^{{m}_{k}}\right)<L\sqrt{\delta}(n1){\pi}^{1},$
hence
${b}_{L}$
is bounded from below by the integer part of
$L\sqrt{\delta}(n1){\pi}^{1}(d1),$
hence
${b}_{L}\ge L\sqrt{\delta}(n1){\pi}^{1}d.$
Since
$L\left({c}_{k}\right)\ge \pi (\lambda +1)/\lambda $
by Theorem 1 we obtain from Proposition 1 , (a) that
${m}_{k}<L{\pi}^{1}\lambda /(\lambda +1).$
Therefore the number of geometrically distinct closed geodesics in the set
${c}_{1},{c}_{2},\dots ,{c}_{{a}_{L}}$
is bounded from below by
$$\begin{array}{ccc}\frac{{a}_{L}}{\frac{L}{\pi}\frac{\lambda}{\lambda +1}}\ge \frac{\frac{L}{\pi}\sqrt{\delta}(n1)d}{2\frac{L}{\pi}\frac{\lambda}{\lambda +1}}\ge \frac{1}{2}\frac{\lambda +1}{\lambda}\sqrt{\delta}\left(n1\frac{\pi}{L}d\right)& & \end{array}$$  
$\square $
We call a Finsler metric bumpy, if all closed geodesics are nondegenerate. If the Finsler metric on a compact and simplyconnected manifold is bumpy and has only finitely many geometric distinct prime closed geodesics
${c}_{1},{c}_{2},\dots ,{c}_{r}$
with average indices
${\alpha}_{1},{\alpha}_{2},\dots ,{\alpha}_{r}$
then the rational cohomology ring
${H}^{*}(M;\mathbb{Q})$
is generated by a single element
$x$
of degree
$d$
with the only relation
${x}^{m+1}=0,$
i.e.
$n=dimM=md.$
The invariants
$d,m$
determine the number
$$B(d,m)=\{\begin{array}{cc}\frac{m(m+1)d}{2d(m+1)4}& ;d\text{even}\\ & \\ \frac{d+1}{2d2}& ;d\text{odd}\end{array}.$$
for which the following formula is derived in [
Ra1,Thm.3]
:
$$\begin{array}{c}B(d,m)={\sum}_{i=1}^{r}\frac{{\gamma}_{i}}{{\alpha}_{i}}\end{array}$$ 
(4)

Here
${\gamma}_{i}\in \{\pm 1/2,\pm 1\}$
is an invariant controling the parity of the sequence
$ind\left({c}_{i}^{m}\right)$
and the orientability of the negative normal bundle
$c$
and
${c}^{2}.$
Let
${c}_{1},\dots ,{c}_{s};s\le r$
be the nonhyperbolic closed geodesics. Then for even
$d$
the following estimate holds with the same argument as in [
Ra1,Thm.3.1(b)]
:
$$\begin{array}{c}{\sum}_{k=1}^{s}\left\{\left{\gamma}_{k}\right\left(\frac{md1}{{\alpha}_{k}}1\right)+2\right\}\ge \frac{1}{4}m(m+1)d\end{array}$$ 
(5)

As a consequence from this formula and Lemma 2 we obtain analogous to [
Ra1,Cor.3.4]
:
Theorem 8
Let
$F$
be a Finsler metric on a compact and simplyconnected manifold
$M$
with
${H}^{*}(M;\mathbb{Q})={T}_{d,m+1}\left(x\right)$
with reversibility
$\lambda $
and flag curvature
$K$
satisfying
$0<\delta \le K\le 1,$
where
$\sqrt{\delta}>\lambda /(\lambda +1)$
provided
$m=1$
and
$d$
is odd.

(a)
If the metric is bumpy then there are at least
$C(m,d,\delta ,\lambda ):=\leftB\right(d,m\left)\right\sqrt{\delta}\frac{\lambda +1}{\lambda}(md1)$
geometrically distinct closed geodesics.

(b)
If the metric is bumpy and there exist only finitely many geometrically distinct closed geodesics then there are at least
$m(m+1)d{\left(\frac{4\lambda}{\sqrt{\delta}(\lambda +1)}+6\right)}^{1}$
nonhyperbolic closed geodesics.
For
$d\in \{2,4,8\}$
and for a fixed value of the lower curvature bound
$\delta $
the function
$C(m,d,\delta ,\lambda )$
grows quadratically in
$m.$
Let us consider Finsler metrics on the
$m$
dimensional complex procjective space
$\mathbb{C}{P}^{m}.$
The flag curvature of the normalized FubiniStudy metric on
$\mathbb{C}{P}^{m}$
satisfies
$1/4\le K\le 1.$
For
$\delta \to 1/4$
and
$\lambda \to 1$
we obtain as maximal value
$C(m,2,1/4,1)=(m+1)(m1/2).$
It is very likely that this bound is not optimal since there are Finsler metrics of Katok type on
$\mathbb{C}{P}^{m}$
with
$m(m+1)$
geometrically distinct closed geodesics, cf. [
Zi,p.139]
.
Another application is the following result analogous to [
BTZ1,Cor.4]
:
Corollary 2
A bumpy Finsler metric on the
$m$
dimensional complex projective space
$\mathbb{C}{P}^{m}$
(
$m\ge 7$
) with reversibility
$\lambda $
and flag curvature
$0<\delta \le K\le 1;\sqrt{\delta}=\frac{2}{m+1}\frac{\lambda}{1+\lambda}$
with only finitely many geometrically distinct closed geodesiscs carries at least
$2m$
geometrically distinct closed geodesics. At least
$(m3)$
of these closed geodesics are nonhyperbolic.
5 Stability properties of closed geodesics
A closed geodesic is called hyperbolic if all all eigenvalues of the linearized Poincaré map have modulus
$\ne 1.$
Then the sequence
$ind\left({c}^{m}\right)$
is linear in
$m,$
i.e.
$ind\left({c}^{m}\right)=mind\left(c\right).$
This was observed by Bott in [
Bo]
and shows immediately part (a) of the following
Lemma 7
Let
$c$
be a closed geodesic of a Finsler manifold with average index
${\alpha}_{c}.$

(a)
If
$ind\left(c\right)\ne {\alpha}_{c}$
then
$c$
is nonhyperbolic.

(b)
If
$ind\left({c}^{2}\right)2ind\left(c\right)=n1$
then
$c$
is of ellipticparabolic type, the linearized Poincaré map splits into
$2\times 2$
blocks of the form
$$\left(\begin{array}{cc}1& 0\\ 1& 1\end{array}\right)\text{and / or}\left(\begin{array}{cc}cos\phi & sin\phi \\ sin\phi & cos\phi \end{array}\right)0\le \phi <\pi ,$$
with respect to a symplectic basis
$({X}_{1},{Y}_{1},{X}_{2},{Y}_{2},\dots ,{X}_{n1},{Y}_{n1})$
satisfying
$\omega ({X}_{i},{Y}_{i})={\delta}_{ij};\omega ({X}_{i},{X}_{j})=\omega ({Y}_{i},{Y}_{j})=0$
for the symplectic form
$\omega .$
Part (b) is shown in [
BTZ1,Lemma3.1]
, since
$ind\left({c}^{2}\right)={I}_{c}\left(1\right)+{I}_{c}(1)$
and
$ind\left(c\right)={I}_{c}\left(1\right),$
cf. the Proof of Lemma 5 Proof of Theorem 4 : By considering the universal covering we can assume
$M$
to be simplyconnected. Since
${\pi}_{l}\left(M\right)\ne 0$
one concludes that
${\pi}_{l1}(\Lambda M)\ne 0,$
hence there is a closed geodesic
$c$
with
$ind\left(c\right)\le l1.$
From Theorem 1 we conclude
$L\left(c\right)\ge \pi (1+{\lambda}^{1}),$
hence Lemma 1 (b) implies that
${\alpha}_{c}>l1.$
Therefore
$c$
is nonhyperbolic by Lemma 7 , part (a).
$\square $
Remark 3
On a compact and notsimplyconnected Riemannian manifold of nonnegative Ricci curvature there is a nonhyperbolic closed geodesic. The proof of this statement (cf. [
BTZ1,Thm.B(a)]
) carries over to the Finsler case without changes, here one does not need Klingenberg's injectivity radius estimate.
Proof of Theorem 5 :
(a) Since
${b}_{n1}(\Lambda {S}^{n},{\Lambda}^{0}{S}^{n})=1$
there is a closed geodesic
$c$
with
$ind\left(c\right)\le n1.$
The length estimate Theorem 1 implies for the second cover
$L\left({c}^{2}\right)\ge 2\pi (1+{\lambda}^{1})>3\pi /\sqrt{\delta}.$
Hence by Lemma 1 we obtain
$ind\left({c}^{2}\right)\ge 3(n1)$
resp.
$ind\left({c}^{2}\right)2ind\left(c\right)\ge n1.$
We conclude from Lemma 7 (b) that
$c$
is of ellipticparabolic type.
(b) Let
$c$
be a shortest closed geodesic which is not homotopically trivial. Then
$c$
is a local minimum for the energy functional, hence
$ind\left(c\right)=0.$
Since
$c$
defines also a closed geodesic on the universal covering we conclude from Theorem 1 that
$L\left({c}^{2}\right)\ge \pi (1+{\lambda}^{1})>\pi /\sqrt{\delta}.$
Hence by Lemma 1 (b) we conclude
$ind\left({c}^{2}\right)\ge (n1)$
which shows that
$c$
is of ellipticparabolic type by Lemma 7 (b).
$\square $
Now we come to a different setting in which one can show the existence of two geometrically distinct closed geodesics on a manifold with nonreversible Finsler metric.
An isometry
$A:M\to M$
of finite order of a compact Finsler manifold
$(M,F)$
has small displacement, if for all points
$p\in M$
the image point
$A\left(p\right)$
does not lie in the cut locus. Let
$\theta :M\times M\to \mathbb{R}$
be the distance function, i.e.
$\theta (x,y)$
is the minimal length of a smooth curve
$c:[0,1]\to M$
from
$x=c\left(0\right)$
to
$y=c\left(1\right).$
Note that this distance function in general is not symmetric since we consider nonreversible Finsler metrics.
Then we define the function
${f}_{A}:M\to \mathbb{R}$
with
${f}_{A}\left(x\right)={\theta}^{2}(x,Ax)$
which is smooth outside the fixed point set
${M}_{A}$
of
$A.$
Then we obtain as in [
BTZ1,p.239]
: The point
$p$
is a critical point of
${f}_{A}$
if and only if the unique minimal geodesic
$\gamma :[0,1]\to M$
from
$p=\gamma \left(0\right)$
to
$A\left(p\right)=\gamma \left(1\right)$
is invariant under
$A,$
i.e.
${A}_{*}\left({\gamma}^{\prime}\left(0\right)\right)={\gamma}^{\prime}\left(1\right).$
It also follows that
$\gamma \left(t\right)$
is a critical point of
${f}_{A}$
for all
$t\in [0,1].$
A critical point
$p\notin {M}_{A}$
determines a closed geodesic, since
$A$
is of finite order. Analogous to [
BTZ1,Thm.3.10]
we obtain
Proposition 2
Let
$A$
be an isometry of finite order and small displacement on a compact Finsler manifold. A local maximum of
${f}_{A}$
determines a closed geodesic of ellipticparabolic type if
$2\theta (p,A(p\left)\right)<\theta (p,Cut(p\left)\right).$
Proof of Theorem 6 :
Let
${\phi}_{t},t\in \mathbb{R}$
define the isometric
${S}^{1}$
action, i.e.
${\phi}_{t}:M\to M$
is a oneparameter group of isometries with
${\phi}_{1}$
equals the identity. Then for a sufficiently large integer
$m$
the isometries
$A={\phi}_{1/m}$
and
${A}^{1}={\phi}_{1/m}$
have small displacement and are of finite order. Then we conclude from Proposition 2 that there are two points
${p}_{\pm}$
such that
${p}_{\pm}$
is a maximum of
${f}_{A}$
resp.
${f}_{{A}^{1}}.$
Then the closed geodesics
${c}_{\pm}$
with
${c}_{\pm}\left(0\right)={p}_{\pm}$
and
${c}_{\pm}(1/m)={A}^{\pm 1}\left({p}_{\pm}\right)$
are of ellipticparabolic type and satisfy
$c\left(t\right)={\phi}_{t}\left({p}_{\pm}\right).$
If
${c}_{},{c}_{+}$
are geometrically equivalent, we can assume without loss of generality that
$p={p}_{}={p}_{+}.$
Hence
${c}_{\pm}\left(t\right)={\phi}_{\pm t}\left(p\right)$
which shows that
${c}_{+}^{\prime}\left(0\right)={c}_{}^{\prime}\left(0\right).$
i.e. the closed geodesics
${c}_{,}{c}_{+}$
are geometrically distinct.
References

W.Ballmann, G.Thorbergsson & W.Ziller: Closed geodesics on positively curved manifolds. Ann. of Math.(2) 116 (1982) 213–247

W.Ballmann, G.Thorbergsson & W.Ziller: Existence of closed geodesics on positively curved manifolds. J. Differ. Geom. 18 (1983) 221–252

W.Ballmann, G.Thorbergsson & W.Ziller: Existence of short closed geodesics and their stability properties. p. 53 – 63, In: Minimal Submanifolds,ed. E.Bombieri, Annals of Mathematics Studies, Princeton Univ. Press, Princeton 1983

V.Bangert: Geodätische Linien auf Riemannschen Mannigfaltigkeiten. Jber.d.Dt.Math.Verein. 87 (1985) 39–66

R.Bott: On the iteration of closed geodesics and the Sturm intersection theory. Comm.Pure Appl.Math. 9 (1956) 171–206

W.Klingenberg: Lectures on closed geodesics. Grundlehren der math.Wiss. 230 Springer–Verlag Berlin Heidelberg New York 1978

W.Klingenberg: Riemannian geometry. de Gruyter Studies Math. 1, 2nd rev. ed., de Gruyter Berlin New York 1995

Y.Long: Index Theory for Symplectic Pahts with Applications. Progress in Math. 207, Birkhäuser Verlag Basel 2002

Y.Long: Multiple Closed Geodesics on Finsler
$2$
–spheres and a Conjecture of D.V.Anosov. Talk at the International Symposium on Finsler Geometry, Nankai University, August 2004

H.B.Rademacher: On the average indices of closed geodesics. J.Differential Geom. 29 (1989) 65–83

H.B.Rademacher:Morse–Theorie und geschlossene Geodätische. Bonner Math. Schriften 229 (1992)

H.B.Rademacher: The Fadell–Rabinowitz index and closed geodesics. J.London Math.Soc.50(1994) 609–624

H.B.Rademacher: A Sphere Theorem for nonreversible Finsler metrics. Math. Annalen 328 (2004) 373 – 387

H.B.Rademacher: Nonreversible Finsler metrics of Positive Curvature. In: A Sampler of RiemannFinsler Geometry. (D. Bao, R. Bryant, S.S.Chern, Z.Shen, eds.), Cambridge Univ. Press, MSRI Series, 2004

G.Thorbergsson: Non–hyperbolic closed geodesics. Math.Scand. 44 (1979) 135–148

W.Ziller: Geometry of the Katok examples. Ergod.Th.& Dyn.Syst. 3 (1982) 135–157
Universitat Leipzig, Mathematisches Institut Augustusplatz 10/11, D04109 Leipzig rademacher@math.unileipzig.de www.math.unileipzig.de/~rademacher