5

4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


Sticks and clubs
Sakae Fuchino, Saharon Shelah and Lajos Soukup
April 3, 1997
Abstract
We study combinatorial principles known as stick and club. Several vari-
ants of these principles and cardinal invariants connected to them are also
considered. We introduce a new kind of side-by-side product of partial or-
derings which we call pseudo-product. Using such products, we give sev-
eral generic extensions where some of these principles hold together with
CH and Martins Axiom for countable p.o.-sets. An iterative version of the
pseudo-product is used under an inaccessible cardinal to show the consistency
of the club principle for every stationary subset of limits of
1
together with
CH and Martins Axiom for countable p.o.-sets.
Keywords: stick principle, club principle, weak Martins axiom, preservation theorem.
1991 Mathematics Subject classication: 03 E 35, 03 E 05.
1 Beating with sticks and clubs
In this paper, we study combinatorial principles known as stick and club, and
their diverse variants which are all weakenings of 3. Hence some of the conse-
quences of 3 still hold under these principles. On the other hand, they are weak
enough to be consistent with the negation of the continuum hypothesis or even
with a weak version of Martins axiom in addition. See e.g. [2], [4], [10] for applica-
tions of these principles. We shall begin with introducing the principles and some
cardinal numbers connected to them.
( [

) (read stick) is the following principle introduced in S. Broverman, J.

Ginsburg, K. Kunen and F. Tall [2]:
( [

): There exists a sequence (x

)
<
1
of countable subsets of
1
such that for
any y [
1
]

1
there exists <
1
such that x

y.
Of course the sequence (x

)
<
1
above is a blu. What is essential here is that
there exists an X [
1
]

0
of cardinality
1
such that for any y [
1
]

1
there is
1
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


an x X with x y. The formulation above is chosen here merely to make the
connection to the principle () introduced later, more apparent.
Note that ( [

) follows from CH.

The principle ( [

) suggests the following cardinal number:

[

= min [ X [ : X [
1
]

0
, y [
1
]

1
x X x y .
We have
1
[

0
and ( [

) holds if and only if [

=
1
. We also consider the
following variants of [

:
[

= min :
1
, there is an X []

0
such that [ X [ = and y []

1
x X x y ;
[

= min :
1
, there is an X []

0
such that [ X [ = and y []

x X x y ;
[

= min [ X [ : X []

0
such that y []

1
x X x y .
We have
1
[

0
and [

0
. ( [

) holds if and only if

[

= [

= [

=
1
. Let us call X as in the denition of [

( [

, [

and [

respectively) a [

-set ( [

-set, [

-set and [

-set respectively).
Lemma 1.1
a) [

.
b) If [

<

1
then [

= [

. In particular, we have then [

.
c) If

then [

.
d) [

|
[

.
Proof a): Let X []

0
be a [

-set of cardinality [

. Then X
0
= X [
1
]

0
is
a [

-set of cardinality [

.
b): By a), it is enough to show [

. We show inductively that, for every

uncountable [

,
()

there exists an X

[]

0
such that [ X

[ [

and
y []

1
x X

(x y).
For =
1
this is clear.
Assume that we have shown ()

for all < . If is a successor then by

induction hypothesis, we can nd X

[]

0
for all < such that [ X

[ [

and y []

1
x X

x y. Let X

=

<
X

. Then X

has the desired

2
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


property: [ X

[ [

is clear. If y []

1
, there is some < such that y []

1
.
Hence there is an x X

X such that x y.
Suppose now that is a limit. By assumption, we have cof() = . Let (
n
)
n
be an increasing sequence of cardinals below such that =

n

n
. For each
n, let X
n
[
n
]

0
be as in ()
n
and let X

=

n
X
n
. Then X

is as desired:
clearly [ X

[ [

. If y []

1
there is an n such that y
n
is uncountable.
Hence there exists an x X
n
X

such that x y
n
y.
In particular we have shown that ()
|
holds and hence [

.
c): Similarly to a).
d): By a) and c), we have [

= [

1
[

|
[

= [

. (Lemma 1.1)
The question, whether [

< [

is consistent, turned out to be a very delicate one:

the problem is connected with some natural weakenings of GCH whose status (i.e.
whether they are theorems in ZFC) is still open. One of them implies that [

= [

(this is essentially stated in [14, 1.2, 1.2A] in the light of [13, 6.1 [D]]; for more see
[15]) while the negation of the other implies that the inequality is consistent. In this
paper, we shall treat the latter consistency proof (Proposition 3.4). In contrast, the
consistency of the inequality [

< [

can be shown without any such additional

set-theoretic assumptions (Proposition 3.5).
The principle () (club), a strengthening of ( [

), was rst formulated in

Ostaszewski [10]. Let Lim(
1
) = <
1
: is a limit . For a stationary
E Lim(
1
),
(E): There exists a sequence (x

)
E
of countable subsets of
1
such that for
every E, x

is a conal subset of with otp(x

) = and for every

y [
1
]

1
there is E such that x

y.
Let us call (x

)
E
as above a (E)-sequence. For E = Lim(
1
) we shall simply
write () in place of (Lim(
1
)). Clearly ( [

) follows from (). Unlike ( [

), ()
does not follow from CH since () + CH is known to be equivalent to 3 (K. Devlin,
see [10]). This equivalence holds also in the version argumented with a stationary
E Lim(
1
).
Fact 1.2 For any stationary E Lim(
1
), (E) + CH is equivalent to 3(E).
Proof The proof in [10] argumented with E works. (Fact 1.2)
S. Shelah [11] proved the consistency of CH + () in a model obtained from
a model of GCH by making the size of (
1
) to be
3
by countable conditions
and then collapsing
1
to be countable. Soon after that, in an unpublished note,
J. Baumgartner gave a model of CH + where collapsing of cardinals is not
3
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


involved: his model was obtained from a model of V = L by adding many Sacks
reals by side by side product. I. Juh asz then proved in an unpublished note that
CH + MA(countable) + () is consistent. Here MA(countable) stands for
Martins axiom restricted to countable partial orderings. Later P. Komj ath [7]
cited a remark by Baumgartner that Shelahs model mentioned above also satises
CH + MA(countable) + (). In Section 3, we shall give yet another model of CH
+ MA(countable) + () in which collapsing of cardinals is not involved (Theorem
3.8). In section 5, we construct a model of CH + MA(countable) + (E) for
every stationary E Lim(
1
) starting from a model of ZFC with an inaccessible
cardinal (Theorem 5.6).
These results are rather optimal in the sense that a slight strengthening of
MA(countable) implies the negation of (). Let MA(Cohen) denote Martins axiom
restricted to the partial orderings of the form Fn(, 2) for some where, as in [8],
Fn(, 2) is the p.o.-set for adding Cohen reals, i.e. the set of functions from some
nite subset of to 2 ordered by reverse inclusion.
Fact 1.3 MA for the partial ordering Fn(
1
, 2) implies [

= [

= 2

0
. Further, if
MA(Cohen) holds, then we have also [

= 2

0
.
Proof Both equations can be proved similarly. For the rst equation, it is enough
to show [

= 2

0
by Lemma 1.1. Suppose that X [
1
]

0
is of cardinality less
than 2

0
. We show that X is not a [

-set. Let P = Fn(

1
, 2). Then for each x X
the set
D
x
= q Fn(
1
, 2) : dom(q) x q() = 0
is dense in P. For each <
1
,
E

= q Fn(
1
, 2) : > ( dom(q) q(b) = 1)
is also a dense subset of P. Let T = D
x
: x X E

: <
1
and G be a
T-generic lter over P. Then the uncountable set
Y = <
1
: q() = 1 for some q G
contains no x X as a subset. (Fact 1.3)
We shall see in Proposition 3.5 that MA for the partial ordering Fn(
1
, 2) is
not enough for the last assertion in Fact 1.3.
(E) is equivalent to the following seemingly much stronger statement. Let
E Lim(
1
) be a stationary set.

(E): There exists a sequence (x

)
E
of countable subsets of
1
such that
for every E, x

is a conal subset of with otp(x

) = and for
every X [
1
]

1
, E : x

X is stationary.
4
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


Fact 1.4 For any stationary E Lim(
1
), (E) and

(E) are equivalent.

Proof Like Fact 1.2, an easy modication of the corresponding proof in [10] will
work. Nevertheless we give here a proof for convenience of the reader.
Clearly it is enough to show (E)

(E). Suppose that (x

)
E
is a (E)-
sequence. We claim that (x

)
E
is then also a

(E)-sequence. Otherwise there

would be a Y [
1
]

1
and a club C Lim(
1
) such that x

, Y for every
C E. By thinning out C if necessary, we may assume that Y is conal
in for each C. For C, denoting by
+
the next element to in C, let
y

[,
+
)Y be a conal subset in
+
with otp(y

) = . Now let Y

C
y

.
Then Y

[
1
]

1
and Y

Y . We show that E : x

= which is a
contradiction: if E C then x

, Y

follows from Y

Y . If E C then
there is C such that < <
+
. By the choice of y

, Y

is not conal in
. Hence again x

, Y

. (Fact 1.4)
Now, let us consider the following variants of the ()-principle:
(
w
): There exists a sequence (x

)
Lim(
1
)
of countable subsets of
1
such that
for every Lim(
1
), x

is conal subset of , otp(x

) = and for
every y [
1
]

1
, there is <
1
such that x

y is nite.
(
w
2): There exists a sequence (x

)
Lim(
1
)
of countable subsets of
1
such
that for every Lim(
1
), x

is conal subset of , otp(x

) = and
for every y [
1
]

1
<
1
: x

y is nite <
1
: x

y is nite
is stationary in
1
.
Clearly () implies (
w
). Similarly to Fact 1.4, we can prove the equivalence of
(
w
) with (

w
) which is obtained from (
w
) by replacing there is an <
1
. . .
with there are stationary may <
1
. . . . Hence (
w
) implies (
w
2). It is also
easy to see that (
w
) implies ( [

): if (x

)
Lim(
1
)
is a sequence as in the denition
of (
w
), then x

u : Lim(
1
), u [
1
]
<
0
is a [

-set of cardinality
1
.
Dzamonja and Shelah [3] gave a model of CH + (
w
) + (). By the remark
above this model also shows the consistency of non-equivalence of ( [

) and ()
under CH. In this paper we prove that (
w
2) is strictly weaker than (
w
) by
showing the consistency of ( [

) + (
w
2) (Corollary 3.12). The partial ordering
used in Corollary 3.12 does not force MA(countable) hence the following problem
remains open:
Problem 1.5 Is MA(countable) + ( [

) + (
w
2) consistent?
5
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


2 Pseudo product of partial orderings
In this section, we introduce a new kind of side-by-side product of p.o.s which will
be used in the next section to prove various consistency results. Let X be any set
and (P
i
)
iX
be a family of partial orderings. For p
iX
P
i
the support of p is
dened by supp(p) = i X : p(i) ,= 1
P
i
. For a cardinal , let

,iX
P
i
be the
set
p
iX
P
i
: [ supp(p) [ <
with the partial ordering
p q p(i) q(i) for all i X and
i X : p(i)
<
=
q(i)
<
=
1
P
i
is nite .
For =
0
this is just a nite support product. We are mainly interested in the
case where =
1
. In this case we shall drop the subscript
1
and write simply

iX
P
i
. Further, if P
i
= P for some partial ordering P for every x X, we shall
write

,X
P (or even

X
P when =
1
) to denote this partial ordering.
For p, q

,iX
P
i
the relation p q can be represented as a combination
of the two other distinct relations which we shall call horizontal and vertical, and
denote by
h
and
v
respectively:
p
h
q supp(p) supp(q) and p[` supp(q) q;
p
v
q supp(p) = supp(q), p(i) q(i) for all i X and
i X : p(i)
<
=
q(i)
<
=
1
P
i
is nite .
For p

,iX
P
i
and Y X let pY denote the element of

,iX
P
i
dened by
pY (i) = 1
P
i
for every i X Y and pY (i) = p(i) for i Y .
The following is immediate from denition:
Lemma 2.1 For p, q

,iX
P
i
, the following are equivalent:
a) p q;
b) There is an r

,iX
P
i
such that p
h
r
v
q;
c) There is an s

,iX
P
i
such that p
v
s
h
q.
Proof b) a) and c) a) are clear. For a) b), let r = psupp q; for a) c),
s = q[` supp(q) p[`(X supp(q)). (Lemma 2.1)
Lemma 2.2
1) If P
i
has the property K for all i X then P =

iX
P
i
preserves
1
.
2) Suppose that . If P
i
has the strong -cc (i.e. for every C [P
i
]

there
is pairwise compatible D [C]

), then P =

,iX
P
i
preserves .
6
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


Proof This proof is a prototype of the arguments we are going to apply repeatedly.
1) and 2) can be proved similarly. For 1), assume that there would be p P and
a P-name

f such that
*) p |
P


f : (
1
)
V
and

f is 1-1 .
Then, let (p

)
<
1
and (q

)
<
1
be sequences of elements of P such that
a) p
0
p and (p

)
<
1
is a descending sequence with respect to
h
;
b) q

v
p

and q

decides

f() for all <
1
;
c) p

[`S

= q

[`S

for every <

1
where
S

= supp(q

) (supp(p)

<
supp(q

)).
For <
1
let d

=

<
supp(q

). Then (d

)
<
1
is a continuously increasing
sequence in [X]
<
1
. Let u

= supp(q

) : q

() ,= p

() for <
1
.
By b), u

is nite and by c) we have u

. Hence by Fodors lemma, there

exists an uncountable (actually even stationary) Y
1
such that u

= u

for all
Y , for some xed u

[X]
<
0
. Since
iu
P
i
has the property K, there exists
an uncountable Y

Y such that q

[`u

: Y

is pairwise compatible. It
follows that q

, Y

are pairwise compatible. For each Y

there exists an
n

such that q

|
P
n

=

f() by b). By *), n

, Y

must be pairwise
distinct. But this is impossible as Y

is uncountable.
For 2), essentially the same proof works with sequences of elements of P of
length , using the -system lemma argument in place of Fodors lemma.
(Lemma 2.2)
Lemma 2.3 If [ P
i
[ 2
<
for all i X, then

,iX
P
i
has the (2
<
)
+
-cc.
Proof By the usual -system lemma argument. (Lemma 2.3)
Corollary 2.4 a) Under CH, if P
i
satises the property K and [ P
i
[
1
for
every i X, then P =

iX
P
i
preserves
1
and has the
2
-cc. In particular P
preserves every cardinals.
b) Suppose that 2
<
= . If P
i
satises the strong -cc for every
1

and [ P
i
[ then

,iX
P
i
preserves every cardinalities and has the
+
-cc.
In particular

,iX
P
i
preserves every cardinals.
Proof By Lemmas 2.2, 2.3. (Corollary 2.4)
Lemma 2.5 For any Y X and x X Y , we have

,iX
P
i

=

,iY
P
i
P
x

,iX\(Y {x})
P
i
.
7
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


Proof The mapping from

,iX
P
i
to

,iY
P
i
P
x

,iX\(Y {x})
P
i
dened by
p (p[`Y, p(x), p[`(X (Y x)))
is an isomorphism. (Lemma 2.5)
In the following we mainly use the partial orderings of the form Fn(, 2) for some
as P
i
in

,iX
P
i
. Note that Fn(, 2) has the property K and strong -cc in the
sense above for every regular .
For a pseudo product of the form

iX
Fn(
i
, 2), Lemma 2.2 can be still im-
proved:
Theorem 2.6 (T. Miyamoto) For any set X, and sequence (
i
)
iX
, the partial
ordering P =

iX
Fn(
i
, 2) satises the Axiom A.
Proof The sequence of partial orderings (
n
)
n
dened by: p
0
q p q
and p
n
q p
h
q for every n > 0 witnesses the Axiom A of P. We omit
here the details of the proof since this assertion is never used in the following. The
idea of the proof needed here is to be found in the proof of Lemmas 2.7 and 5.2.
(Theorem 2.6)
Lemma 2.7 Suppose that [ P
i
[ for every i X and P =

+
,iX
P
i
. Then
1) If x is a P-name with |
P
x V , then for any p P there is q P such
that q
h
p and
() for any r q, if r decides x then rsupp(q) already decides x.
2) Let G be P-generic. If u V [G] is a subset of V of cardinality <
+
, then
there is a ground model set X

X of cardinality (in the sense of V ) such

that u V [G (

+
,iX

P
i
)].
Proof 1): Let : ; (
1
(),
2
()) be a surjection such that

1
() for every < . Let (p

)
<
, (p

)
<
and (r
,
)
<,<
be sequences of
elements of P dened inductively by:
a) p
0
= p; (p

)
<
is a descending sequence with respect to
h
;
b) for a limit < , p

is such that supp(p

) =

<
supp(p

) and, for i
supp(p

), p

(i) = p

(i) for some < such that i supp(p

);
c) (r
,
)
<
is an enumeration of r P : r
v
p

;
d) let r = r

1
(),
2
()
and
p

= r[` supp(r) p

[`(X supp(r)).
8
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


If there is s
h
p

such that s decides x, then let

p
+1
= p

[` supp(p

) s[`(X supp(p

)).
Otherwise let p
+1
= p

.
Let q

,iX
P
i
be dened by supp(q) =

<
supp(p

) and, for i supp(q),

q(i) = p

(i) for some < such that i supp(P

). We show that this q is as

desired: suppose that r q decides x. Then there is some < such that
rsupp(q) = p

[` supp(p

) q[`(X supp(p

)).
By d), it follows that rsupp(q) rsupp(p
+1
) decides x.
2): Let u be a P-name for u and let x

, < be P-names such that |

P
x

V
for every < and |
P
u = x

: < . By 1), for each p P, we can

build a sequence (p

)
<
of elements of P decreasing with respect to
h
such that
p
0

h
p and
()

for any r p

, if r decides x

, then rsupp(p

) already decides x

.
Let q P be dened by supp(q) =

<
supp(p

) and, for i supp(q), q(i) = p

(i)
for some < such that i supp(p

). Then q satises:
() for any r q, if r decides x

for some < , then rsupp(q) already

decides x

.
The argument above shows that qs with the property () are dense in P. Hence, by
genericity, there is such q G. Clearly, G

,isupp(q)
P
i
contains every information
needed to construct u. (Lemma 2.7)
3 Consistency results
Proposition 3.1 (CH) For any innite cardinal , let P =

Fn(
1
, 2). Then
|
P
[

= .
Proof
Claim 3.1.1 |
P
[

If =
1
this is clear. So assume that
2
. For < , let

f

be the
P-name of the generic function from
1
to 2 added by the -th copy of Fn(
1
, 2)
in P. Let G be a P-generic lter over V . In V [G] let X [
1
]

0
be such that
[ X [ < . Then by
2
-cc of P there exists < such that X V [G

] for
G

= G

\{}
Fn(
1
, 2). Since (

f

)[G] is Fn(
1
, 2)-generic over V [G

] by Lemma
2.5, we have x , ((

f

)[G])
1
0 for every x X.

(Claim 3.1.1)
9
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


Claim 3.1.2 |
P
[

For u []
<
0
, let

T
u
be a P-name such that
|
P


T
u
= ([
1
]

0
)
V [(

f

)
u
]

where

f

is as in the proof of the previous claim. Let

T be a P-name such that
|
P


T =

T
u
: u []
<
0
.
For each u []
<
0
, (

f

[G])
u
corresponds to a generic lter over

u
Fn(
1
, 2)
Fn(
1
, 2). Hence, by CH, we have |
P
[

T
u
[ =
1
. It follows that |
P
[

T [ =
. Thus it is enough to show that |
P


T is a [

-set .
Let p P and

A be a P-name such that p |
P


A [
1
]

1
. We show that
there is an r p such that r |
P
x

T x

A.
Now we proceed as in the proof of Lemma 2.2. Let (p

)
<
1
, (q

)
<
1
be
sequences of elements of P and (

)
<
1
be a strictly increasing sequence of ordinals
<
1
such that
a) p
0
p and (p

)
<
1
is a descending sequence with respect to
h
;
b) q

v
p

and q

|
P

A for all <
1
;
c) p

[`S

= q

[`S

for every <

1
where
S

= supp(q

) (supp(p)

<
supp(q

)).
For <
1
let u

= supp(q

) : q

() ,= p

() . As in the proof of Lemma

2.2, there exists u

[]
<
0
such that S =
1
: u

= u

is stationary.
Now (q

[`u)
S
is an innite sequence of elements of P
u
=
u
Fn(
1
, 2). Since
P
u
satises the ccc, there exists an S and <
1
such that q

[`u

|
P
u

S : p

[`u

G is innite . Let r = q

[`(supp(p

) supp(p

)). Let

b be a
P-name such that
r |
P


b = S : q

[`u

p[`u

: p

G .
Let x be a P-name such that r |
P
x =

:

b . Then r |
P
[ x [ =
0
.
Since

b can be computed in V [(

f

[G])
u
] we have r |
P
x

T
u
. It is also clear
by denition of x that r |
P
x

A.

(Claim 3.1.2)
(Proposition 3.1)
Proposition 3.1 shows that [

can be practically every thing. In particular we

obtain:
Corollary 3.2 The assertion cof( [

) = is consistent with ZFC.

10
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


Actually, Fn(, 2) forces almost the same situation:
Lemma 3.3 Suppose that is a cardinal such that

0
for every < . Then,
for P = Fn(, 2), we have |
P
[

= .
Proof |
P
[

can be proved similarly to Claim 3.1.1. For |

P
[

,
let G be a P-generic lter and let G

= G Fn(, 2) for < . In V [G], let

X =

V [G

] [
1
]

0
: < . Then [ X [ = (here we need SCH in general).
We show that X is a [

-set. For this, it is enough to show the following:

Claim 3.3.1 In V [G], if y [
1
]

1
, then there is

< and innite y

V [G

]
such that y

y.

In V , let y be a P-name of y which is nice in the sense of [8]. For < , let
y

= y

: <
1
Fn(, 2). Then |
P
y =

<
y

. Hence |
P
<
y

is innite . It follows that there is some

< such that y

= y

[G] is
innite. Since y

is an Fn(

, 2)-name, y

[G] V [G

]. Thus these

and y

are
as desired.

(Claim 3.3.1)
(Lemma 3.3)
Proposition 3.4 (CH) Suppose that
()
,
There is a sequence (A
i
)
i<
of elements of []

1
such that [ A
i
A
j
[ <

0
for every i, j < , i ,= j
holds for some > 2

0
. Then there exists a partial ordering P such that
a) P preserves
1
and and has the
2
-cc;
b) |
P
[

= and
c) |
P
[

.
In particular, if ()
,
is consistent with ZFC for some > 2

0
, then so is
[

< [

.
Remark. By [12, 6], ()

and ()
,
for some < are equivalent, where
()

there are nite a

i
Reg
2
for i <
1
such that, for any A [
1
]

0
,
max pcf(
iA
a
i
) .
For more see [15].
Proof Let P be as in Proposition 3.1. We claim that P is as desired: a) follows
from Corollary 2.4 and b) from Proposition 3.1. For d), if X []

0
is a [

-set
then for each i < there is an x
i
X such that x
i
A
i
. Since A
i
, i < are
almost disjoint x
i
, i < must be pairwise distinct.
The last assertion follows from Lemma 1.1, d). (Proposition 3.4)
Now we show the consistency of the inequality [

< [

:
11
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


Proposition 3.5 Assume 2

1
=
2
. Then for any cardinal
2
there exists a
partial ordering P such that
a) P satises the
3
-cc;
b) P preserves
1
and
2
;
c) if

0
= in addition, then |
P
MA(Fn(
1
, 2)) ;
d) |
P
[

= and
e) |
P
[

=
2
.
Proof Without loss of generality let be regular and let P =

2
,
Fn(
1
, 2). Then
a) and b) follow from Corollary 2.4. For c), note that |
P
2

0
= under

0
= .
Hence, by Lemma 2.7 and Lemma 2.5, we see easily that |
P
MA(Fn(
1
, 2)) .
An argument similar to the proof of of Proposition 3.1 shows that |
P
[

= .
For e), we prove rst the following:
Claim 3.5.1 Let X = [
2
]

0
. Then we have |
P
X is a [

-set . In particular
|
P
[

2
.

Suppose that, for some p P and a P-name y we have p |

P
y [
2
]

2
.
Let

f be a P name such that p |
P


f :
2
y and

f is 1-1 . Let (p

)
<
2
and
(q

)
<
2
be sequences of elements of P such that
f) p
0
p and (p

)
<
2
is a descending sequence with respect to
h
;
g) q

v
p

and q

decides

f() for all <
2
;
h) p

[`S

= q

[`S

for every <

2
where
S

= supp(q

) (supp(p)

<
supp(q

)).
For <
2
, let

2
be such that q

|
P


f() =

. Let u

=
supp(q

) : q

() ,= p

() for <
2
. Just like in the proof of Lemma 2.4, we
can nd u

[]
<
0
such that S = <
2
: u

= u

is stationary in
2
. Since
[ Fn(
1
, 2) [ =
1
, there exists T S of cardinality
2
such that q

[`u

, T are
all the same. Let
n
, n be elements of T and let q =

n
q
n
. Then q p
and q |
P

n
: n y .

(Claim 3.5.1)
Now by d), we have |
P
[

>
1
. Hence, by the claim above, it follows that
|
P
[

=
2
. (Proposition 3.5)
Modifying the proofs of Propositions 3.1 and 3.5 slightly, we can also blow up
the continuum while setting [

strictly between
1
and 2

0
. For example:
Proposition 3.6 Assume CH and 2

1
=
2
. Then for any cardinals , such
that
2
and

1
= , there exists a partial ordering P such that
12
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


a) P satises the
3
-cc;
b) P preserves
1
and
2
;
c) |
P
MA(countable) ;
d) |
P
[

= ;
e) |
P
[

=
2
and
f) |
P
2

0
= .
Proof For i < let
P
i
=

Fn(
1
, 2), if i < ,
Fn(, 2), otherwise.
Then P =

2
,i<
P
i
is as desired. e) can be proved by almost the same proof as
that of Claim 3.5.1. a), b), c) can be shown just as in Proposition 3.5. Since P
adds (at least) many Cohen reals over V and [ P [ = , f) follows from a). d) is
proved similarly to Claims 3.1.1 and 3.1.2. For |
P
[

we need the following

modication of Claim 3.1.2: let

T be dened as in the proof of Claim 3.1.2. As
there, we can show easily that |
P
[

T [ = . To show that |
P


T is a [

-set ,
let p P and

A be a P-name such that p |
P


A [
1
]

1
. Now let (p

)
<
1
,
(q

)
<
1
, (

)
<
1
, u

[]
<
0
and S be just as in the proof of Claim 3.1.2. Let
v

= u

. Since P
v
=
iv
P
i
is countable, we may assume without loss of
generality that q

[`v

, S are all the same. Now we can proceed just like in the
proof of Claim 3.1.2 with u

replaced by u

.
The following Lemmas 3.7 and 3.9 show that, in spite of typographical similarity,

Fn(
1
, 2) and

Fn(, 2) are quite dierent forcing notions: while the rst one
destroys () or even ( [

) by Lemma 3.1, the second one not only preserves a

()-sequence in the ground model but also creates such a sequence generically.
Lemma 3.7 Let S = (x

)
E
be a (E)-sequence for a stationary E Lim(
1
).
Let P =

Fn(, 2) for arbitrary . Then we have |

P
S is a (E)-sequence .
Proof Let p P and

A be a P-name such that p |
P


A [
1
]

1
. We show
that there is q p and E such that q |
P
x

A. Let

f be a P-name such
that p |
P


f :
1


A and

f is 1-1 . Let (p

)
<
1
and (q

)
<
1
be sequence of
elements of P satisfying the conditions a) c) in the proof of Lemma 2.2. Also, let
u

, <
1
be as in the proof of Lemma 2.2. As there, we can nd an uncountable
Y
1
and u

[]
<
0
such that u

= u

for all Y . Since

u
Fn(, 2) is
countable we may assume that q

[`u

are all the same for Y . Now for each

Y let

be such that q

|
P


f() =

and let Z =

: Y . Since
q

, Y are pairwise compatible,

, Y are pairwise distinct and so Z is

uncountable. Note that Z is a ground model set. Hence there exists E such
13
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


that x

Z. Let q =

Y
q

. Then q p. Since sup

: < and
|
P

: < is an initial segment of Z , we have q |

P
Z

A. Hence
q |
P
x

A. (Lemma 3.7)
Theorem 3.8 CH+ MA(countable) + there exists a constructible -sequence
is consistent.
Proof We can obtain a model of the statement by starting from a model of
V = L and force with P =

Fn(, 2) for a regular . By Corollary 2.4, every

cardinal of V is preserved in V [G]. Since P adds many Cohen reals over V
while [ P [ = and P has the
2
-cc, we have V [G] [= 2

0
= . By Lemma
2.5, V [G] [= MA(countable) . By Lemma 3.7, the 3-sequence in V remains a
-sequence in V [G]. (Theorem 3.8)
In fact, we do not need a -sequence in the ground model to get () in the
generic extension by

Fn(, 2) :
Lemma 3.9 Let be uncountable and P =

Fn(, 2). Then for any stationary

E Lim(
1
) we have |
P
(E) holds .
Proof For E let
f

: [, +)
be a bijection and let
S

= x : x is a conal subset of , otp(x) = .

For each x S

let p
x
P be dened by
p
x
= ( +n, (0, i) ) : n , i 2, i = 1 f

( +n) x.
For distinct x, x

, p
x
and p
x
are incompatible. Hence, for each E, we can
nd a P-name x

such that
|
P
x

is a conal subset of and otp( x

) =
and
p
x
|
P
x

= x for each x S

.
We show that |
P
( x

)
E
is a (E)-sequence . For this, it is enough to show
that, for any p P and a P-name

A, if p |
P


A [
1
]

1
, then there is q p
and E such that q |
P
x

A. Let

f be such that
p |
P


f :
1


A and

f is 1-1 .
14
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


Now let (p

)
<
1
, (q

)
<
1
, (u

1
, Y and u

be as in the proof of Lemma 2.2.

For each Y let

be such that q

|
P


f() =

and let Z =

: Y .
Let
C = Lim(
1
) :

Y
(supp(q

)
1
)
and Z is unbounded in .
Then C is closed unbounded in
1
and hence there exists a

C E. Let
q

=

Y
q

. Then we have q

q and q

|
P
Z

A. Now let x S

be such that x Z

. Finally let q = q

q
x
. Then we have q p and
q |
P
x

= x Z

A. (Lemma 3.9)
Note that Es in Lemmas 3.7 and 3.9 are ground model sets. To force (E)
for every stationary E Lim(
1
) which may be also added generically, we need a
sort of iteration described in the next section.
Toward the consistency of (
w
) + (
w
2 ), we consider rst the following lemma
which should be a well-known fact. Nevertheless, we include here a proof:
Lemma 3.10 Assume that there is a sequence (C

)
<
of elements of [
1
]

1
such
that [ C

[
0
for all < < . Then there exists a partial ordering P with
the property K such that in V
P
there is a sequence (B

)
<
of elements of [
1
]

1
such that B

and [ B

[ <
0
for all < < .
Proof Let
P = (D, f) : D []
<
0
, f : D Fn(
1
, 2),
f() Fn(C

, 2) for all D.
For (D, f), (D

, f

) P, let
(D

, f

) (D, f) D D

, f() f

() for all D and

(f

())
1
[1] (f())
1
[1], D are pairwise
disjoint.
By the usual -system lemma argument, we can show that P has the property K.
Since C

, < are pairwise disjoint modulo countable, the set

T
,
= (D, f) P : D, dom(f()) and
> ( dom(f()) f()() = 1)
is dense in P for every < and <
1
. Hence if G is a V -generic lter over P,
then
B

= <
1
: f()() = 1 for some (D, f) G
15
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


is conal in
1
and hence uncountable. Also by the denition of on P, we have
[ B

[ <
0
for every < < . (Lemma 3.10)
Note that if there is a sequence (B

)
<
as in Lemma 3.10 then by the argument
in the proof of Proposition 3.4, we have [

.
Lemma 3.11 There is a partial ordering Q with the property K such that
|
Q
(
w
2) .
Proof Let (Q

,

R

1
be the nite support iteration of partial orderings with
the property K such that for each Lim(
1
), there is a Q

name

U

such that
Q

forces:

is an ultralter over ,

U

for all < ,

R

is a p.o.-set with the

property K and there is an

R

-name x

such that
|

R
x

is a conal subset of of ordertype and

[ x

a [ <
0
for all x

U

.
For example, we can take the Mathias forcing for the ultralter

U

as

R

. For
successor <
1
let |
Q


R

= 1 .
Let Q = Q

1
. As (Q

,

R

1
is a nite support iteration of property K
p.o.s, Q satises also the property K (see e.g. [9]). Now let G be a V -generic
lter over Q. In V [G], if X [
1
]

1
then the set <
1
: X V [G

]
contains a club subset C of Lim(
1
). Let S
0
= C : X

U

[G] and
S
1
= C : X

U

[G] . Since

U

[G] is an ultralter over in V [G

] for
every C, we have C = S
0

S
1
. We have [ x

[G] X [ <
0
for S
0
and
[ x

[G] X [ <
0
for S
1
. Thus ( x

[G])
Lim(
1
)
is a (
w
2)-sequence in V [G].
Actually this proof shows that ( x

[G])
Lim(
1
)
is even a (
w
2)-sequence in the
stronger sense that it satises the assertion of the denition of (
w
2) with is
stationary replaced by contains a club. (Lemma 3.11)
Corollary 3.12 There is a partial ordering R with property K such that |
R
[

2
but (
w
2) holds . In particular ( [

) + (
w
2) is consistent with ZFC. Further
if CH holds then for any cardinal , there exists a cardinals preserving proper partial
ordering R

such that |
R
[

but (
w
2) holds .
Proof Let R = P
1


P
2
where P
1
is as P in Lemma 3.10 for =
2
and

P
2
as Q
in Lemma 3.11 in V
P
1
.
For the second assertion, we let R

= Fn(, 2,
1
)

P
1


P
2
. Note that under
CH, Fn(, 2,
1
) is cardinals preserving and forces that 2

1
= . Hence there is a
sequence (C

)
<
as in Lemma 3.10 in the generic extension. Thus in V
Fn(,2,
1
)
,

P
1
can be taken as in Lemma 3.10 for our . Finally, in V
Fn(,2,
1
)

P
1
let

P
2
be as
in Lemma 3.11. (Corollary 3.12)
16
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


4 CS

-iteration
In this section, we introduce an iterative construction of p.o.s which is closely
related to the pseudo product we introduced in section 2. We adopt here the
conventions of [5] on forcing. In particular, a p.o. (or forcing notion) P is a pre-
ordering with a greatest element 1
P
. In the following, we just try to develop a
minimal theory needed for Theorem 5.6. More general treatment of the iterations
like the one described below should be found in [16].
We call a sequence of the form (P

,

Q

a CS

-iteration if the following

conditions hold for every :
*0) P

is a p.o. and, if < , then

Q

is a P

name such that |

P


Q

is a
p.o. with a greatest element 1

Q
.
*1) P

= p : p is a function such that dom(p) []

0
;
p[` P

for any < and,

if dom(p) then p]restr |
P

p()

Q

.
*2) For p, q P

, p
P
q if and only if
i) for any < , p[` |
P

p() q() ;
ii) di (p, q) = dom(p)dom(q) : p[` / |
P

p() = q() is nite.

We rst show that such a sequence (P

,

Q

is really an iteration in the

usual sense. In the following we assume always that (P

,

Q

is a CS

-iteration
as dened above.
Lemma 4.1 Suppose that . Then
0) if p P

, then p[` P

;
1) P

;
2) for p, q P

, we have p
P
q p
P

q;
3) for p, q P

, if p
P

q then p[`
P
q[`.
Proof 1) can be proved by induction on . Other assertions are clear from the
denition of CS

-iteration. (Lemma 4.1)

Lemma 4.2 Suppose that and p, q P

. Then p
P
q p
P

q.
Proof Suppose that p and q are compatible in P

, say r
P
p, q for some r P

.
Then r P

by Lemma 4.1, 1) and r

P

p, q by Lemma 4.1, 2). Hence p and q

are compatible in P

.
Conversely, suppose that p and q are compatible in P

, say s
P

p, q for
some s P

. Then we have s[` P

by Lemma 4.1, 0), s[`

P
p[` = p and
17
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


s[`
P
q[` = q. Hence p and q are compatible in P

. (Lemma 4.2)
Suppose that , p P

. By Lemma 4.1, 0), we have p[` P

. For
r
P
p[`, let
p

r = p[`(dom(p) ) r.
For p, q P

, p
h
P
q p
P
q and p[`dom(q) = q; p
v
P
q p
P
q and
dom(p) = dom(q) (h and v stand for horizontal and vertical respectively).
Lemma 4.3 1) Let , , p, r be as above. Then p

r P

and p

r
P

r,
p.
2) For p, q P

, r = q[`(dom(q) dom(p)) p is an element of P

and r
h
P
p.
3) If p
n
P

for n and p
n+1

h
P
p
n
for every n , then q =

p
n
: n
is an element of P

and q
h
P
p
n
for every n
Proof 1): By induction on . If = then p

r = r p[` = p. Suppose
that we have shown the inequality for every

< . Let p and r be as above. If

is a limit then we obtain easily p

r P

and p

r
P

r, p by checking
*1) and *2) of the denition of CS

-iteration. In particular, *2), ii) holds for the

inequality p

r
P

r, p since di (p

r, p) = di (r, p[`) and di (p

r, r) = .
If = +1 for some , then p[`

r P

, p[`

r
P
r, p[` by induction
hypothesis. If , dom(p) then it follows p = p[` P

and p

r
P

r, p.
Otherwise (p

r)[` |
P
p()

Q
p() . Hence again it follows that p

r P

and p

r
P

r, p.
2) and 3) are trivial. (Lemma 4.3)
Lemma 4.4 Suppose that , p P

and q P

. If p and q are
incompatible in P

then p and q[` are incompatible in P

.
Proof Suppose that p and q[` are compatible in P

. Then there is r P

such
that r
P
p, q[`. Let s = q

r. By Lemma 4.3, we have s
P

q, r. Hence p
and q are compatible in P

. (Lemma 4.4)
Lemma 4.5 Suppose that and that A is a maximal antichain in P

.
Then A is also a maximal antichain in P

.
Proof By Lemma 4.1, 1), we have A P

. By Lemma 4.2, A is an antichain

in P

. Suppose that A were not a maximal antichain in P

. Then there is some

q P

such that q is incompatible with each of p A. By Lemma 4.4, it follows

that q[` is incompatible with each of p[` = p, p A. This is a contradiction to
the assumption that A is a maximal antichain in P

. (Lemma 4.5)
18
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


5 CS

-iteration of Cohen reals

In the rest, we consider the CS

-iteration (P

,

Q

for a cardinal such that

|
P

= Fn(, 2)
for every < .
Lemma 5.1 Let p, q P

be such that p q. Then there is r P

such that r p
and for any di (r, q), there is t Fn(, 2) such that r[`|
P
r() =

t .
Proof We dene inductively a decreasing sequence (
n
)
n<
of ordinals and a
decreasing sequence (p
n
)
n
of elements of P

as follows: Let
0
= max di (p, q).
Choose p

0
P

0
so that p

0
p[`
0
and that p

0
decides p(
0
). Let p
0
= p

p

0
.
If
n
and p
n
have been chosen, let D
n
= di (p
n
, q)
n
. If D
n
= we are done.
Otherwise, let
n+1
= max D
n
. Choose p

n+1
P

n+1
such that p

n+1
p
n
[`
n+1
and p

n+1
decides p
n
(
n+1
). Let p
n+1
= p
n

n+1
. This process terminates after
m steps for some m , since otherwise we would obtain an innite decreasing
sequence of ordinals. Clearly r = p
m
is as desired. (Lemma 5.1)
Lemma 5.2 P

satises the axiom A.

Proof Let
n
, n be the relations on P

dened by p
n
q p
h
P
q for
p, q P

and every n (in Ishiu [6] an axiom A p.o., for which the
n
s can
be taken to be all the same, is called uniformly axiom A). (
n
)
n
has the fusion
property by Lemma 4.3, 3). Hence it is enough to show the following:
Claim 5.2.1 For any p P

and maximal antichain D P

, there is q
h
P
p
such that r D : r is compatible with q is countable.

Let : ; n (
1
(n),
2
(n)) be a surjection such that
1
(n) < n
for all n > 0 and, for any k, l , there are innitely many n such that
(n) = (k, l). We construct inductively p
k
, t
k
, u
k
P

and a sequence (s
k,l
)
l
for k as follows: let p
0
= p. If p
k
has been chosen then let (s
k,l
)
l
be an
enumeration of Fn(dom(p
k
), Fn(, 2)). If there are t D and u P

such that
u t, p

, di (u, p
k
) = doms

1
(k),
2
(k)
and u[`di (u, p
k
) = s

1
(k),
2
(k)
(of course we
identify here elements t of Fn(, 2) with corresponding P

-name

t), then let t
k
and
u
k
be such t and u and let p
k+1
= p
k
u[`(dom(u
k
) dom(p
k
)). By Lemma 4.3, 2),
we have p
k+1
P. Otherwise let t
k
= u
k
= 1
P
and p
k+1
= k
k
.
Now, let q =

k
p
k
. Then by Lemma 4.3, 3), we have q P

and q
P
p.
We show that this q is as desired.
Suppose that t D is compatible with q. Then by Lemma 5.1, there is u
P
t, q such that u[`di (q, r) has its values in Fn(, 2). Let n be such that
19
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


di (q, r) q
n
and k n be such that s

1
(k),
2
(k)
= u[`di (q, r). Clearly t
k
D
by construction. We claim that t = t
k
: otherwise t and t
k
would be incompatible.
Hence u
k
and u should be incompatible. But this is a contradiction.
It follows that
r D : r is compatible with q t
k
: k .

(Claim 5.2.1)
(Lemma 5.2)
In particular, P

is proper and hence the following covering property holds:

Corollary 5.3 Suppose that G is a P

-generic lter over V . Then for any a

V [G] such that V [G] [= a is a countable set of ordinals , there is a b V such
that a b and V [= b is a countable set of ordinals .
Lemma 5.4 If is strongly inaccessible, then P

satises the -cc.

Proof Suppose that p

for < . We show that there are compatible

conditions among them. Without loss of generality we may assume that dom(p

) :
< is a -system with the root x []

0
Let
0
= sup + 1 : x.
Then
0
< and p

[`x P

0
for every < . Since [ P

[ < there are ,

< ,
,=

such that p

[`x = p

[`x. But then q = p

and q
P
p

, p

.
(Lemma 5.4)
Lemma 5.5 Suppose that E Lim(
1
) is stationary. Then |
P
(E) .
Proof For each E let f

: [, +) be a bijection and let

S

= x : x is a conal subset of , otp(x) = .

For each x S

, let p
x
P

be dened by
p
x
= ( +n, q

x,n
) : n
where q

x,n
is the standard P
+n
-name for (0, i) with i 2 and i = 1 f

(+n)
x. For distinct x, x

, p
x
and p
x
are incompatible. Hence there is a P

-name x

such that |
P
x

is a conal subset of with otp( x

) = and p
x
|
P
x

= x
for every x S

.
We show that |
P
( x

)
E
is a (E)-sequence . Suppose that p P

and

A
is a P

-name such that p |

P


A [
1
]

1
. We have to show that there is q
P
p
and E such that q |
P
x

A.
Let

f be a P

-name such that p |

P


f :

A is 1-1 . Choose p

, q

, u

for
<
1
inductively such that
20
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


a) p
0

P
p and (p

)
<
1
is a decreasing sequence with respect to
h
P
;
b) q

v
P
p

and q

decides

f();
c) u

= di (q

, p

) dom(p)

<
dom(q

);
d) q

[`u

Fn(, Fn(, 2)).

The condition d) is possible because of Lemma 5.1. By Fodors lemma, there
is Y [
1
]

1
and r Fn(, Fn(, 2)) such that q

[`u

= r for every Y .
For each Y , there is

1
such that q

|
P


f() =

by b). Let
Z =

: Y . Let
C = Lim(
1
) :

Y
(sup(q

)
1
)
and Z is unbounded in .
Then C is closed unbounded in
1
. Since E was stationary, there exists a

C E. Let q

Y
q

. Then we have q

|
P
Z

A. Now let x S

be such that x Z

. Finally let q = q

p
x
. Then we have q
h
P
p and
q |
P
x

= x Z

A. (Lemma 5.5)
Let (P

,

Q

be a CS

-iteration as above. For < let P

/

G

be a P

-
name such that |
P
P

/

G

= p

P

: p[`

G

with the ordering p

,
q p
P
q . As in [5], we can show that P

/

G

. Also, by Corollary
5.3, practically the same proof as in [5] shows that
|
P
P

/

G

is to a CS

-iteration of Fn(, 2) .
Now we are ready to prove the main theorem of this section:
Theorem 5.6 Suppose that ZFC + there exists an inaccessible cardinal is con-
sistent. Then ZFC + CH + MA(countable) + (E) for every stationary E
Lim(
1
) is consistent as well.
Proof Suppose that is strongly inaccessible. For P

as above, let G

be a P

-
generic lter over V . We show that V [G

] models the assertions. Let E Lim(

1
)
be a stationary set in V [G

]. Since P

has the -cc by Lemma 5.4, there is some

< such that E V [G

] where G

= G

. Hence by the remark before

this theorem, we may assume without loss of generality that E V . But then, by
Lemma 5.5, we have V [G

] [= (E) .
Finally, we show that MA(countable) holds in V [G

]. Let T be a family of
dense subsets of Fn(, 2) in V [G

] of cardinality < . Again by the -cc of P

, we
can nd an < such that T V [G

]. Since we have
P

/

G
+1
,
21
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


the generic set over V [G

] added by

Q

[G

] = Fn(, 2) is T-generic over Fn(, 2)

in V [G

]. (Theorem 5.6)
At the moment we or more precisely the rst and the third author do
not know if an inaccessible cardinal is really necessary in Theorem 5.6. As for
CS-iteration, is collapsed to be of cardinality
2
in the model above, since the
continuum of each of the intermediate models is collapsed to
1
in the following
limit step of conality
1
. Thus the following problem seems to be a rather hard
one:
Problem 5.7 Is the combination MA(countable) + (E) for every stationary
E Lim(
1
) consistent with 2

0
>
2
?
Acknowledgments
The research of this paper began when the rst author (S.F.) was at the Hebrew
University of Jerusalem. He would like to thank The Israel Academy of Science and
Humanities for enabling his stay there. He also would like to thank T. Miyamoto
for some quite helpful remarks.
The second author (S.S.) was partially supported by the Deutsche Forschungs-
gemeinschaft(DFG) grant Ko 490/71. He also gratefully acknowledges partial
support by the Edmund Landau Center for research in Mathematical Analysis,
supported by the Minerva Foundation (Germany). The present paper is the second
authors Publication No. 544.
The third author (L.S.) is partially supported by the Hungarian National Foun-
dation for Scientic Research grant No. 16391 and the Deutsche Forschungsgemein-
schaft (DFG) grant Ko 490/71.
References
[1] B. Balcar, F. Franek: Independent families in complete Boolean algebras,
Trans. Amer. Math. Soc., Vol. 274 (1982), 607618.
[2] S. Broverman, J. Ginsburg, K. Kunen, F. Tall: Topologies determined by
-ideals on
1
, Can. J. Math., 30 No. 6 (1978), 13061312.
[3] M. Dzamonja and S. Shelah: Similar but not the same: versions of do not
coincide, preprint.
[4] S. Fuchino, S. Shelah, L. Soukup: On a theorem of Shapiro, Mathematica
Japonica, Vol. 40, No. 2 (1994).
22
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


[5] M. Goldstern: Tools for your forcing construction, in: H. Judah (ed.): Set
theory of the reals, Israel Mathematical Conference Proceedings, Bar Ilan
University (1992), 305360.
[6] T. Ishiu: Uniform axiom A, preprint.
[7] P. Komj ath: Set systems with nite chromatic number, European Journal of
Combinatorics, 10 (1989), 543549.
[8] K. Kunen: Set Theory (North-Holland, Amsterdam, New York, Oxford, 1980).
[9] K. Kunen, F. Tall: Between Martins axiom and Souslins hypothesis, Funda-
menta Mathematicae, Vol. 102 (1979), 173181.
[10] A.J. Ostaszewski: On countably compact perfectly normal spaces, J. London
Mathematical Society(2), 14 (1976), 505516.
[11] Saharon Shelah: Whitehead groups may not be free, even assuming CH. II,
Israel Journal of Mathematics, 35 (1980), 257285.
[12] : More on cardinal arithmetic, Archive for Mathematical Logic, 32 (1993),
399428.
[13] : Advances in cardinal arithmetic.
[14] : Further cardinal arithmetic, to appear in Israel Journal of Mathematics.
[15] : PCF and innite free subsets, in preparation.
[16] : Proper and Improper Forcing.
23
5
4
4


r
e
v
i
s
i
o
n
:
1
9
9
8
-
0
4
-
2
0







m
o
d
i
f
i
e
d
:
1
9
9
8
-
0
4
-
2
1


Authors addresses
Sakae Fuchino
Institut f ur Mathematik II,
Freie Universit at Berlin
14195 Berlin, Germany
[email protected]
Saharon Shelah
Institute of Mathematics,
The Hebrew University of Jerusalem
91904 Jerusalem, Israel
and
Department of Mathematics,
Rutgers University
New Brunswick, NJ 08854, USA
[email protected]
Lajos Soukup
Mathematical Institute
of the Hungarian Academy of Sciences
[email protected]
24