8

0
7


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







m
o
d
i
f
i
e
d
:
2
0
0
2
-
1
1
-
0
2


STRONGLY MEAGER SETS OF SIZE CONTINUUM
TOMEK BARTOSZYNSKI AND SAHARON SHELAH
Abstract. We will construct several models where there are no strongly mea-
ger sets of size 2

0
.
1. Introduction
In this paper we work exclusively in the space 2

equipped with the standard

product measure denoted as . Let ^ and / denote the ideal of all measure
zero sets, and meager subsets of 2

, respectively. For x, y 2

, x + y 2

is
dened as (x +y)(n) = x(n) +y(n) (mod 2). In particular, (2

, +) is a group and
is an invariant measure.
Denition 1 ([3]). A set X of real numbers or more generally, a metric space,
is strong measure zero if, for each sequence
n
: n of positive real numbers
there is a sequence X
n
: n of subsets of X whose union is X, and for each n
the diameter of X
n
is less than
n
.
The family of strong measure zero subsets of 2

is denoted by o^.
The following characterization of strong measure zero is the starting point for
our considerations.
Theorem 2 ([7]). The following are equivalent:
(1) X o^,
(2) for every set F /, X +F ,= 2

.
This theorem indicates that the notion of strong measure zero should have its
category analog. Indeed, we dene after Prikry:
Denition 3. Suppose that X 2

. We say that X is strongly meager if for every

H ^, X +H ,= 2

. Let o/ denote the collection of strongly meager sets.

Observe that if z , X + F = x + f : x X, f F then X (F + z) = .
In particular, a strong measure zero set can be covered by a translation of any
dense G

set, and every strongly meager set can be covered by a translation of any
measure one set.
Let Borel Conjecture denote the statement o^ = [2

]
0
and Dual Borel Con-
jecture the statement: o/ = [2

]
0
.
The question whether Borel Conjecture and Dual Borel Conjecture are jointly
consistent is the motivation for this paper. In particular, we are interested in what
kinds of strongly meager sets exist in various models.
First author partially supported by NSF grant DMS 0200671.
Second author partially supported by Israel Science Foundation and NSF grant DMS 0072560.
Publication 807.
1
8
0
7


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







m
o
d
i
f
i
e
d
:
2
0
0
2
-
1
1
-
0
2


2 TOMEK BARTOSZYNSKI AND SAHARON SHELAH
2. Preservation of not being strongly meager
A small modication [10] of the denition of strongly meager sets leads to a
family that captures the concept of strongly meager sets while it has some additional
properties.
Denition 4. We say that X o/
+
if for every H ^, there exists a countable
set Z 2

such that
x X Z , x +H.
Lemma 5. (1) o/ o/
+
,
(2) o/
+
is a -ideal,
(3) it is consistent that o/ = o/
+
,
(4) it is consistent that o/ ,= o/
+
.
Proof. (1) and (2) are obvious.
(3) o/ = o/
+
holds in Cohens model, and in fact every known model where
o/ = [2

]
0
[4].
(4) By the results of [2], assuming CH, o/ is not a -ideal.
Denition 6. Suppose that N H() is a countable model. A G

set H ^ is
big over N if 2

N H.
We say that a proper forcing notion P is nice if for every p P, a countable
model N H() containing p, P and a set H which is big over N there exists a
condition q p such that
(1) q is (N, P)-generic over N,
(2) q
P
H is big over N[

G].
Theorem 7. Suppose that V [= X , o/
+
and P is nice. Then V
P
[= X , o/
+
.
Proof. Suppose that X , o/
+
and let H ^ be a G

set such that

Z [V 2

x X Z H +x.
Let x
n
: n be a P-name for a countable set of reals, and let p P. Find a
countable model N H() containing all relevant objects and let x X be such
that N2

H+x. Since P is nice, there exists q p such that q

P
n x
n
H+x.
It follows that V
P
[= X , o/
+
.
The following theorem is a particular instance of a preservation result due to
Eisworth, [5]. We give a sketch of the proof for completeness.
Theorem 8. Let T

,

Q

: < be a countable support iteration of proper

forcing notions such that every < ,

is nice. Then T

is nice.
Proof. The proof follows standard proof of preservation of properness.
Lemma 9 ([8], [1] lemma 6.1.2). Let T

,

Q

: < be a countable support

iteration. Assume < , p T

and is a T

-name for an ordinal. There

is a T

-name for an ordinal and a condition q

p such that q = p and

q

= .
Let N be a countable submodel of H() containing P

and let H =

n

m>n
U
m
be big over N, where U
m
s are basic open sets.
8
0
7


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







m
o
d
i
f
i
e
d
:
2
0
0
2
-
1
1
-
0
2


STRONGLY MEAGER SETS OF SIZE CONTINUUM 3
Lemma 10. Let T

,

Q

: < be a countable support iteration and assume that

for all < ,

is nice. Then for all N , for all N and for

all p T

N, whenever q

p is (N, T

)-generic, there is an (N, T

)-generic
condition r

p such that r = q and r N[G

] 2

H.
Proof. The proof is by induction on .
Successor step. Let =

+1. We assume that N so

N too. Using
the induction hypothesis on ,

we can extend q to a (N, T

)-generic condition
q

such that q

and q

= q. Hence without loss of generality we

can assume = + 1.
Since

N[

G

] H()
V[

G]
, q forces that there is an (N[

G

],

Q

)-generic
condition p() which forces that N[G

] 2

H. It has a T

-name, r() and

r = q

r() is (N, T

)-generic and r N[G

] 2

H.
Limit step. Let N be a limit ordinal and =
0
<
1
< be a sequence
of ordinals in N such that sup(N ) = sup
n

n
. Let
n
: n ) enumerate all
T

-names for ordinals which are in N, let x

n
: n ) enumerate all T

-names for
reals in N with innitely many repetitions. By induction on n using lemma
9, we dene conditions p

n
T

N, T
n
-names

n
N for ordinals such that for
each n,
(1) p

0
= p and p
n+1
p
n
,
(2) p

n+1

n
=
n
,
(3) p

n+1

n
= p
n

n
,
Up to now the proof followed standard proof of preservation of properness. Now
we will make a small change relevant to the current setup.
We will rene the sequence p
n
: n as follows. Let p
1
m
: m and
x
0
2

N be such that
(1) p
1
0
= p

1
,
(2) p
1
m+1
p
1
m
, p
1
m
T

,
(3) p
1
m
x
0
m = x
0
m.
Since x
0
U
k
for innitely many k, it follows that for some m, p
1
m
x
0
U
m
.
Let G be a T
0
-generic lter over N containing p
0

0
. Construct sequence
p
2
m
: m and x
1
2

N[G] be such that

(1) p
2
0
= p

1
[
0
, ),
(2) p
2
m+1
p
2
m
, p
2
m
T
0,
,
(3) p
2
m
x
1
m = x
1
m.
As before, N[G] [= x
1
U
k
for innitely many k, so it follows that for some
m > 0, N[G] [= p
2
m

0,
x
1
U
m
. Thus there is a T
0
-name m for an integer
such that
p
2
= p
1

0
p
2
m
m > 0 x
1
U
m
.
We continue in this fashion and construct a sequence p
n
: n ) such that
(1) p
0
= p and p
n+1
p
n
,
(2) p
n+1

n
=
n
,
(3) p
n+1

n
= p
n

n
,
(4) p
n+1
m > n x
n
U
m
.
Note that all conditions p
n
are in N (while the sequence is not). Now using the
induction hypothesis we dene conditions q
n
T
n
such that q
0
= q, q
n
p
n

n
,
q
n+1

n
= q
n
and q
n
is an (N, T
n
)-generic condition. We set r =

n
q
n
.
8
0
7


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







m
o
d
i
f
i
e
d
:
2
0
0
2
-
1
1
-
0
2


4 TOMEK BARTOSZYNSKI AND SAHARON SHELAH
Now for each n, since supp(p
n+1
) N, supp(p
n+1
) dom(r) and by (1) and
(3), r p
n+1
. Hence r

n
=

n
, and q
n
= r
n

n

n
N. Likewise
r

n
N for each n , and r is (N, T

)-generic. Similarly, for every n,

r

m x
n
U
m
.

The following denition gives an easy to verify property of forcing notion which
implies niceness.
Denition 11. Suppose that P is a forcing notion satisfying Axiom A. P has
PPP-property if for every P-name x such that
P
x 2

and every p P, n
there exists k such that for every there exists A [2

]
<k
and q
n
p,
q
P
x A.
Observe that PPP-property is a weak version of Laver property (which is equiv-
alent to the requirement that k depends only on n but not on p). Thus Laver,
Miller and Sacks forcings have property PPP (since they have the Laver property),
as well as many tree forcings.
Lemma 12. If P has PPP-property then P is nice. In particular, if P has Laver
property then P is nice.
Proof. Suppose that N H() is a countable model containing P and p P.
Suppose that H ^ is big over N. Let U
j
: j ) be open sets such that
H =

j
U
j
. We want to nd a condition q p which is (N, P)-generic such that
q
P
2

N[

G] H.
Basic step: Suppose that x N is a P-name for a real, n and p P N.
Using PPP-property we construct reals x
1
, . . . , x
k
N 2

and a sequence of
conditions p
m
: m ) N such that
(1) p = p
0
,
(2) p
m+1

n
p
m
,
(3) p
m

P
j k xm = x
j
m.
Since sets U
j
are open it follows that for every j there is m such that p
m

P
x U
j
. In other words, given p, n , j , and x as above we can nd q
n
p,
q N and q
P
x U
j
.
The rest of the construction is standard: we build in V a sequence (j
n
, x
n
, A
n
) :
n ) of all triples (j, x, A) where j , x N is a P-name for a real and A N
is a maximal antichain of P. Using the basic step above, we build by induction a
sequence p
n
: n ) such that
(1) p
0
= p,
(2) p
n
N,
(3) p
n+1

n
p
n
,
(4) p
n+1

P
x
n
U
jn
,
(5) p
n+1
is compatible with countably many elements of A
n
.
Finally, let q be such that q p
n
for all n. Clearly, q has the required properties.

3. No small sets of size continuum

Suppose that is a -ideal of sets of reals having Borel basis. Dene

=
X 2

: A X +A ,= 2

. In particular, /

= o^ and ^

= o/.
8
0
7


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







m
o
d
i
f
i
e
d
:
2
0
0
2
-
1
1
-
0
2


STRONGLY MEAGER SETS OF SIZE CONTINUUM 5
Denition 13. We say that T has property P
J
if for every condition p T, and
a T-name for a real x such that p
P
x , V 2

there exists a set H V

such that for all z V 2

, there exists q
z
p such that
q
z
x H +z.
Theorem 14. Let T

,

Q

: <
2
be a countable support iteration of proper
forcing notions. If for every <
2
, T
2
/T

has property P
J
, then V
P
2
[=

[R]
<2

0
.
Proof. Suppose that X 2

V
P
2
and
2
X

. Working in V
P
2
, for every
Borel set H nd a real x
H
such that V
P
2
[= x
H
, H +X. Granted that for
every <
2
, V
P

[= CH, we can nd <

2
such that x
H
V
P
for H V
P
.
Suppose that p
2
x , V
P
2

. Apply property P
J
to nd H with the
required properties. In particular, there is q p such that q
2
x H + x
H
. It
follows that q
2
x
H
H + x, thus q
2
x , X. Since x and p were arbitrary,
we conclude that X V
P
2

.
Denition 15. Suppose that P satises Axiom A. We say that P is strongly

-
bounding if for every p P, n and a P-name n, if p n then there exists
q
n
p and a set A []
<
such that q n A.
The following is (essentially) proved in [9], see also [1] theorem 8.2.14.
Theorem 16. If P is strongly

-bounding then P has property P

M
. Let T

,

Q

:
<
2
be a countable support iteration of proper forcing notions such that for
<
2
,

is strongly

-bounding. Then V
P
2
[= /

= o^ [R]
<2

0
.
Laver forcing (and many other forcing notions) have property P
N
, a somewhat
weaker result was proved in [11], but whether this property is preserved, particularly
for posets adding unbounded reals, is unclear. Therefore we will use the following
notions.
Denition 17. Suppose that N H(). We say that a sequence of clopen sets
C = C
n
: n ) is big over N if
(1) C
n
s have pairwise disjoint supports,
(2) (C
n
) 2
n
for n ,
(3) for every innite set X 2

, X N, there exists innitely many n such

that X +C
n
= 2

.
Lemma 18. Suppose that C
n
: n ) is big over N. Then for every x 2

,
C
n
+x : n ) is big over N.
Proof. For a clopen set C, if C +X = 2

then C +x +X = 2

+x = 2

.
Theorem 19. For every countable model N H() there exists a set H ^
which is big over N.
Proof. We will need the following theorem:
Theorem 20 ([6], [1] theorem 3.3.9). Suppose that X 2

is an innite set and

> 0. Then there exists a clopen set C 2

such that (C) < and X +C = 2

.
Let N H() be a countable model. Fix an enumeration (with innitely many
repetitions), X
n
: n of all innite subsets of N 2

that are in N. By
theorem 20, for each n there exists a clopen set C
n
of measure < 2
n
such that
C
n
+X
n
= 2

.
8
0
7


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







m
o
d
i
f
i
e
d
:
2
0
0
2
-
1
1
-
0
2


6 TOMEK BARTOSZYNSKI AND SAHARON SHELAH
Corollary 21. Suppose that c is a Cohen real over V. There exists a sequence
C
n
: n ) V[c] which is big over V.
Denition 22. We say that a proper forcing notion P is good if for every p P, a
countable model N H() containing p, P and a sequence C = C
n
: n ) which
is big over N there exists a condition q p such that
(1) q is (N, P)-generic over N,
(2) q
P
C
n
: n ) is big over N[

G],
(3) q
P
F
C
N = F
C
N[

G], where F
C
= x 2

n x , C
n
.
Theorem 23. If T

,

Q

: <
2
is a countable support iteration and for every
<
2
,

is good then T
2
is good.
Proof. Similar to the proof of theorem 8. Note that the requirements are repre-
sented by G

sets.
Theorem 24. Let T

,

Q

: <
2
be a countable support iteration of proper
forcing notions such that every <
2
,

is good. Then V
P
2
[= o/
[R]
<2

0
.
Proof. Suppose that X 2

V
P
2
and V
P
2
[= X o/. As in the proof of
theorem 14, for H ^ let x
H
2

be such that x
H
, H+X. We can assume that
x
H
V if H V. Suppose that x is a T
2
-name for a real and p
P
2
x , V. Let
N H() be a countable model containing p, T
2
, and x. Let C
n
: n ) V be
big over N. Put H =

m

n>m
C
n
^ and let C = C
n
+x
H
: n ). Sequence
C is also big over N, and H +x
H
=

m

n>m
C
n
+x
H
. Since T
2
is good we can
nd q p such that
(1) q is (N, T
2
)-generic over N,
(2) q
2
C is big over N[

G],
(3) q
2
F
C
N = F
C
N[

G].
Since p x , N it follows that q x , F
C
= 2

(H + x
H
). In particular,
q x H + x
H
, and as before, q x , X. Thus X V 2

, which nishes the

proof.
4. Laver and Miller forcings
In this section we will show that Laver and Miller forcings are good, and therefore
there are no strongly meager sets of size continuum in Lavers or Millers model.
For a tree p 2
<
let [p] denote the set of branches of p, and let p
s
= t p : s
t or t s. For s p and m > [s[ let succ
m
p
(s) = t p : [s[+[t[ = m and s

t p.
We will suppress the superscript m if its value will be clear from the context.
Let split(p) = s p : [succ
p
(s)[ > 1 =

n
split
n
(p), where split
n
(p) =
_
s split(p) :

_
t s : t split(p)
_

= n
_
. The unique element of split
0
(p) is called
stem(p). For s p let A
s
= n : s

n p
Denition 25. The Laver forcing L is the following forcing notion:
p L p
<
is a tree & s p
_
[s[ stem(p) [succ
p
(s)[ =
0
_
.
For p, q L, p q if p q.
For every p L and s
<
dene a node p(s) in the following way: p() =
stem(p) and for n let p(s

n) be the n-th element of succ

p
_
p(s)
_
.
8
0
7


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







m
o
d
i
f
i
e
d
:
2
0
0
2
-
1
1
-
0
2


STRONGLY MEAGER SETS OF SIZE CONTINUUM 7
For p, q L and n dene
p
n
q p q & s n
n
p(s) = q(s).
In particular, p
0
q is equivalent to p q and stem(p) = stem(q).
The rational perfect forcing (Miller forcing) M is the following forcing notion:
p M p
<
is a perfect tree &
s p t p
_
s t & [succ
p
(t)[ =
0
_
.
For p, q M, p q if p q. Without loss of generality we can assume that
[succ
p
(s)[ = 1 or [succ
p
(s)[ =
0
for all p M and s p. Conditions of this type
form a dense subset of M.
For p, q M, n , we let
p
n
q p q & split
n
(q) = split
n
(p).
Theorem 26. Laver and Miller forcings are good.
Proof. We will prove the theorem for the Laver forcing L, the proof for Miller
forcing is similar (and easier).
Lemma 27. Suppose that p
L
x 2

. There exists q
0
p a set of reals x
s
: s
q, stem(q) s such that
(1) if stem(q) s q then for every n there is r
0
q
s
such that r
L
xn =
x
s
n,
(2) for every s, mapping n x
s

n
(n A
s
) is either one-to-one or constant,
(3) lim
nAs
x
s

n
= x
s
.
Proof. Induction on levels of p. Suppose that [stem(p)[ = k and dene trees p
n
:
n such that
(1) p
0
= p,
(2) p
n+1

k+n
= p
n

k+n
for every n,
(3) reals x
s
are dened for s p
n+1

k+n
and satisfy conditions (1) and (3).
Suppose that p
n
is given, and s p
n

k+n
. For each l A
s
let r
s
l

0
(p
n
)
s

n
and s
l
2
l
be such that r
s
l
xl = s
l
. Let A

s
A
s
be an innite set such that

lA

s
s
l
= x
s
2

. Finally let p
n+1
=

spn
k+n

lA

s
r
s
l
. Clearly, q

=

n
p
n
satises (1) and (3). By pruning it further we get q
0
q

that satises (2) as

well.
Recall that B p is a front in p, that is for every branch x [p] there is n
such that xn B.
Lemma 28. Suppose that p L and B p. Then either there exists q
0
p such
that q B = or there exists q
0
p such that B is a front in q.
Proof. Dene rank function on p as follows, for s p
(1) rank
B
(s) = 0 if n [s[ sn B,
(2) If rank
B
(s) ,= 0 then
rank
B
(s) = min : A [A
s
]

n A rank
B
(s

n) < .
If rank
B
(stem(p)) is dened then let q
0
p be such that rank
B
is dened on q and
rank
B
(s) > rank
B
(t) whenever t s. It follows that B is a front in q.
If rank
B
(stem(p)) is not dened then there is q
0
p such that rank
B
(s) is
undened for s q. Thus q B = .
8
0
7


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







m
o
d
i
f
i
e
d
:
2
0
0
2
-
1
1
-
0
2


8 TOMEK BARTOSZYNSKI AND SAHARON SHELAH
Suppose that N H(), C
n
: n ) is big over N.
We will need the following two basic constructions.
Basic Step I. Given p N L, k, m and an L-name for a real x N such
that p
L
x , N 2

we need q
k
p and n > m such that q
L
x C
n
.
We can assume that k = 0, otherwise we repeat the construction described below
for each protected node.
Working in N we can shrink p (without changing the stem) so that it satises
conclusion of lemma 27. Let B = s p : i, j A
s
x
s

i
,= x
s

j
. Applying
lemma 28 we can further shrink p (without changing the stem) so that B is a front
in [p]. Note that if p B = then p
L
x = x
stem(p)
, which contradicts the choice
of x.
Let Z B be a maximal antichain in p. Fix s Z and let X
s
= x
s

l
: l A
s
.
Since N [= X
s
[2

the set
B
s
= l A
s
: n > m x
s

l
C
n

is innite. For each l B

s
x n
l
such that x
s

l
C
n
l
. Let q
s
l

0
p
s

l
be such
that q
s
l

L
xsupp(C
n
l
) = x
s

l
supp(C
n
l
). In particular, q
s
l

L
x C
n
l
.
Let q
s
=

lBs
q
s
l
and let q =

sZ
q
s
. Clearly q
0
p and q
L
n > m x C
n
.
Basic Step II. Given p N L, k, l and an L-name

X = x
j
: j for a
countable set of reals in N
L
we need q
k
p and n > l such that q
L

X+C
n
= 2

.
As before we can assume that k = 0. Dene sequences q
j
: j ), s
j
i
: i
j, j such that
(1) q
0
= p,
(2) q
j+1

0
q
j
,
(3) [s
j
i
[ > j for i j,
(4) s
j
u
,= s
j
v
if u, v j, u ,= v,
(5) s
j
u
s
i
u
if u j i.
(6) q
j+1

L
i n s
j
i
x
i
,
Let x
i
=

j>i
s
j
i
. Reals x
i
are pairwise dierent and X = x
i
: i N. By
the assumption, there exists n > l such that X + C
n
= 2

. Find l such that

C
n
2
l
and let j be so large that q
j

L
xl : x X = s
j
i
l : i j. Clearly,
q
j

L

X +C
n
= 2

.
The rest of the proof is standard induction. We proceed as in the proof of lemma
12. Let (

X
n
, x
n
, A
n
) : n be an enumeration of all sets (

X, x, A) in N where

X is an L-name for a countable sets of reals in N, x is a L-name for a real such

that x , N 2

, A is a maximal antichain in L. Using the basic steps above, we

build by induction a sequence p
n
: n ) such that
(1) p
0
= p,
(2) p
n+1

n
p
n
,
(3) p
n+1
m > n x
n
C
m
,
(4) p
n+1
m > n

X
n
+C
m
= 2

,
(5) p
n+1
is compatible with countably many elements of A
n
.
Note that conditions produced in step I are not necessarily in N, but they are of
form

sZ
q
s
where Z N and q
s

l
N for s Z, l . This suces to carry
out the construction.
It is not clear whether every forcing notion having Laver property (or PPP
property) is nice, but many of them are.
8
0
7


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







m
o
d
i
f
i
e
d
:
2
0
0
2
-
1
1
-
0
2


STRONGLY MEAGER SETS OF SIZE CONTINUUM 9
5. Strongly meager sets of size continuum
The simplest case when we have large strongly meager sets is when there are
many random reals around.
Lemma 29. If cov(^) = 2
0
then [2

] o/.
Proof. Suppose that [X[ < 2
0
and H ^. It follows that H+X =

xX
H+x ,=
2

. Therefore X o/.
To construct a strongly meager set of size continuum x an enumeration H

:
< 2
0
of Borel measure zero sets and construct inductively reals x

, z

: <
2
0
such that
(1) x

: < (H

+z

) = ,
(2) x

,

<
H

+z

.
Note that X = x

: < 2
0
is the required set since z

, H

+X.
Theorem 30. It is consistent with ZFC that cov(^) < 2
0
and there exists a
strongly meager set of size continuum.
Proof. Let V

be a model obtained by adding

2
Cohen reals to a model V
0
[= CH,
and let V be obtained by adding
1
random reals (side-by-side) to V

. It is easy to
see that V [= V

o/ (as witnessed by random reals), and it is well known

([12], [1] theorem 3.3.24) that V [= cov(^) =
1
.
References
[1] Tomek Bartoszynski and Haim Judah. Set Theory: on the structure of the real line. A.K.
Peters, 1995.
[2] Tomek Bartoszynski and Saharon Shelah. Strongly meager sets do not form an ideal. Journal
of Mathematical Logic, 1:134, 2001.
[3] Emil Borel. Sur la classication des ensembles de mesure nulle. Bulletin de la Societe Math-
ematique de France, 47:97125, 1919.
[4] Timothy J. Carlson. Strong measure zero and strongly meager sets. Proceedings of the Amer-
ican Mathematical Society, 118(2):577586, 1993.
[5] Todd Eisworth. Contributions to the theory of proper forcing, 1994. Ph.D. thesis.
[6] Paul Erdos, K. Kunen, and R. Daniel Mauldin. Some additive properties of sets of real
numbers. Fundamenta Mathematicae, 113(3):187199, 1981.
[7] Fred Galvin, Jan Mycielski, and Robert Solovay. Strong measure zero sets. Notices of Amer-
ican Mathematical Society, pages A280, 1973.
[8] Martin Goldstern. Tools for your forcing constructions. In Haim Judah, editor, Set theory of
the reals, Israel Mathematical Conference Proceedings, pages 305360. Bar Ilan University,
1992.
[9] Martin Goldstern, Haim Judah, and Saharon Shelah. Strong measure zero sets without Cohen
reals. The Journal of Symbolic Logic, 58(4):13231341, 1993.
[10] Marcin Kysiak. Masters Thesis, Warsaw University, 2000. in Polish.
[11] Andrzej Nowik and Tomasz Weiss. Strongly meager sets of real numbers and tree forcing
notions. Proceedings of the American Mathematical Society, 130(4):11831187, 2002.
[12] Janusz Pawlikowski. Why Solovay real produces Cohen real. The Journal of Symbolic Logic,
51(4):957968, 1986.
Department of Mathematics and Computer Science, Boise State University, Boise,
Idaho 83725 U.S.A.
E-mail address: tomek@math.boisestate.edu, http://math.boisestate.edu/~tomek
Department of Mathematics, Hebrew University, Jerusalem, Israel and Department
of Mathematics, Rutgers University, New Brunswick, NJ
E-mail address: shelah@sunrise.huji.ac.il, http://math.rutgers.edu/~shelah/