% LaTeX 2.09 \documentstyle[12pt]{article} \input amssym.def \input amssym %% if amssym.def and amssym.tex are not available uncomment lines below %\def\Bbb#1{{\bf #1}} %\def\goth#1{{\bf #1}} %\def\upharpoonright{\uparrow} %\def\blacksquare{\Box} \newcommand{\res}{\upharpoonright} \newcommand{\qed}{\nopagebreak\par\noindent\nopagebreak$\blacksquare$\par} \newcommand{\bairespace}{\omega^{\omega}} \newcommand{\binseq}{2^{<\omega}} \newcommand{\cantorspace}{2^{\omega}} \newcommand{\concat}{{\;\hat{ }\;}} \newcommand{\continuum}{{\goth c}} \newcommand{\cross}{\times} \newcommand{\forces}{\mid\vdash} \newcommand{\Intersect}{\bigcap} \newcommand{\Intersection}{\bigcap} \newcommand{\intersect}{\cap} \newcommand{\intersection}{\cap} \newcommand{\proof}{\par\noindent proof:\par} \newcommand{\rationals}{{\Bbb Q}} \newcommand{\reals}{{\Bbb R}} \newcommand{\seq}{\omega^{< \omega}} \newcommand{\Union}{\bigcup} \newcommand{\union}{\cup} \newtheorem{theorem}{Theorem} \newtheorem{lemma}[theorem]{Lemma} \newtheorem{corollary}[theorem]{Corollary} \newcount\quesno \quesno=0 \def\ques{\advance\quesno by 1 \par\bigskip \noindent(Q\the\quesno)} \begin{document} \begin{center} {\Huge Special sets of reals} \footnote{in {\bf Set Theory of the Reals}, ed Haim Judah, Israel Mathematical Conference Proceedings, vol 6 (1993), 415-432, American Math Society.} \end{center} \begin{center} Arnold W. Miller \end{center} \section*{Introduction} This is a survey paper which will update Miller [1984b]. It is concerned with totally imperfect subsets of the reals. A set of reals is totally imperfect iff it does not contain an uncountable closed set. Examples of such sets are: Luzin sets, Sierpi\'{n}ski sets, concentrated sets, strong measure zero sets, universal measure zero sets, perfectly meager sets, strong first category sets, $\sigma$-sets, $\lambda$-sets, $Q$-sets, $\gamma$-sets, $s_0$-sets, and $C^\prime$-sets. We have concentrated this survey on answering the questions which appeared in the earlier one. \section*{Forcing Axiom} \ques Is $\neg CH$ consistent with either of: (P1) For every ccc poset $\Bbb P$ of cardinality less than or equal to the continuum $\goth c$, there exists a Luzin set of filters, i.e., $\langle G_\alpha:\alpha<\omega_1\rangle$ such that each $G_\alpha$ is a $\Bbb P$-filter and for every dense set $D\subseteq \Bbb P$ for all but countably many $\alpha$ we have $G_\alpha\intersect D\not=\emptyset$. (P2) For all ccc $\sigma$-ideals $I$ in the Borel subsets of $\reals$ there exists an $I$-Luzin set, i.e., an uncountable set $X\subseteq\reals$ such that for every $A\in I$ we have $A\intersect X$ is countable. \bigskip These two properties seem related because of the following result of Martin and Solovay [1970]. \begin{theorem} (Martin, Solovay) MA is equivalent to the statement that for every ccc $\sigma$-ideal $I$ in the Borel subsets of $\reals$ and for every $J\in [I]^{<\continuum}$ we have $\Union J\not=\reals$. \end{theorem} \begin{corollary} \label{corms} MA is equivalent to the statement that $\continuum$ is regular and for every ccc $\sigma$-ideals $I$ in the Borel subsets of $\reals$ there exists a $\continuum$-$I$-Luzin set, i.e., $X\in[\reals]^{\continuum}$ such that for every $A\in I$ we have $|A\intersect X|<\continuum$. \end{corollary} In Miller-Prikry [1984a] it is shown that (P2) is consistent with $\neg CH$. It is also shown that the regularity of $\continuum$ is necessary in Corollary \ref{corms}. Surprisingly Todorcevic [1991] has shown that (P1) implies CH. \begin{theorem} (Todorcevic) \label{tod} The continuum hypothesis is equivalent to the proposition that every ccc poset $\Bbb P$ of cardinality less than or equal to the continuum $\goth c$ has a Luzin set of filters. \end{theorem} \proof That this is true assuming CH is easy since such a poset contains only $\omega_1$ maximal antichains. To prove the other direction we will use the following four lemmas. First define for $f,g\in\bairespace$, $f\leq g$ iff for all $n\in\omega$ we have $f(n)\leq g(n)$, and $f\leq^* g$ iff for all but finitely many $n\in\omega$ we have $f(n)\leq g(n)$. \begin{lemma}\label{tod1} Suppose $ F\subseteq\bairespace$ are increasing functions well-ordered by $\leq^*$ and unbounded. Then there exists distinct $f,g\in {F}$ such that $f\leq g$. \end{lemma} \proof Since $F$ is unbounded its order type under $\leq^*$ cannot have countable cofinality. Consequently, we can find $g_0\in { F}$ such that $\{f\in { F}: f\leq^* g_0\}$ is dense in $ F$. Let ${ F}^\prime\subseteq F$ be unbounded such that there exists $n_0$ such that for every $f\in F^\prime$ and $n\geq n_0$ we have $g_0(n)\leq f(n)$. Let $$T=\{s\in\seq : \{f\in F^\prime :f\supseteq s\}\; {\rm unbounded }\}.$$ It is easy to see that $T$ cannot be finitely branching. Consequently, there exists $s\in T$ with $|s|\geq n_0$ and for which there exists infinitely many $n$ such that $s\concat n\in T$. Let $f\supset s$ with $f\leq^*g_0$ and $f\in F$. For some $k>n_0$ we have that $g_0(l)\geq f(l)$ for all $l\geq k$. Let $g\in F^\prime$ be such that $g\supseteq s\concat n$ where $n>f(k)$. Then $f\leq g$ because: \begin{itemize} \item $f(l)=g(l)=s(l)$ for $l\leq |s|$, \item $f(l)\leq f(k)\leq g(|s|)=n\leq g(l)$ for $|s|\leq l\leq k$ (by choice of $n$ and since they are increasing), \item $f(l)\leq g_0(l)\leq g(l)$ for $l>k$ (by choice of $k$ and definition of $F^\prime$). \end{itemize} \qed \begin{lemma}\label{tod2} Suppose $ F\subseteq\bairespace$ are increasing functions ordered by $\leq^*$ in type $\omega_1$ and unbounded. Suppose $\Bbb P$ is the poset of finite pairwise $\leq$-incomparable subsets of $ F$. Then $\Bbb P$ has ccc. \end{lemma} \proof First note that for every $A,B\subseteq F$ unbounded there exists unbounded $A^\prime\subseteq A$ and $B^\prime\subseteq B$ such that for every $f\in A^\prime$ and $g\in B^\prime$ we have $f$ and $g$ are $\leq$-incomparable. This is easy to see since if we let $$T_A=\{s\in\seq : \{f\in A :f\supseteq s\}\; {\rm unbounded }\}$$ $$T_B=\{s\in\seq : \{f\in B :f\supseteq s\}\; {\rm unbounded }\}$$ then these are both infinite branching trees and so we can choose $s\in T_A$ and $t\in T_B$ with $s(k) h_\beta(\alpha)$ such that $f_\gamma\in p$. \end{itemize} Order $\Bbb P$ by inclusion on each coordinate. \bigskip\noindent Claim. $\Bbb P$ has ccc. \proof If $p\union p^\prime\in \Bbb P_{F }$, then by Lemma \ref{tod3} $$(p\union p^\prime\union \{f_\gamma\}, A \union A^\prime)\leq (p,A),\; (p^\prime,A^\prime)$$ for all sufficiently large $\gamma$.\qed Note that $\{(p,A):\gamma\in A\}$ is dense for each $\gamma\in\omega_2$. If there is a Luzin set of filters, then one of them must meet at least $\omega_2$ of these dense sets. Consequently there exists $\Bbb P$-filter $G$ with $|G|=\omega_2$. Now look at $X=\Union\{ A:\exists p\; (p,A)\in G\}$. Since $X$ has cardinality $\omega_2$, it has an $\omega_1^{th}$ element and so the set $\Union\{p:\exists A\; (p,A)\in G\}$ has cardinality $\omega_1$. But this contradicts Lemma \ref{tod1}. This proves Theorem \ref{tod}. \qed Note that this proof also gives a counterexample to Theorem 2.3 of Miller and Prikry [1984a]. The mistake in the proof occurs in the first sentence where it is stated that ``without loss of generality we may assume $\Bbb P$ is a boolean algebra''. To correct that theorem we must change it to read: \begin{theorem} Suppose GCH and $\Bbb P$ is ccc of cardinality $\omega_2$, then there exists a countably closed poset $\Bbb Q$ with $\omega_2$-cc and cardinality $\omega_2$ such that in $V^{\Bbb Q}$ there exists $\langle G_\alpha:\alpha<\omega_1\rangle$ centered subsets of $\Bbb P$ such that for every dense $D\subseteq \Bbb P$ for all but countably many $\alpha$ we have $G_\alpha\intersect D\not=\emptyset$. \end{theorem} Now, of course, in a boolean algebra centered subsets generate filters, but in a partial order this may not be the case. This seems to be one of the ingredients of Todorcevic's example. The following example sheds more light on this. \begin{theorem} (Juhasz-Kunen) There exists a poset $\Bbb P$ which is $\sigma$-centered but not $\sigma$-filtered. \end{theorem} \proof Let $\Bbb P=\{(C,F): C\subseteq\cantorspace$ clopen, $\emptyset\not= F\in [C]^{<\omega}, \mu(C)\leq {1\over |F|}\}$. Order $\Bbb P$ by: $(C_0,F_0)\leq (C_1,F_1)$ iff $C_0\subseteq C_1$ and $F_0\supseteq F_1$. To see that $\Bbb P$ is $\sigma$-centered, note that there are only countably many clopen sets and any finite set of conditions with the same clopen set have a lower bound. To see that it cannot be written as the union of countably many filters, for any filter $G$ define $$F_G=\Union\{F:\exists C\; (F,C)\in G\}$$ and $$C_G=\Intersection\{C:\exists F\; (F,C)\in G\}.$$ Note that $F_G\subseteq C_G$ and either $F_G$ is finite or $C_G$ has measure zero. So in either case $F_G$ has measure zero. Consequently, given any family $\{G_n:n\in\omega\}$ of filters there exists $x\in\cantorspace\setminus \Union_{n\in\omega} F_{G_n}$. Since $(\cantorspace,\{x\})\in \Bbb P$ we have $\Bbb P\not=\Union_{n\in\omega}G_n$. \qed \section*{Strong first category} \ques (Galvin) Does every Sierpinski set have strong first category? \bigskip An uncountable set of reals is a Sierpinski set iff it meets every measure zero set in a countable set. A set has strong first category iff it can be translated away from every measure zero set. This problem remains open although there are some partial results. \begin{theorem} (Bartoszynski-Judah [1990]) (CH) Every Sierpinski set is the union of two sets of sets of strong first category. \end{theorem} \proof Let $X=\{x_\alpha:\alpha<\omega_1\}$ be any Sierpinski set. Construct $X_i^\alpha$ countable for $\alpha<\omega_1$ and $i<2$ so that \begin{itemize} \item $X_i^\alpha\subseteq X_i^\beta$ if $\alpha<\beta$, \item $ X_0^\alpha\intersect X_1^\alpha =\emptyset$, \item $ X_0^\alpha\union X_1^\alpha \subseteq X$, and \item $x_\alpha\in X_0^{\alpha+1}\union X_1^{\alpha+1}$. \end{itemize} Afterwards $X_i=\Union_{\alpha<\omega_1}X_i^\alpha$ will give us our partition $X=X_0\union X_1$ into sets of strong first category. Let $\langle G_\alpha,i_\alpha\rangle$ for $\alpha<\omega_1$ list all pairs $\langle G,i\rangle$ for $G$ a $G_\delta$-set of measure zero and $i\in \{0,1\}$. Let $X_i^0=\emptyset$ and for limit ordinals $\alpha$ let $X_i^\alpha=\Union_{\beta<\alpha}X_i^\beta$. At stage $\alpha+1$ proceed as follows. Let $i=i_\alpha$ and $G=G_\alpha$. First find $z$ such that $$(z+G)\intersect X^\alpha_{i}=\emptyset.$$ This is easy to do because the set $X^\alpha_{i}$ is countable, so any $z$ not in the measure zero set $X^\alpha_{i}- G$ will do. Now let \smallskip $X^{\alpha+1}_{i}= X^{\alpha}_{i}$ and let \smallskip $X^{\alpha+1}_{1-i}= X^{\alpha}_{1-i}\union ((z+G)\intersect X)\union\{x_\alpha\}$ (where $\{x_\alpha\}$ is added only if it is not in $X^{\alpha+1}_{i}$. \smallskip \noindent So $(z+G)\intersect X\subseteq X_{1-i}$ and therefore $(z+G)\intersect X_i=\emptyset$. \qed \ques It is also an open question of Galvin if the union of two strong first category sets must have strong first category. \bigskip \begin{theorem} (Bartoszynski-Judah [1990]) \label{BJ90} If ZFC is consistent, then so is ZFC + there exists a Sierpinski set + every Sierpinski set has strong first category. \end{theorem} This model is obtained by starting with a model of MA+$\neg$CH and adding $\omega_1$ random reals. Note that by a Lowenheim-Skolem argument, if there exists a Sierpinski set which fails to have strong first category, then there exists an inner model of CH in which there exists a Sierpinski set which fails to have strong first category. Woodin [to appear] has shown that $\Sigma^2_1$ statements are absolute between models of CH, assuming there is a Woodin cardinal. He has noted that the existence of a Sierpinski set which fails to have strong first category is a $\Sigma^2_1$ statement, consequently, it is unlikely to be independent of CH. Note that in the model of Theorem \ref{BJ90} CH fails. \begin{theorem} (Jasinski-Weiss [1991]) \label{JW99} If $S$ is a Sierpinski set and $\Union_{n<\omega}C_n$ is a union of compact sets of measure zero, then there exists $x$ such that $(x+\Union_{n<\omega}C_n)\intersect S=\emptyset$. \end{theorem} \proof In fact, $\{x: (x+\Union_{n<\omega}C_n)\intersect S=\emptyset\}$ is comeager. The following is copied from Miller [1990/1]. Let the $C_n$ be increasing and compact. Let $\Intersect_{n<\omega}U_n$ be decreasing, open, $\mu(U_n)< {1\over n}$, and $\rationals + \Union_{n<\omega}C_n\subseteq \Intersect_{n<\omega}U_n$. \medskip \noindent Claim. $Y=\{x:x+\Union_{n<\omega}C_n\subseteq \Intersect_{n<\omega}U_n\}$ is a dense $G_\delta$ and hence comeager. \proof Since $Y$ contains the rationals $\rationals$ it is enough to see that it is a $G_\delta$. But note that for each $n$ $$\{x:x+C_n\subseteq U_n\}$$ is open, since $C_n$ is compact and $U_n$ is open. Since $$Y=\{x:x+\Union_{n<\omega}C_n\subseteq \Intersect_{n<\omega}U_n\}= \Intersect_{n<\omega}\{x:x+C_n\subseteq U_n\}$$ it is $G_\delta$ and the Claim is proved. To prove the theorem let $X=\Intersect_{n<\omega}U_n\intersect S$ and let $Z=Y\setminus(X-\Union_{n<\omega}C_n)$. Since $X$ is countable and so $X-\Union_{n<\omega}C_n$ is meager, it is enough to see that $(z+\Union_{n<\omega}C_n)\intersect S=\emptyset$ for every $z\in Z$. But $z\in Y$ implies $z+\Union_{n<\omega}C_n\subseteq\Intersect_{n<\omega}U_n$ implies $(z+\Union_{n<\omega}C_n)\intersect S\subseteq X$ implies (if nonempty) that $z+c=x$ where $c\in \Union_{n<\omega}C_n$ and $x\in X$ so finally $z\in X-\Union_{n<\omega}C_n$. \qed The referee has informed me that Theorem \ref{JW99} was also proved by Tim Carlson. The following is an open question. \ques Is it consistent to have every strong measure zero set countable (Borel conjecture) and every first category zero set countable (Dual Borel conjecture)? \bigskip Carlson's theorem that in the Cohen real model the Dual Borel conjecture holds has been extended by Judah-Shelah [1989] to show that it is consistent with MA($\sigma$-centered) that the Dual Borel conjecture holds. This argument has been improved by Pawlikowski [1990] to show that any model obtained by the finite support iteration of $\sigma$-centered forcing the dual Borel conjecture holds. Judah-Shelah-Woodin [1990] have shown that the Borel conjecture is consistent with the continuum arbitrarily large. See Bartoszynski-Shelah [to appear] or Bartoszynski-Judah [to appear] for a correction to that argument. \section*{Q-sets and strong measure zero} A set of reals is a Q-set iff every subset is a relative $G_\delta$. A set of reals $X$ has strong measure zero iff for any sequence of positive reals $\langle \epsilon_n:n\in\omega\rangle$ the set $X$ can be covered by sequence of sets $\langle I_n:n\in\omega\rangle$ where each $I_n$ has diameter less than $\epsilon_n$. \ques (Fleissner) Does every Q-set have strong measure zero? \bigskip This question was answered by the following theorem. \begin{theorem} (Judah-Shelah [1988]) If we add one Mathias or Laver real $f$ to a model $M$ of MA+$\neg$CH, then in $M[f]$ every uncountable set of reals $X\in M$ remains a Q-set and furthermore does not have strong measure zero. \end{theorem} Judah-Shelah [to appear] constructed a Q-set in a model $V[G]$ where $V$ is a model of GCH and the extension satisfies the Sack's property: for every $f\in\bairespace\intersect V[G]$ there exists $\langle H_n:n\in\omega\rangle\in V$ such that for every $n$ $f(n)\in H_n$ and $|H_n|\leq 2^n$. Strong measure zero sets are also denoted as $C$-sets. Rothberger defined $C^\prime$ and $C^{\prime \prime}$ as follows. A set of reals $X$ is $C^{\prime \prime}$ iff for any sequence $\langle {\cal U}_n : n\in\omega\rangle$ of open covers of $X$ there exists $\langle {U}_n : n\in\omega\rangle$ with each $U_n\in {\cal U}_n$ such that $X\subseteq\Union_{n\in\omega}{U}_n$. The definition of $C^\prime$ is the same except the open covers ${\cal U}_n$ are also required to be finite. \ques (Rothberger) Does every set with property $C^\prime$ have property $C^{\prime \prime}$? \bigskip This question is answered in the negative in Miller-Fremlin [1988] where it is shown that assuming CH that there is a $C^\prime$ set which is not $C^{\prime \prime}$. Another question about Q-sets was answered by Knight [to appear] where it is shown to be consistent to a have a $\Delta$-set which is not a Q-set. A $\Delta$-set is a set of reals $X$ with the property that for every sequence $X_n\subseteq X$ with the property that $\Intersection_{n\in\omega}X_n=\emptyset$ there exists open sets $U_n$ with $X_n\subseteq U_n$ and $(\Intersection_{n\in\omega} U_n) \intersect X=\emptyset$. For more about $\Delta$-sets see Tanaka [1986], [1990]. \section*{Hierarchy order} If $X$ is a separable metric space, then define $ord(X)$ (Borel rank or Baire order) to be the smallest $\alpha<\omega_1$ such that every Borel set in $X$ is $\Sigma^0_\alpha$ in $X$. A set $A$ is analytic in $X$ iff $A=\Union_{f\in\bairespace}\Intersect_{n\in\omega}B_{f\res n}$ for some Borel sets $B_{s}$ in $X$ for $s\in\seq$. \ques (Mauldin) Is it consistent to have a space $X$ where every $ord(X)<\omega_1$ but not every analytic set is Borel? \bigskip This question remains open. A related result is proved in Miller [1990a]: assuming CH there exists a separable metric space $X$ such that every analytic in $X$ set is Borel in $X$ but there exists a set analytic in ${X^2}$ set which is not Borel in $X^2$. This doesn't answer the question since the space $X^2$ has Borel subsets of arbitrarily large rank while $X$ has bounded Borel rank. The following is another open question along similar lines. \ques Is it consistent to have a space $X$ where $ord(X)>3$ but the difference hierarchy inside the $\Delta_3^0$ sets is bounded? \bigskip The paper Miller [1990a] contains some related results about projective hierarchies. \section*{Universal measurable sets} A set of reals is universally measurable iff it is measurable with respect to any measure on the real line. \ques (Mauldin) Is it consistent to have only $\continuum$ many universally measurable sets? \bigskip This question remains open. It is theorem of Laver that it is consistent that there are only $\continuum$ universal measurable zero sets. This holds, for example, in the random real model (see Miller [1983]) The following theorem of Grzegorek and Ryll-Nardzewski seems relevant. \begin{theorem} (Grzegorek, Ryll-Nardzewski [1980]) There are universally measurable sets which are not equal to a Borel set modulo a universal measure zero set. \end{theorem} \proof Let $W$ be the $\Pi^1_1$ set of well-orderings. Then $W$ is universally measurable. On the other hand if $B$ is any Borel set then $W\Delta B$ cannot be universally measurable. $$W\Delta B= (W\setminus B)\union (B\setminus W)$$ The set $B\setminus W$ is $\Sigma^1_1$, hence if it is uncountable it would contain a perfect set and not be universal measure zero. So without loss we may assume $B\subseteq W$. But then by the boundedness theorem it is easy to get a perfect subset of $W\setminus B$ and so it doesn't have universal measure zero. \qed \section*{Perfectly meager} A set of reals $P$ is perfect iff it is homeomorphic to $\cantorspace$. A set of reals $X$ is perfectly meager iff for every perfect set $P$ the set $X\intersect P$ is meager in $P$. \ques (Marczewski) Is the product of perfectly meager sets perfectly meager? \bigskip This was answered by Reclaw. \begin{theorem} \label{rec3} (Reclaw [to appear]) If there exists a Luzin set, for example assuming CH, then there exists $X$ and $Y$ are perfectly meager such that $X\cross Y$ is not perfectly meager. \end{theorem} \proof Let $\langle C_s:s\in\seq\rangle$ be a family of perfect subsets of $\cantorspace$ such that \begin{itemize} \item $C_{\langle\rangle}=\cantorspace$, \item for each $s\in\seq$ $\langle C_{s\concat n}:n\in\omega\rangle$ is a disjoint family of nowhere dense subsets of $C_s$, \item for each $s\in\seq$ every $t\in\binseq$ if $C_s\intersect [t]\not=\emptyset$, then for some $n\in\omega$ we have $C_{s\concat n}\subseteq [t]$. \end{itemize} The Luzin function $f:\bairespace\rightarrow \cantorspace$ is defined by $\{f(x)\}=\Intersect_{n\in\omega}C_{x\res n}$. We will use the following forcing Lemma about $f$. Let $\Bbb P=\seq$, that is Cohen forcing for $\bairespace$. \begin{lemma} \label{rec2} Let $(x,y)$ be the pair of Cohen reals added by forcing with $\Bbb P^2$. For every perfect $P\subseteq(\cantorspace)^2$ and $p\in\Bbb P$ there exists $q\leq p$ and $C\subseteq P$ closed nowhere dense in $P$ such that $q\forces$``$(x,f(y))\in C$ or $(x,f(y))\notin P$". \end{lemma} \proof Let $p=\langle p_o,p_1\rangle$. \medskip\noindent Case 1. $\{x\}\cross C_{p_1}\not\subseteq P$. Since $P$ is closed there exists $t\in\binseq$ such that $[t]\intersect C_{p_1}\not=\emptyset$ and $q_0\leq p_0$ such that $q_0\forces$``$(\{x\}\cross [t])\intersect P=\emptyset$''. For some $m\in\omega$ we have that $C_{p_1\concat m}\subseteq [t]$. Let $q_1=p_1\concat m$. Then $q\forces$``$(x,f(y))\notin P$''. \medskip\noindent Case 2. $\{x\}\cross C_{p_1}\subseteq P$. Let $q_0\subseteq x$ with $q_0\leq p_0$ such that $q_0\forces$``$\{x\}\cross C_{p_1}\subseteq P$''. Let $$H=\{u\in\cantorspace:\{u\}\cross C_{p_1}\subseteq P\}.$$ It is easy to see that $H$ is closed. Since there is a comeager in $[q_0]$ set of Cohen reals in $H\intersect [q_0]$ it must be that $[q_0]\subseteq H$ and thus $[q_0]\cross C_{p_1}\subseteq P$. If we let $q_1=p_1\concat n$ for an arbitrary $n$ and $C=[q_0]\cross C_{q_1}$, then $C\subseteq P$ is nowhere dense in $P$ and $q\forces$``$(x,f(y))\in C$''. \qed Finally we show how the lemma proves Theorem \ref{rec3}. Suppose $$K=\{(x_\alpha,y_\alpha) : \alpha<\kappa\}$$ is a Luzin set. This means that $\kappa$ is uncountable, but for every countable transitive model $M$ of a fragment of set theory, all but countably many of the $(x_\alpha,y_\alpha)$ are $\Bbb P^2$ generic over $M$. It follows from Lemma \ref{rec2} that $$K_1=\{(x_\alpha,f(y_\alpha)):\alpha<\kappa\}$$ is perfectly meager. To see this, let $P\subseteq(\cantorspace)^2$ be any perfect set. Let $M$ be a countable transitive model of a partial fragment of set theory which contains a code for $P$. Let $\{C_n:n\in\omega\}$ be all closed nowhere dense in $P$ sets which are coded in $M$. By Lemma \ref{rec2} for all but countably many $\alpha$ we have that $(x_\alpha,f(y_\alpha))\notin P$ or $(x_\alpha,f(y_\alpha))\in \Union_{n\in\omega}C_n$. Hence $K_1\intersect P$ is meager in $P$. Similarly $$K_2= \{(f(x_\alpha),y_\alpha):\alpha<\kappa\}$$ is perfectly meager. However consider $$H=\{\langle (x_\alpha,f(y_\alpha)),(y_\alpha,f(x_\alpha))\rangle :\alpha<\kappa\}.$$ $H\subseteq K_1\cross K_2$ and $H$ is homeomorphic to the Luzin set $K$ since it is essentially just the graph of the function $f\cross f:K\rightarrow\cantorspace$. It follows that $H$ is Luzin in its closure and so $K_1\cross K_2$ is not perfectly meager. \qed A generalization of this result appears in Pawlikowski [1989]. \section*{$\sigma$-sets} A set of reals is a $\sigma$-set iff every relative Borel set is a relative $F_\sigma$-set, i.e. $ord(X)=2$. \ques Can you map a $\sigma$-set continuously onto the reals? \bigskip This was answered by Reclaw. \begin{theorem} \label{recl2} (Reclaw) If a set of reals $X$ can be mapped continuously onto $\cantorspace$, then $ord(X)=\omega_1$. \end{theorem} \proof For a countable $H\subset P(Y)$ define the Borel hierarchy generated by $H$ as follows. Let $$\Sigma_0(H)=H$$ and for $\alpha<\omega_1$ let $$\Sigma_{\alpha}(H)=\{\union_{n\in\omega}(Y\setminus A_n): A_n\in\Sigma_{\beta_n}(H),\beta_n<\alpha\}.$$ Let Borel($H$) be the union of all $\Sigma_{\alpha}(H)$ for $\alpha<\omega_1$. We will need the following theorem which generalizes the classical theorem of Lebesgue. \begin{theorem} (Bing, Bledsoe, Mauldin [1974]) \label{BBM} Suppose $H\subseteq P(\cantorspace)$ is a countable family such that every clopen set is in Borel($H$). Then $ord(H)=\omega_1$. \end{theorem} \proof To prove this theorem we will need the following two lemmas. Given a countable $H\subseteq P(\cantorspace)$ let $$R=\{C\cross A:A\in H, C\subseteq \cantorspace\;\mbox{ is clopen}\}.$$ \begin{lemma} (Universal sets) \label{univ} For each $\alpha$ with $1\leq\alpha<\omega_1$ there exists $U\in\Sigma_\alpha(R)$ which is universal for $\Sigma_\alpha(H)$ sets, i.e., for every $A\in\Sigma_\alpha(H)$, there exists $x\in\cantorspace$ such that $A=\{y: (x,y)\in U\}$. \end{lemma} \proof For $\alpha=1$: Let $H=\{A_n:n\in\omega\}$ let $$U=\Union_{n\in\omega}\{x: x(n)=1\}\cross (\cantorspace\setminus A_n).$$ For $\alpha>1$: Let $x\mapsto \langle x_n:n\in\omega\rangle$ be a nice recursive coding taking $\cantorspace\rightarrow(\cantorspace)^\omega$. Let $\beta_n$ for $n\in\omega$ be cofinal in $\alpha$, and $U_n\in\Sigma_{\beta_n}(R)$ be universal for $\Sigma_{\beta_n}(H)$ sets. Define $U_n^{\prime}$ by $(x,y)\in U_n^{\prime}$ iff $(x_n,y)\in U_n$. It is easy to check that $U_n^\prime$ is also $\Sigma_{\beta_n}(R)$ and universal for $\Sigma_{\beta_n}(H)$. But now taking $$U=\Union_{n\in\omega}(\cantorspace\setminus U_n^\prime)$$ gives us a set in $\Sigma_\alpha(R)$ which is universal for $\Sigma_\alpha(H)$ sets. \qed \begin{lemma} (Diagonalization) \label{diag} Suppose that every clopen set is in Borel($H$). Then for every $B\in$Borel($R$) the set $\{x:(x,x)\in B\}$ is in Borel($H$). \end{lemma} \proof For $B=C\cross A$ where $A\in H, C\subseteq \cantorspace$ is clopen. Note that $\{x:(x,x)\in B\}=C\intersect A$. Since the by assumption $C\in$Borel($H$), we have the result for elements of $R$. To do Borel($R$) is an easy induction. \qed Now we give a proof of Theorem \ref{BBM}. Suppose Borel($H)=\Sigma_\alpha(H)$. By Lemma \ref{univ} there exist $U$ in Borel($R$) which is universal for $\Sigma_\alpha(H)$ and hence Borel($H$). By Lemma \ref{diag} the set $$D=\{x:(x,x)\notin U\}$$ is in Borel($H$). But this means that for some $x$ that $D=\{y:(x,y)\in U\}$. But then $x\in D$ iff $x\notin D$. \qed Now we give a proof of Theorem \ref{recl2}. Suppose that $f:X\rightarrow \cantorspace$ is onto and continuous. Let $\cal C$ be a a countable clopen basis for $X$ and let $H=\{f^{\prime\prime}C: C\in\cal C\}$. Since it is clear that $H$ contains all clopen sets, by Theorem \ref{BBM}, the $ord(H)=\omega_1$. But the map $f$ takes the Borel hierarchy of $X$ directly over the hierarchy on Borel($H$), so $ord(X)=\omega_1$. \qed An extension of Reclaw's result to Souslin (operation A) sets appears in Miller [1990a]. \section*{$\lambda$-sets and $\lambda^\prime$-sets} A set of reals $X$ is a $\lambda$-set iff every countable subset of $X$ is a relative $G_\delta$ in $X$. A set of reals $X\subseteq Y$ is a $\lambda^\prime$-set with respect to $Y$ iff for every countable $F\subseteq Y$ the set $X\union F$ is a $\lambda$-set. It is a theorem of ZFC that there are $\lambda^\prime$ sets of cardinality $\omega_1$. This section will clear up some confusion about a remark (p 266) in Reed [1983] which is attributed to me. The cardinal $\goth b$ is the minimum cardinality of a set $F\subseteq \bairespace$ such that $F$ is unbounded, i.e. there does not exist $g\in\bairespace$ such that $f\leq^* g$. In Reed [1983] this is referred to as the least $\alpha$ for which $M(\alpha)$ holds. The cardinal $\goth d$ is the minimum cardinality of a set $F\subseteq \bairespace$ such that $F$ is a dominating family, i.e., for every $g\in\bairespace$ there exists a $f\in F$ such that $g\leq^* f$. Obviously $\goth b\leq \goth d\leq \goth c$. \begin{theorem} (Rothberger [1939]) There exists a $\lambda$-set $X$ of cardinality $\goth b$, in fact, $X$ is a $\lambda^\prime$-set with respect to the irrationals. \end{theorem} \proof Let $F=\{f_\alpha:\alpha<\goth b\}\subseteq\bairespace$ be an unbounded family such that $\alpha\leq\beta$ implies $f_\alpha\leq^*f_\beta$. We claim that for every countable $H\subseteq\bairespace$ the set $F\union H$ is a $\lambda$-set. It is enough to see that for every countable $H\subseteq\bairespace$ $H$ is a relative $G_\delta$ in $F\union H$. Since $F$ is unbounded there exists $\alpha<\goth b$ such that for every $h\in H$ we have $f_{\alpha}\not\leq^* h$. This also holds for every $\beta>\alpha$ since $f_\beta\geq^* f_\alpha$. So if $$K=\{g\in\bairespace: f_\alpha(n)\not\leq^* g\}$$ then $K$ is a $G_\delta$ such that $H\subset K$ and $|K\intersect (X\union H)|<\goth b$. So it now suffices to note the following: \medskip\noindent Claim: Any set of reals $Y$ of cardinality less than $\goth b$ is a $\lambda$-set. \medskip\noindent Let $Y=\{x_n:n<\omega\}\union\{y_\alpha: \alpha<\kappa\}$. Let $\{U_n(x_m):n\in\omega\}$ be a decreasing neighborhood base for each $x_m$. For each $\alpha<\kappa$ define $g_\alpha\in\bairespace$ so that $y_\alpha\notin U_{g_\alpha(n)}(x_n)$. Since $\kappa<\goth b$ there exists $g\in\bairespace$ such that $g_\alpha\leq^*g$ for all $\alpha<\kappa$. Let $U=\Intersection_{n<\omega}(\Union_{m>n} U_{g(n)}(x_n)$. Then $U$ is a $G_\delta$-set such that $U\intersect Y=\{x_n:n\in\omega\}$. \qed \begin{theorem} (Miller [1983]) In the Cohen real model there are no $\lambda$-sets of cardinality greater than $\omega_1$. \end{theorem} \proof It is shown that every set of reals of cardinality $\omega_2$ contains the one-to-one continuous image of a Luzin set. But such a set cannot be a $\lambda$-set. If $L$ is Luzin and $D$ a countable dense subset of $L$, then $D$ cannot be relatively $G_\delta$ in $L$. But if $f:L\rightarrow X$ is one-to-one and continuous, then $f^{\prime\prime}D$ cannot be $G_\delta$ in $X$. \qed In the Cohen real model $\goth b=\omega_1$ and $\goth d=\goth c$. \begin{theorem} \label{lav} In the Laver model for the Borel conjecture ($\goth b=\goth d=\goth c=\omega_2$) there does not exist a $\lambda^\prime$-set with respect to the reals of cardinality $\omega_2$. \end{theorem} \proof Lemma 14 of Laver [1976] implies the following lemma. \begin{lemma} Suppose $p\forces$``$\tau\in [0,1]$'', then there exists $q\leq p$ and a finite set $U_s$ for each splitting node $s$ of $q(0)$, such that for all $\epsilon >0$ and all but finitely many n with $s\concat n\in q(0)$ $$q(0)_{s\concat n}\concat q\res[1,\omega_2) \forces \exists u\in U_s |u-\tau|<\epsilon$$ \end{lemma} \begin{lemma} \label{lav2} Given the above $q,\tau,$ and $U=\Union_{s\in q(0)} U_s$, for any $G_\delta$ set $G\supseteq U$ there exists $r\leq q$ such that $r\forces$``$\tau\in G$''. \end{lemma} \proof This is an easy fusion argument on $q(0)$. If $G=\Intersection_{n\in\omega}G_n$ where $G_n$ is open, then make sure that everything in the $n^{th}$ forces that $\tau\in G_n$. \qed Let $V[f_\beta:\beta<\omega_2]$ be the Laver model. To prove Theorem \ref{lav} suppose that $X\subseteq [0,1]$ is a $\lambda^\prime$-set in $V[f_\beta:\beta<\omega_2]$. Via a Lowenheim-Skolem type argument there exists $\alpha<\omega_2$ such that $$(X\intersect V[f_\beta:\beta<\alpha]) \in V[f_\beta:\beta<\alpha]$$ and a function $f\in V[f_\beta:\beta<\alpha]$ such that for every countable $D\subseteq [0,1]$ in $V[f_\beta:\beta<\alpha]$ we have $f(D)$ is a $G_\delta$ code for a set $G$ such that $$G\intersect (X\union D)=D.$$ But now Lemma \ref{lav2} (applied with $V[f_\beta:\beta<\alpha]$ as the ground model) implies that $X\subseteq V[f_\beta:\beta<\alpha]$. But since $V[f_\beta:\beta<\alpha]$ satisfies CH, the set $X$ has cardinality $\omega_1$. \qed \section*{Bibliography} This bibliography includes corrections from Miller [1984b] as well as related papers not mentioned in the body of this paper. \medskip Avraham, Uri; Shelah, Saharon, Martin's axiom does not imply that every two $\aleph_1$-dense sets of reals are isomorphic, Israel J. Math., 38 (1981), 161-176. \medskip Balogh, Zoltan; Junnila, Heikki, Totally analytic spaces under V=L, Proc. Amer. Math. Soc., 87 (1983), 519-527. \medskip Bartoszynski, Tomek; Judah, Haim, On Sierpinski sets, Proceedings of the American Mathematical Society, 108 (1990), 507-512. \medskip Bartoszynski, Tomek; Judah, Haim, Measure and category: the asymmetry, to appear. \medskip Bartoszynski, Tomek; Shelah, Saharon, Borel Conjecture and $2^{\aleph_0}>\aleph_2$, to appear. \medskip Baumgartner, James E., Iterated forcing. in { Surveys in set theory}, 1-59 London Math. Soc. Lecture Note Ser., 87, Cambridge Univ. Press, Cambridge-New York, 1983. \medskip Baumgartner, James E., Sacks forcing and the total failure of Martin's axiom, Topology and its Applications, 19 (1985), 211-225. \medskip Bing, R. H.; Bledsoe, W. W.; Mauldin, R. D., Sets generated by rectangles, Pacific Journal of Mathematics, 51 (1974), 27-36. \medskip Brown, Jack B.; Gardner, Richard J., Generalized Lusin sets and the Baire order problem, Bull. Acad. Polon. Sci. Ser. Sci. Math., 27 (1979), 745-750. \medskip Brown, J. B.; Cox, G. V., Continuous images of Lusin sets, Proc. Amer. Math. Soc., 89 (1983), 52-54. \medskip Brown, J. B.; Cox, G. V. Baire category in spaces of probability measures. II., Fundamenta Mathematicae, 121 (1984), 143 -148. \medskip Brown, Jack B., Singular sets and Baire order, Real Analysis Exchange, 10 (1984/85), 78-84. \medskip Brown, Jack B., Continuous restrictions of Marczewski measurable functions, Real Analysis Exchange, 11 (1985/86), 64-71. \medskip Brown, Jack B., Marczewski null sets and intermediate Baire order, in { Measure and measurable dynamics (Rochester, NY, 1987)}, 43-50, Contemp. Math., 94, Amer. Math. Soc., Providence, RI, 1989. \medskip Brown, Jack B.; Prikry, Karel, Variations on Lusin's theorem. Transactions of the American Mathematical Society, 302 (1987), 77-86. \medskip Brown, Jack B., Differentiable restrictions of real functions, Proceedings of the American Mathematical Society, 108 (1990), 391-398. \medskip Brown, Jack B., The Ramsey sets and related sigma algebras and ideals, to appear. \medskip Brzuchowski, Jan; Cichon, Jacek; Weglorz,Bogdan, Some applications of strong Lusin sets, Compositio Math., 43 (1981), 217-224. \medskip Cichon, Jacek, On Banach numbers, Bull. Acad. Polon. Sci. Ser. Sci. Math., 29 (1981), 531 - 534. \medskip Cichon, J.; Szymanski, A.; Weglorz, B., On intersections of sets of positive Lebesgue measure, Colloquium Mathematicum, 52 (1987), 173-174. \medskip Cichon,Jacek, On two-cardinal properties of ideals, Transactions of the American Mathematical Society, 314 (1989), 693-708. \medskip Corazza, Paul, The generalized Borel conjecture and strongly proper orders, Transactions of the American Mathematical Society, 316 (1989),115-140. \medskip Corazza, Paul, Ramsey sets, the Ramsey ideal, and other classes over $\reals^1$, to appear. \medskip Daniels, Peg; Gruenhage, Gary, The point-open type of subsets of the reals, Topology and its Applications, 37 (1990), 53-64. \medskip Davies, Roy O., Two remarks on the measure of product sets, in { Classical real analysis (Madison, Wis., 1982)}, 45-48, Contemp. Math., 42, Amer. Math. Soc., Providence, R.I., 1985. \medskip van Douwen, Eric K.; Fleissner, William G., Definable forcing axiom: an alternative to Martin's axiom, Topology and its Applications, 35 (1990), 277-289. \medskip Erdos, Paul; Kunen, Kenneth; Mauldin, R. Daniel Some additive properties of sets of real numbers, Fundamenta Mathematicae, 113 (1981),187-199. \medskip Fleissner, William G., Squares of $Q$ sets, Fundamenta Mathematicae, 118 (1983), 223-231. \medskip Fremlin, D. H.; Hansell, R. W.; Junnila, H. J. K., Borel functions of bounded class, Trans. Amer. Math. Soc., 277 (1983), 835-849. \medskip Fremlin, David H., { Consequences of Martin's axiom}, Cambridge Tracts in Mathematics, 84 Cambridge University Press, Cambridge-New York, (1984). \medskip Fremlin, D. H.; Jasinski, J., $G_\delta$-covers and large thin sets of reals, to appear. \medskip Galvin, Fred, Miller, Arnold W., $\gamma $-sets and other singular sets of real numbers, Topology and its Applications, 17 (1984), 145-155. \medskip Goldstern, M.; Judah, H., Iteration of Souslin forcing, projective measurability, and the Borel conjecture, to appear in the Israel Journal of Mathematics. \medskip Grzegorek, Edward; Ryll-Nardzewski, Czeslaw, A remark on absolutely measurable sets, Bull. Acad. Polon. Sci. Ser. Sci. Math., 28 (1980), 229-232. \medskip Grzegorek, Edward, Remarks on some Borel structures, in { Measure theory}, Oberwolfach 1983, 69-74, Lecture Notes in Math., 1089, Springer, Berlin-New York, (1984). \medskip Grzegorek, Edward, Symmetric sigma-fields of sets and universal null sets, in { Measure theory}, Oberwolfach 1983, 101-109, Lecture Notes in Math., 1089, Springer, Berlin-New York, 1984. \medskip Grzegorek, Edward, Always of the first category sets, Proceedings of the 12th winter school on abstract analysis (Srni, 1984). \medskip Grzegorek, Edward, Always of the first category sets. II., Proceedings of the 13th winter school on abstract analysis (Srni, 1985). Rendiconti del Circolo Matematico di Palermo. Serie II. Supplemento, (1985), 43-48. \medskip Jasinski, Jakub; Weiss, Tomasz, Sierpinski sets and strong first category, Proceedings of the American Mathematical Society, 111 (1991), 235-238. \medskip Judah, Haim, Strong measure zero sets and rapid filters, The Journal of Symbolic Logic, 53 (1988), 393-402. \medskip Judah, Haim,; Shelah, Saharon, Souslin forcing, The Journal of Symbolic Logic, 53 (1988), 1188-1207. \medskip Judah, Haim,; Shelah, Saharon, $Q$-sets do not necessarily have strong measure zero, Proceedings of the American Mathematical Society, 102 (1988), 681-683. \medskip Judah, Haim; Shelah, Saharon, ${\rm MA}(\sigma$-centered): Cohen reals, strong measure zero sets and strongly meager sets, Israel Journal of Mathematics, 68 (1989), 1-17. \medskip Judah, Haim; Shelah, Saharon; Woodin, W. H., The Borel conjecture, Annals of Pure and Applied Logic, 50 (1990), 255-269. \medskip Judah, Haim,; Shelah, Saharon, $Q$-sets, Sierpinski sets, and Rapid Filters, to appear in the Proceedings of the American Mathematica Society. \medskip Just, Winifred, Subsets of the plane of strong measure zero- variations on the theme, to appear. \medskip Knight, R. W., $\Delta$-sets, to appear. \medskip Kunen,Kenneth, Miller, Arnold W., Borel and projective sets from the point of view of compact sets, Mathematical Proceedings of the Cambridge Philosophical Society, 94 (1983), 399-409. \medskip Laver, Richard, On the consistency of Borel's conjecture, Acta. Math., 137 (1976), 151-169. \medskip Martin,D.; Solovay R., Internal Cohen extensions, Annals of Mathematical Logic, 2 (1970), 143-178. \medskip van Mill, Jan; Mogilski, Jerzy, Property C and fine homotopy equivalences, Proc. Amer. Math. Soc., 90 (1984), no. 1, 118-120. \medskip Miller, Arnold W., Some properties of measure and category, Trans. Amer. Math. Soc., 266, (1981), 93-114. Corrections and additions to: ''Some properties of measure and category'', Trans. Amer. Math. Soc., 271 (1982), 347 - 348. \medskip Miller, Arnold W., Mapping a set of reals onto the reals, J. Symbolic Logic 48, (1983), 575-584. \medskip Miller, Arnold W., Prikry, Karel When the continuum has cofinality $\omega\sb 1$, Pacific Journal of Mathematics, 115 (1984a), 399-407. \medskip Miller, Arnold W., Special subsets of the real line, in Handbook of set-theoretic topology, (1984b), 201-233, North-Holland, Amsterdam-New York. \medskip Miller, Arnold W., Rational perfect set forcing, in { Axiomatic set theory} (Boulder, Colo., 1983), 143--159, Contemp. Math., 31, (1984c) Amer. Math. Soc., Providence, R.I. \medskip Miller, Arnold W.; Fremlin, David H., On some properties of Hurewicz, Menger, and Rothberger, Fundamenta Mathematicae, 129 (1988), 17-33. \medskip Miller, Arnold W., Projective subsets of separable metric spaces, Annals of Pure and Applied Logic, 50 (1990a), 53-69. \medskip Miller, Arnold W., Set-theoretic properties of Loeb measure, The Journal of Symbolic Logic, 55 (1990b), 1022-1036. \medskip Miller, Arnold W., Abstract in Real Analysis Exchange, 16 (1990/1), 20. \medskip Natkaniec, Tomasz, On projections of planar sets, Real Analysis Exchange, 11 (1985/86), 411-415. \medskip Pawlikowski, Janusz, Products of perfectly meager sets and Lusin's function, Proceedings of the American Mathematical Society, 107 (1989), 811 -815. \medskip Pawlikowski, Janusz, Small subset of the plane which almost contains almost all Borel functions, Pacific Journal of Mathematics, 144 (1990), 155-160. \medskip Pawlikowski, Janusz, Finite support iteration and strong measure zero sets, The Journal of Symbolic Logic, 55 (1990), 674 -677. \medskip Pfeffer, Washek F.; Prikry, Karel, A topological concept of smallness, Proceedings of the National Academy of Sciences of the United States of America, 84 (1987), 3957-3958. \medskip Pfeffer, Washek F.; Prikry, Karel, Small spaces, Proceedings of the London Mathematical Society. Third Series, 58 (1989), 417-438. \medskip Plewik, Szymon, On universally measure zero and always first category sets, to appear. \medskip Przymusinski, Teodor C., On the equivalence of certain set-theoretic and topological statements, in Collection: Topology, Vol. II (Proc. Fourth Colloq., Budapest, 1978), North-Holland, Amsterdam, (1980), pp. 999 - 1003, Colloq. Math. Soc. Janos Bolyai, 23. \medskip Przymusinski, Teodor C., The existence of Q-sets is equivalent to the existence of strong Q-sets, Proc. Amer. Math. Soc. 79 (1980), 626-628. \medskip Rothberger, F., Sur un ensemble de premiere categorie qui est depourvu de la propriete $\lambda$, Fundamenta Mathematicae, 32 (1939), 50-55. \medskip Reclaw, Ireneusz, A note on the $\sigma$-ideal of $\sigma$-porous sets, Real Analysis Exchange, 12 (1986/87), 455-457. \medskip Reclaw, Ireneusz, On small sets in the sense of measure and category, Fundamenta Mathematicae, 133 (1989), 255-260. \medskip Reclaw, Ireneusz, Products of perfectly meagre sets, to appear in Proceedings of the American Mathematical Society. \medskip Reclaw, Ireneusz, Some additivity properties of special sets of reals, to appear. \medskip Reed, G. M. , Collectionwise Hausdorff versus collectionwise normal with respect to compact sets, Topology and Its Applications, 16 (1983), 259-272. \medskip Seredynski, Witold, Some operations related with translation, Colloquium Mathematicum, 57 (1989), 203-219. \medskip Shortt, Rae M., Universally measurable spaces: an invariance theorem and diverse characterizations, Fundamenta Mathematicae, 121 (1984), 169-176. \medskip Shortt, Rae M., Borel density, the marginal problem and isomorphism types of analytic sets, Pacific Journal of Mathematics, 113 (1984), 183-200. \medskip Shortt, R. M.; Rao, K. P. S. Bhaskara, Blackwell spaces and generalized Lusin sets, Real Analysis Exchange, 10 (1984/85), 163-179. \medskip Shortt, R. M.; Rao, K. P. S. Bhaskara, Generalized Lusin sets with the Blackwell property, Fundamenta Mathematicae, 127 (1987), 9-39. \medskip Shortt, Rae Michael, Sets with no uncountable Blackwell subsets, Czechoslovak Mathematical Journal, 37(112) (1987), 320-322. \medskip Stepanek, Petr, Souslin's hypothesis, Borel's conjecture and the inner model HOD, Bull. Acad. Polon. Sci. Ser. Sci. Math., 29 (1981), no. 5-6, 193 - 197. \medskip Steprans, Juris, Cardinal arithmetic and $\aleph_1$-Borel sets, Proc. Amer. Math. Soc., 84 (1982), 121-126. \medskip Tanaka, Hidenori, Normality and hereditary countable paracompactness of Pixley-Roy hyperspaces. Fundamenta Mathematicae, 126 (1986), 201-208. \medskip Tanaka, Hidenori, Countable paracompactness of Pixley-Roy hyperspaces, Proceedings of the American Mathematical Society, 108 (1990), 1115-1120. \medskip Todorcevic, S., Remarks on Martin's axiom and the continuum hypothesis, Canadian Journal of Mathematics, 43 (1991), 832-851. \medskip Woodin, H., to appear. \bigskip \begin{flushright} Department of Mathematics \\ University of Wisconsin \\ 480 Lincoln Drive \\ Madison, WI 53706, USA \\ miller@math.wisc.edu. \\ \end{flushright} \end{document}