See [[Jech - Set theory_ The third millennium edition]] pg. 557.
**Theorem (Jensen).** There is a generic extension $L[a]$ of $L$ such that a is a $\Delta_3^1$ real.
>[!note] Remark
>By [[Shoenfield's Absoluteness Theorem]], every $\Pi_2^1$ or $\Sigma_2^1$ real is constructible; on the other hand $0^{\sharp}$ is a $\Delta_3^1$ real, but it cannot be obtained by forcing over $L$.
# Definition
Recall the definition of [[Sacks forcing|forcing with perfect trees]].
$\mathbb{P}$ is a set of perfect trees, ordered by $\subseteq$, which is carefully constructed using the $\diamondsuit$-[[Diamond|principle]] in $L$.
A generic $G\subseteq \mathbb{P}$ corresponds to a branch $a\in 2^{\omega}$ such that in $L[a]$:
- $a$ is the unique $\mathbb{P}$-generic
- $\{ n<\omega \mid a(n)=1\}$ is a $\Delta_{3}^{1}$ definable set.
%% , and ![[Sacks forcing#^stem|Sacks real]]
>[!info] Definition
> Let $T=$ $\{T(s): s \in \operatorname{Seq}(\{0,1\})\}$ be a collection of perfect trees. We say that $T$ is *fusionable* if that for every $s$,
> 1. $T(s)$ is a perfect tree whose stem has length $\geq \operatorname{length}(s)$.
> 2. $T\left(s^{\frown} 0\right) \subset T(s)$ and $T\left(s^{\frown} 1\right) \subset T(s)$.
> 3. $T(s^\frown 0)$ and $T\left(s^{\frown} 1\right)$ have incompatible stems.
> and we let the *fusion of $T$* be
> $
> \mathcal{F}(T)=\bigcap_{n=0}^{\infty} \bigcup_{s \in\{0,1\}^n} T(s) .
> $
For each fusionable $T, p=\mathcal{F}(T)$ is a perfect tree; and for each $s$, if $t$ is the stem of $p_s=T(s)$, then $p \upharpoonright t$ is stronger than both $p$ and $p_s$. %%
# Properties
- $\mathbb{P}$ and $\mathbb{P}\times\mathbb{P}$ satisfy [[Chain conditions#^ccc|ccc]]