This is the differential equation the Legendre polynomials satistky; you may use the recursive relation, known as Bonnet's recursion formula, in order to recreate them:
$(n+1)P_{n+1}(x) = (2n+1)xP_n(x) - nP_{n-1}(x)$
You can use this recursion in order to get them, as long as you get the first 2 terms: you can find them using explicitly the differential equation:
For $n=0$, the equation becomes $\frac{d}{dx}\Big[ (1-x^2) \frac{d}{dx}P_0(x) \Big] = 0$ This clearly has $P_0(x)=$ constant as a solution, and if you use the normalization
$P_n(1)=1$ for every $n$
you get $P_0(x)=1$.
For $n=1$, the equation becomes $\frac{d}{dx}\Big[ (1-x^2) \frac{d}{dx}P_1(x) \Big] +2P_1(x)= 0$. This can seem a little messy: let's just try polynomials of the simplest form (non-constant), that is, of the form $Ax+B$ (linear); we get
$-2Ax + 2Ax + 2B=0$
which clearly tells us that the polynomial $P_1(x) = Ax$ is a solution; using again the normalization condition, we get $P_1(x) = x$.
From now on, is trivial to get the other ones, using the above recursion formula: for example for the 2nd polynomial we have
$2P_2(x) = 3xP_1(x)-P_0(x) = 3x^2 - 1$
$P_n(x) = \frac{1}{2^n n!} \frac{d^n}{dx^n} \Big[(1-x^2)^n \Big]$, I have given the recursive equation because I read you wanted to recreat all the polynomials one by one, from 0 to $n=4 or 5...$
– Klaramun Nov 05 '14 at 20:01$P_n(x) = \sum_{k=0}^{+\infty} a_k(n) x^k$
and incert this into the differential equation: you will get a recurrence relation for the coefficients.
You do not need to worry about the convergence, this is only a trick: you can start by assuming $|x| < 1$, using the method and then try to find a "cute" expression for the $P_n(x)'s$; by obtaining these polynomials you only need to check that these are indeed solutions, and you are done. (The normalization condition will be still $P_n(1)=1$, it's the standard one)
– Klaramun Nov 05 '14 at 20:30