# A question on an Erdős proof

I am currently going through a proof by Erdős and I am having difficulty understanding one of his arguments. He first gives the following Lemma:

If ${n}\choose{k}$ is divisible by a prime power $p^a$, then $p^a \leq n$.

Now he says this:

Let $\pi(k)$ denote the number of primes less than or equal to $k$. It is clear that for $k \geq 8$, $\pi(k) \leq k/2$. Hence, if ${n}\choose{k}$ does not have prime factors greater than $k$, we should have, from the lemma, ${n}\choose{k}$ $\leq n^{k/2}$.

I do not follow this argument. Any help would be greatly appreciated.

The first statement (that $\pi(k) \leq \frac{k}{2}$) comes from the fact that prime numbers (except 2) are odd.
Now suppose that $\binom{n}{k}$ has only prime factors $\leq k$. There are at most $\frac{k}{2}$ such factors by the above statement. On the other hand, by the lemma, for each such prime factor $p$, $p^a \leq n$ where $a$ is largest such that $p^a | \binom{n}{k}$.
Therefore, $\displaystyle \binom{n}{k} = \prod_{i=1}^m p_i^{a_i} \leq \prod_{i=1}^m n = n^m \leq n^{\frac{k}{2}}$, where $\{p_i\}_{1\leq i\leq m}$ are the prime factors of $\binom{n}{k}$ and $a_i$ are the powers to which they appear in its prime factorization.
Suppose that $\binom{n}k$ has no prime factors greater than $k$, where $k\ge 8$; $\pi(k)\le\frac{k}2$, so $\binom{n}k$ has at most $\frac{k}2$ prime factors. Let these prime factors be $p_1,\ldots,p_m$, where $m\le\frac{k}2$, and for $i=1,\ldots,m$ let $p_i^{a_i}$ be the highest power of $p_i$ dividing $\binom{n}k$. Then by the lemma $p_i^{a_i}\le n$ for $i=1,\ldots,m$, so
$$\binom{n}k=\prod_{i=1}^mp_i^{a_i}\le n^m\le n^{k/2}\;$$