Problem 116
Medium Conditional Probability and Independence

A box contains two balls, one white and one black. Balls are…

A box contains two balls, one white and one black. Balls are drawn one at a time until the black ball appears; each time a white ball is drawn, it is returned to the box together with two more white balls. Find the probability that the black ball is not drawn in the first $50$ trials.

Explanation of Method

The composition of the box changes after every white draw, so the draws are dependent. We track the number of white balls before each trial: $1,3,5,\dots$ The probability that all first $50$ draws are white is a product of conditional probabilities (multiplication rule). The product of fractions $\frac{2k-1}{2k}$ can be written with a binomial coefficient and estimated with Stirling's formula.

Step 1: The box before each trial

At the start: $1$ white, $1$ black. After each white draw the white ball is returned and $2$ more are added, so the number of white balls grows by $2$. Given that trials $1,\dots,k-1$ were all white, before trial $k$ the box contains

\[ 2k-1 \text{ white balls and } 1 \text{ black ball}. \]

Trial $k$White balls before the trial$P(\text{white}\mid\text{earlier whites})$
$1$$1$$\frac{1}{2}$
$2$$3$$\frac{3}{4}$
$3$$5$$\frac{5}{6}$
$4$$7$$\frac{7}{8}$
$k$$2k-1$$\frac{2k-1}{2k}$

Step 2: Multiplication rule

Let $W_k$ = "the $k$-th ball drawn is white". Then

\[ P(W_1W_2\cdots W_{50}) = \prod_{k=1}^{50}\frac{2k-1}{2k} = \frac{1\cdot3\cdot5\cdots99}{2\cdot4\cdot6\cdots100}. \]

Step 3: A closed form

Multiply numerator and denominator by $2\cdot4\cdots100=2^{50}\cdot50!$:

\[ \frac{1\cdot3\cdots99}{2\cdot4\cdots100} = \frac{100!}{(2^{50}\,50!)^2} = \frac{1}{2^{100}}\binom{100}{50}. \]

\[ \boxed{P = \frac{1}{2^{100}}\binom{100}{50}\approx0.0796.} \]

Step 4: Stirling's approximation

For large $n$, $\binom{2n}{n}\approx\dfrac{4^n}{\sqrt{\pi n}}$. With $n=50$:

\[ P\approx\frac1{\sqrt{50\pi}}\approx0.0798, \]

very close to the exact value.

Step 5: Interpretation

At first it seems that the black ball should appear quickly, because white balls "flood" the box. But the probability of surviving $n$ trials decreases only like $\frac1{\sqrt{\pi n}}$: after $50$ trials, the black ball has still not appeared in about $8\%$ of cases.


Result

Answer
Product$\prod_{k=1}^{50}\frac{2k-1}{2k}$
Probability$\binom{100}{50}/2^{100}\approx0.0796$