Probability Problem - How many events needed?

I am trying to figure out how to calculate how many samples of a population I would need to take in order to get a true representation of the total population. For example, say I have a population of balls made up of 10 colors and each color is equal to 10% of the total population. How many balls at random would need to be drawn from a container containing the 'population' of balls to obtain a good representation of the total population (ie at a 90% confidence interval or 95% confidence interval)? Hopefully that makes some sense. I know it's probably not too difficult of a problem, but my stats background is pretty weak. Thanks for any help.

edit - After thinking a little more, the total population size is probably important. For the above, lets assume there are 1000 balls that make up the population in the container.

The usual rule is n = (t^2)*(s^2) / (d^2)

You need to pick or calculate t (score based on error and sample size - often takes as '2'), s (s^2 is the population variance, estimate if nexessary), and d (acceptable error) - roughly speaking.

Next, if your sample size exceeds maybe 5% of your population, you may need to consider a Finite Population Correction Factor.

Let's see your work.