True and false. You don't need anywhere near a million samples to get a good approximation for your normal distribution. Far fewer than 100 is sufficient (and 14 is a fine place to start if you are really constrained on data and need to get to 90-10).
