Understanding significance in science

Consider a fixed die such that on average after 61 rolls the outcomes are 11 sixes and 10 of the remaining integers one through five. How many rolls would it take to determine with 95 % certainty that the die was biased? Meta-analysis studies (who get a final p value less than 0.05 ie significant)often take smaller studies with a p value greater than 0.05 (ie insignificant) and combine them to get a to big enough sample to determine significance. Due to picking one study and excluding another, one could get different answers to the same question. The cochrane collaboration tries to set criteria for w/c studies are to be included before performing the meta-analysis thereby avoiding some bias.