N Hisograms of Random numbers to aproximate Gaussian distribution

How to can I make a fuction that for a parameter N draws N histograms of some random numbers (in interval -1 to 1) generated by random.random() genrator in numpy so that the first graph contins only xi random numbers, the second xi+xj and the N-th histogram is a histogram of xi+xj+...xN and I should also show that for big N the histogram will look similar to the gaussian distiribution, how do I do this?