If we have a population having a very large number of elements and a large number of distinct values, we could visualize drawing its histogram with a large number of intervals and then drawing a smooth curve through the tops of the bars in the histogram. This smooth curve is the distribution of the population.
An example of this is the population of the heights of all the people in the U.S. If we had a very precise measuring instrument, then there would be a large number of distinct heights and lots of people of each height. If we could actually construct a histogram of the whole population using a large number of intervals, the resulting distribution would be bell shaped where the bell would be centered at the average height.
Another example of the bell shaped curve is the population of all
IQ's in the US which scientists have determined are well modeled
by a bell shaped curve with average value approximately equal
to 100 and standard deviation (see the next section) approximately
15. In the figure above we have drawn the histogram of a
sample of 500 IQ's. Note that it is centered at the average IQ value and most
IQ's are within two standard deviations (2 
15 = 30) of the mean.
If a population is discrete, that is, it has only a few distinct values (for example, only 0's and 1's), we could draw a bar graph with a bar for each distinct value and the height of the bar equal to the proportion of elements in the population having that value. An example of this is if repeatedly flipped a coin 10 times, each time recording the number of heads we obtained. The resulting population would only have 11 distinct values (0 heads through 10 heads) but if we did the 10 flips millions of times, there would be millions of values in the population. Using probability ideas, it can be shown that the resulting distribution would look like: