A stratified sample divides the population into strata and does a proportional random sample in each stratum.
In this example, each column is a stratum. It samples two gems from each column.
Strata are usually divided by population groups.
Stratified sampling is used to ensure a representative sample.
The sample size is n=16. This choice was arbitrary to make some of the other examples easier.
It is possible to pick the same gem more than once in each column.
If you click "Resample" several times, you will see the proportions change. This is sampling error and is part of the sampling process. We will spend time later talking about how to deal with sampling error.