Just about every platform in JMP includes an optional analysis role called the BY role. You can use this when you want to obtain separate analyses or graphs for every level or group in a column. For example, suppose you're working with the Big Class data table, and you want a separate bivariate fit of weight by height for each of the different age groups. When you put age in the BY role, the launch produces a window for a collection of platforms, one platform for each unique value of age. All of the individual platforms -- the bivariate fits, in this case -- appear in the same window. When you use a BY variable, a list of object references is returned, one for each platform in the window. In this example, there are six levels of age, so what's returned is a list of six bivariate objects, and these references are stored in the variable bivGroup. You can send a message to all of the platforms by sending the message to the list stored in bivGroup, which then distributes the message to each platform. Or, you can send a message to just one platform, or a specific subset of the platforms, by using subscripts.
