When you specify a categorisation variable, the default is to create the split-out columns as the alphabetical list of the unique items in the category variable. You can over-ride this (to change the ordering or use only a subset of the available data) by setting an explicit list of values to be split into.

chPlot yield by farm into 'University Farm' 'Waseca' 'Morris'
View chClose 

See also ...
CategoriseBy – category variable