This task builds on the tower representation which Don Steward utilises in some of his tasks on mean.  

This representation is used here to help pupils develop a sense of spread.  Later, pupils have to use their understanding to generate sets of data to satisfy the given criteria.  

Further, there is careful variation of the questions throughout with, mainly, one one attribute changing between each question.  This allows the effect on changes in either mean or standard deviation to be compared while the other remains invariant.  

Credit: @Chrismcgrane84 with thanks to @garyl82 for some suggestions.