MTB > # Gelman et al example - page 47 
MTB > # estimating an unknown normal variance from football
MTB > # spread data

MTB > name c1 'chisqs' c2 'sigma2' c3 'sigma'
MTB > rand 1000 'chisqs';
SUBC> chisq 672.
MTB > let 'sigma2'=672*13.85**2/'chisqs'
MTB > let 'sigma'=sqrt('sigma2')
MTB > dotplot 'sigma'

Each dot represents 4 points
                                     .
                                ..: .: :
                            :  ::::::::: :
                            ::.::::::::::::.
                            ::::::::::::::::..
                          .:::::::::::::::::::.: .
                        .:::::::::::::::::::::::::: .
               .  ..:.:::::::::::::::::::::::::::::::..:.    .
         +---------+---------+---------+---------+---------+-------sigma   
     12.50     13.00     13.50     14.00     14.50     15.00

MTB > notitle
MTB > median 'sigma'
   Median of sigma   =      13.861
MTB > sort 'sigma' 'sigma'
MTB > let k1='sigma'(25)
MTB > let k2='sigma'(975)
MTB > prin k1 k2
K1       13.1937
K2       14.6357