Fw: Mail delivery failed: returning message to sender

["Helmut" <helmut@cam.wits.ac.za>]

>>
>>I figure the variance should consist of terms from the following
>>product:
>>
>>/     \       /       \
>>| m^3 |       | H_m^2 |
>>| m^2 |   x   |  H_m  |
>>|  m  |       |   1   |
>>|  1  |       |       |
>>\     /       \       /
>>
>>and be symmetric. It might be possible to compute the coefficients by
>>starting with the coefficient of j^3 and (n-j+1)^3 and working one's
>>way down to the lower-order terms. I'm sure the variance can be proved 
>
>
>
>Judging from the number of comparisons, this is NOT true, and
>a lot more of unpleasant terms appear. 
>
>Regards, Helmut Prodinger
>
>