Allin, correct dataset is here: http://eubie.sweb.cz/gretl_forum/allrng_allin.gdt

The first 22 missing observation were removed in HAR modeling. 

Take care, 

Daniel