I'm glad it has been solved using the page method.

I've been confronted a couple of times with this issue and I've been able to filter the troublesome rows using this kind of query construct:

select "all columns except the troubled one"

-- into save_data_table

from the_table

where isnumeric(isnull(troubled_column,0)) = 0

Until now, I've been lucky every time only a single column of the row had purity issues.

As Gail stated, you need to investigate the rows and determine its influence at application-data level.

Retain the row and correct the column value if you can, else ... remove the row.

Test it before your delete any rows !!

Off course, after fixing the rows, re-run your data_purity check !