But the last figure must be influenced somehow. As i understand the figure
is all translated packets through time? (but perhaps I'm wrong on this one).
And as packets translated before 2006 is no longer in the data set, it must
end up as a lower number?
That number is just the result of a "select count(*) from table where
language" query, so that number won't change. The numbers from before
2006 are from an unknown script that we don't know what it did.
Essentially it will look like in 2006 lots of translations came out of
nowhere, which is what happened since there was an import of the old
translations.