I am not a Twain scholar (my field is CS and Stat) but I think Dagnall's
point is that before someone asserts a new-found "truth" there ought to
be more going for the author than "maybe", "perhaps", "seems", etc. I know
in stat words like that just don't belong in a paper except in describing
the situation before your paper or in the section on future directions.
Dave Gomberg