I think your grading criteria need a little nudging. We should grade based upon what the intent of the exercise is -- in this case, the ability to convey and support an idea. This includes the elements of style (spelling, punctuation, grammar, etc.).
For example, if you took a written exam and made 15 punctuation errors, you should be ineligible to receive an 'A' (95%). That level of error suggests the writer does not understand how to use punctuation, which is a large barrier to ensuring your readers understand your points.
Spelling is less critical than punctuation -- misspellings do not generally obscure meaning.
I think the scoring should look something like this (on a 100-point scale, for a simple thesis paper ):
Thesis: 15 points maximum; points awarded for clarity in expression (style impacts clarity).
Supporting arguments: 60 points maximum; evaluated on logic, factual accuracy, clarity of expression, and significance to the thesis.
Summation: 25 points maximum; scoring based on clarity of expression of the thesis and consistency with the supporting arguments.
This grading scale would be for a class teaching composition, of course -- nowhere is mastery of subject knowledge evident in the grading. The reason we write theses and other essays is to communicate ideas and information. The scoring must therefore reflect how well the student communicated; style is sometimes critical to the communication of ideas.