ISUCKATTHIS, can you please provide an affirmative proposal for what you think should happen?
That's a much harder question and I am not qualified to address it. I'll take a stab anyway.
In order to answer this question comprehensively, you'd have to know how many students were affected.
Probably the best solution, if it is possible, is to curve separately those students taking the test under different conditions. That would mean having a separate curve for all those students whose exams were misformatted. The obvious problem with this idea is that there may be an insufficient number of such students to form a normal distribution.
Another, less fair solution would be to re-weight the questions for all or a portion of the students according to their exam's format. Students with the proper format would get the exams with the original weighting scheme, students with an improper format would get a weighting scheme that compensated for their misallocation of time during the exam. This would mean de-emphasizing the latter two questions in the weighting scheme.
Finally, if all else fails, one or all of the following imperfect solutions could be applied: 1) the formatting and/or organization of essay answers to the latter two questions could be officially disregarded by the administration in the grading process (i.e., grading for content only), 2) students with misformatted exams could be given compensatory points or 3) the last two questions could be disregarded in the grading process.