T-test to compare two means with only means and n known

#1
Hi statistics geniuses! I've been looking around not being able to find out how, if at all, it is possible to test whether two means are different from each other. I only know the two means and the n for each of the two samples... How can you do that?
 
#3
Dason is right, without any information about the variability (i.e., standard deviation or standard error) in each group this is not possible. Also, are these two groups independent or dependent (matched)?