The rationale behind extended norms is that there is as much diversity in the top 1% of the population as there is between the 50th %ile and the 99th %ile, but much less data--and ExNorms are an attempt to "spread the curve" in that last percentile (or fraction thereof).
Do we have enough data to make meaningful distinctions (from 108 students)? Well, more data would, of course, be better, but 108 is certainly a whole lot better than what exists in the standard norms. There were 2,200 students in 11 age groups in the original standardization pool, which means about 200 students per age bracket--which you could think of as about three students-worth in the entire pool, at all age levels, representing everyone of IQ 145+ (or, if you like, 0.07 of a person representing everyone in the data set of 160+). I think it's fair to say that 108 is a much better sample size than 3 (let alone 0.07).
We readily acknowledge that there is a significant functional difference (including in educational needs) between someone of IQ 100 and someone of IQ 130. The ExNorms are in service to the idea that a similar functional difference exists between the individuals of IQ 160 and 190. (BTW, having assessed individuals with IQs below the floor of 40, at the opposite extreme, I can attest to analogously significant functional differences between someone assessed in the 40s and someone who is not formally assessable, such as adolescents for whom one can obtain standardized-ish data only using infant measures.)
