I did not figure out how to get sub-millisecond-precision timing data from 
MIT/GNU Scheme.

With millisecond timing data, I needed to user larger numbers to get meaningful 
results.

However, around 10¹² to 10¹⁶ this starts to give meaningful results, which do 
indeed bear out the timing prediction that each additional factor of ten runs 
about three times as long.

I also changed the functions provided, I'm not sure how one was expected to 
write search-for-primes using the provided functions which return no value and 
only dump output to the console...

(define (start-prime-test n start-time)
  (if (prime? n)
      (report-prime (- (process-time-clock) start-time))
      #f))

(define (report-prime elapsed-time)
  (display " *** ")
  (display elapsed-time)
  #t)

(define (search-for-primes n)
  (if (timed-prime-test n)
    n
    (search-for-primes (+ n 1))))