It is testable and therefore is empirical science. For example a human skeleton found in the cambrian would falsify the explaination that humans arose through minor variations over millions of years.
If I claim that the current through a resistor will be proportional to the voltage, not only is it possible that such claim might be disproven if false, it is basically certain. While it is true that many of the claims posed by the historical sciences might turn out to be provably false, there remains the very real possibility that they might be false but in such fashion as can never be proven.
Suppose I hand you a box of cross-cut shredded paper and tell you that it contains half of the shreddings from an accurate copy of the "novel" Gatsby (which is notable for its non-use of the letter "e"). If examination of the shreddings reveals that some of the pieces contain the letter "e", that would constitute proof that the shreddings are not what I claim. If examination of the shreddings reveals that none of the pieces contain the letter "t", that would also disprove my claim. On the other hand, no amount of analysis would be able to prove my claim to be true, and it is possible that my claim could be false and yet no amount of analysis would be able to disprove it.
When I am saying that Evolution is not an empirical science I am not discussing the sequence of species found in the fossil record. Rather I am most specifically referring to the naturalistic explanation of how and why this came about. It is this part of the theory that is untestable and therefore not an empirical science. This does not mean that it is not a proper science, only that it is a speculative one, not an empirical one. Speculation and conjecture are a necessary part of the scientific endeavor. Indeed, without them science could not advance. Natural Evolution should just be honest about the limits of its study and the speculative nature of its conclusions.