TexasCHLforum.com

This idea involves a bit of forensic metallurgy and less "gritty street smarts", so if that's not what you need then don't bother reading below.

When a round is fired, the brass stretches and then mostly springs back to its former shape; this allows it to seal ("obturate") the flow of propellant gas out of the breech of the gun. When it is sized in a forming die during reloading, the brass is stretched again, formed back into its original shape to be reused. Brass undergoes a phenomenon called "work hardening" wherein as the metal is formed,it becomes more brittle and less ductile. Over a sufficient number of cycles, which varies based upon the stresses to which the case is subjected, the case neck will become too brittle and crack. This is never a problem for the first 2-3 firings, but after that point the number varies greatly by cartridge.

A reloaded case would have a measurably harder case neck versus one that had only been fired once. You'd need to test how brittle the case neck of the case from evidence is compared with other once fired cases of the same type AND reloaded cases of the same type. A pair of averages would emerge between the control cases, but I am not sure how different the means and SDs would be. By comparing the ductility of the case from evidence to those means, you could determine if it had been reloaded. A lot of range ammunition is made with reloaded cases which are not annealed (see below), so it would be imperative to be able to identify and test the same sort of ammunition from the factory. Given the ubiquity of military brass, the shell would need a rather unique headstamp (i.e. the markings on the rear of the case, which are manufacturer specific) to find a comparison sample.

There is a technique called annealing which returns the case neck to its normal, ductile state. It is commonly done for high end target reloading, but not for garden variety range or even hunting ammo. A machine to quickly do a lot of cases consistently is around $500 and up to thousands of dollars. If the suspect had no such machine, it would be very unlikely that he annealed cases to give to friends (given how laborious that process is), which would make the above test accurate.