No matter what method you use, to have them graded as such, they have to pass the standards of the TPG. What is that (are those) method (s)?
I personally am not sure. I've heard many thing from others but I've never really felt that i've gotten a consensus. Keep in mind this is essentially just to tell you if they absolutely are not.
Update: I purchased a MS 64 PL to compare to my other coins. As you can see the outlines of both coins devices are very clear. But when comparing the field to device difference in brightness the second coin is a little bit more pronounced. The fields are brighter and the devices are darker. I also noted that all devices are seen in the PL coin. I feel this test can fairly accurately determine if a coin is proof like if it meets 4 criteria. 1. All (95%?) devices on the coin are clearly distinguishable in the reflection. 2. There is a noticable difference in the darkness of devices and the brightness of the fields in the reflection. 3. The fields are reflective to the edge of the coin showing a full circle in It's reflection. (For example the reverse of the first coin does not have a full circle of reflection, it is weak on the bottom side) 4. The first three criteria must be met on both obverse and reverse.