The only way this could work is a resisitance measurement (electrical conductivity) of an item. Gold conducts electricity with little resistance. The purer the gold the less resistance it would have had. Has to be a pretty sensitive bridge balance to detect minute differences in conductivity in gold thats pre 24K to say 10K gold that has a mix of metal in it...
Gold was often cut with silver, copper, etc. All these "impurities" are also good conductors as well....
I would be highly skeptical of a device like this.
What's the brand name??
Here's a recent writeup
This tester is not reliable. It gives false positives and cannot distinguish gold-filled from solid karat gold. It gave a reading of 721 on a US gold eagle coin and a reading of 945 on a sterling ring electroplated in 18k gold. Out of three readings on the same piece of plated Chinese fake "10k white gold" it read 435, then 509, then 415. When I put 18k testing solution on the piece, it started to cook out, with yellow smoke and green bubbles, after the GT 3000 tester had given three readings within range of low karat gold.
It also gives false negatives and low readings on known high grade gold coins and bullion bars.
If I had known what an unreliable piece of junk this thing is, I would never have bought it. Too bad that one star is the lowest rating available, or I would have given it a negative 3 stars.
The only reliable gear I know of utilizes an Xray technique. That device can be found here..
http://www.starstruckllc.com/Olympus-Delta-Inspector-HandHeld-Gold-Tester.php
Good luck with your device. Chemical testing is also known to be accurate. Involves acids and is destructive....
