My guess is slight carbon ring or measurement error on initial measurement. I've had it happen also with Hornady tool: used different force on new vs subsequent measures and ended with different readings.
I measure with 5 different bullets using firm pressure and use the shortest value as reference. My typical variation is -+.002 from reading to reading with Hornady/Sierra. Berger's tend to be closer to +-.001.
Also if your using the actual exact same bullet as reference for multiple readings, it's very possible it's slightly dented and would give short readings with successive measurements.