The statistical difference between 25,000 and 43,000 samples being... small. Basically, any effect visible at less than 2 in 40,000 might be missed by the 25,000 test.

Seriously, do you think this is a huge difference in quality of measurements?