主题：求助，有两个问题。有花相谢。 -- 黄河故人

If I understand the post correctly, essentially you want to know how to get the F value so that next time you pull a random sample and do the measurement(Fx), F and Fx will most likely to be the same.

First you need to determine what contributes most to the variation of B. You could go ahead and do a Gage Analysis, but I will take a guess and say it's your measurement system rather than you sample variation based on the fact that "每个样品上面在三个位置取三个值也在1.90-2.10之间波动".

If so, your sure bet is do as many measurement as you can, to average them out. By method one you only do three measurement and two fifteen, clearly two is better. Since the calculation F=A/B is so simple, I doubt which step you average would make a difference.

If you want to be more precise and acurate about your measurement, you really need to do a GAGE R&R, then target specifc aspects: operator, equipment,method etc.