Here is the problem I been wondering. I have a manufacturer we work with at work. They have a manufacturing process to take ten different measurements of resistance, of different heights, on a pin component. They take two different measurements at each height. Each pin of ten measurements appears to follow a normal distribution (even though the sample is small). They do this same process with twenty different pins. When I look at the resistance data of all twenty pins, it doesn't have a normal distribution. It looks like a straight line in a normality plot, but fails the common normality tests in Minitab. This got me to wondering why data, obtained from a machine in manufacturing, isn't statistically normal. Can anyone enlighten me with some ideas?