Hi,

I assume you need to write more specific questions

I will give you some directions:

1. "Max drift" is (right/positive skew) and not normally distributed, so the writer uses LN transformation (natural log).

Now the reslt is still not symetrical (left/negative skew) and not normaly distributes but much better.

The writer assumed it can be count as normally distributed. so instead of treating Max drift it treats LN(Max drift)

I attached histograms and QQ plots

2. "F:I" continue to "A : D" and together this is the raw data

3. "L:M" is the raw data ("A : D" +" F:I") sorted by Sa

4. "N" is LN(Max drift)

5. "P:R" calculate mean and standard deviation of LN(Max drift) per each Sa value.

6. Now he compares the average to specific values (0.007 (IO),0.025(LS), 0.05(CP) ) to using Z=(x-μ/σ)

(say Z is a normal distribution and now you can easily calculate probabilities that x>μ (using excel NORMDIST function see a graphic demonstration of this function at

www.statskingdom.com/normal.html)

And you should continue and ask specific questions...