Help me out: Computing SE CV SED LSD

#1
Dear All,
In calculating standard error of mean (SE) and coefficient of variation (CV) for summary table of means exported to excel from SAS anova output of treatments with multiple response variables in RCBD design, which of these methods is correct formula for computing SE and CV:
1.
a) SE= root mean square error/root of no of replicates
where root mean square error is from anova table

b) SE= standard deviation of range of treatments mean/root of number of treatment.

2. a cv =(root mean square error/grand mean )*100%
b cv=stdev(range of treatment means)/grand mean) *100%
where the standard deviation is got from excel stdev function.

In summary, I wish to know if it is advisable to use root of mean square error from anova table in SAS output to compute SE and CV of exported table of means in excel rather than using standard deviation of range of treatment means in computing SE and CV.

question 2:
For scientific paper presentation in journal, is it advisable to compute standard error of difference of mean(SED) and least significant difference(LSD) for unbalanced rcbd design.

I look forward to your prompt and favourable response.